\
+ \
+
\
+ ';
+
+ var replyTemplate = '\
+ \
+
\
+ \
+ \ + <%username%>\ + \ + <%time.delta%>\ +
\ +<#text#>
\
+ \ + \ + reply ▿\ + proposal ▹\ + \ + \ + \ +
\ +\ +<#proposal_diff#>\ +\ +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/_modules/futuregrid/virtual/cluster/FGCluster.html b/doc/build/html/_modules/futuregrid/virtual/cluster/FGCluster.html
new file mode 100644
index 0000000..4319b5f
--- /dev/null
+++ b/doc/build/html/_modules/futuregrid/virtual/cluster/FGCluster.html
@@ -0,0 +1,2305 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ Source code for futuregrid.virtual.cluster.CloudInstances
+#! /usr/bin/env python
+# -*- coding: utf-8 -*-
+
+'''
+CloudInstances.py (python)
+'''
+
+'''
+More ....
+-------------------------
+
+Operations on clusters based on backup file
+
+Backup file is the file which stores information of
+a list of virtual clusters
+
+Backup file format
+-------------------------
+[{cluster1}, {cluster2}, {cluster3}, {}]
+
+Each cluster is a dictionary, it has the following format:
+{'name':'cluster1', 'status':'run', '0':'{}', '1':'{}', ...}
+The first two keys are name, and status, the rest are instances.
+Name represents the name of virtual cluster and status represents
+the status of cluster. Key '0', '1', ... represent instances, each
+instance is associated with a number which can be used to refer to.
+Instance key starts at 0.
+
+Each instance is a dictionary, it has the following format:
+{'id':'', 'image':'', 'type':'', 'ip':''}
+'''
+
+import pickle
+import os
+
+
+[docs]class CloudInstances:
+ '''
+ Class CloudInstances
+ '''
+
+ cloud_instances = {}
+ backup_file = None
+ RUN = 'run'
+ SAVED = 'saved'
+ DOWN = 'terminated'
+
+ def __init__(self):
+ self.clear()
+ return
+
+[docs] def set_backup_file(self, filename):
+ '''
+ Sets backup file for reading and writing
+
+ Parameters:
+ filename -- backup file name
+
+ Logic:
+ Checks if the backup file has the correct format by
+ checking if all keys are present in each dictionary.
+
+ Return:
+ true -- If can not find backup file, so the file has not
+ been created.
+ false -- If file exists, but can not find all the keys
+ which are needed.
+ '''
+
+ self.backup_file = filename
+
+ # check if file is old version or corrupted
+ try:
+ src_file = open(os.path.expanduser(self.backup_file), "r")
+ cloud_list = pickle.load(src_file)
+ for cloud in cloud_list:
+ # check basic info
+ if not 'name' in cloud or \
+ not 'status' in cloud:
+ return False
+ # if is an instance, check all the keys
+ for element in cloud.values():
+ if type(element) is dict:
+ if not 'id' in element or \
+ not 'image' in element or \
+ not 'type' in element or \
+ not 'ip' in element:
+ return False
+ except (KeyError, TypeError):
+ return False
+ except IOError:
+ return True
+ return True
+
+[docs] def set_cloud_instances_by_name(self, name):
+ '''
+ Sets a new cloud instance
+
+ Parameters:
+ name -- virtual cluster name
+
+ Logic:
+ Granted this cluster is new and about to be created,
+ sets cluster name to name, and sets status to run.
+
+ No returns
+ '''
+
+ self.cloud_instances['name'] = name
+ # if create cluster, set status to RUN
+ self.cloud_instances['status'] = self.RUN
+
+[docs] def get_cloud_instances_by_name(self, name):
+ '''
+ Gets cloud instance list by cluster name
+
+ Parameters:
+ name -- virtual cluster name
+
+ Logic:
+ Granted we check cluster existence every time before
+ each operation (create, save, restore, terminate, ...)
+ so here no need to check if backup file is existed
+
+ Opens backup file for reading, loads cluster lists,
+ find cluster instances list given cluster name, then
+ sets cloud_instances to the cluster loaded
+
+ No returns
+ '''
+
+ # return cluster by given name
+ # using key 'name'
+ src_file = open(os.path.expanduser(self.backup_file), "r")
+ cloud_list = pickle.load(src_file)
+ for cloud in cloud_list:
+ if cloud['name'] == name:
+ self.cloud_instances = cloud
+
+[docs] def checkpoint_cloud_instances(self,
+ cluster_name,
+ control_node_id,
+ compute_node_id,
+ instance_type,
+ cluster_size):
+ '''
+ Sets cluster saving information after actually saves it
+
+ Parameters:
+ cluster_name -- virtual cluster name
+ control_node_id -- control node image id
+ compute_node_id -- compute node image id
+ instance_type -- instance type
+ cluster_size -- size of cluster (control node not included)
+
+ Logic:
+ Sets all possible values that could be used to restore cluster
+ in the future before saves it into backup file
+
+ Sets cluster status to SAVED
+
+ No returns
+ '''
+
+ # if save cluster, set basic info in backupfile
+ # resotre this cluster based on this info
+ self.clear()
+ self.cloud_instances['name'] = cluster_name
+ self.cloud_instances['control'] = control_node_id
+ self.cloud_instances['compute'] = compute_node_id
+ self.cloud_instances['type'] = instance_type
+ self.cloud_instances['size'] = cluster_size
+ self.cloud_instances['status'] = self.SAVED
+
+[docs] def get_all_cloud_instances(self):
+ '''
+ Gets all cloud instances lists
+
+ No parameters
+
+ Logic:
+ Loads all cloud instances from backup file.
+
+ Return:
+ cloud_list -- if file exists, return all the cloud lists
+ saved
+ [] -- if file does not exist
+ '''
+
+ try:
+ src_file = open(os.path.expanduser(self.backup_file), "r")
+ cloud_list = pickle.load(src_file)
+ return cloud_list
+ except IOError:
+ return []
+
+[docs] def get_cluster_size(self, cluster=None):
+ '''
+ Gets virtual cluster size
+
+ Parameters:
+ cluster -- virtual cluster name
+ default: None
+
+ Logic:
+ Default cloud instance is the dictionary self.cloud_instances,
+ if cluster is specified, then returns the size of cluster given
+ name.
+
+ Given cluster_instance has the format of:
+ {'name':'', 'status':'', '0':'', '1':'',...}
+ The size of cluster is the length of this dictionary subtracts
+ length of name and status which is 2
+
+ Return:
+ length of self.cloud_instances -- if cluster is set to default
+ length of cluster -- if cluster is set to any other clusters
+ '''
+
+ # get cluster size
+ # not include name and status fileds
+ if not cluster:
+ return len(self.cloud_instances) - 2
+ else:
+ return len(cluster) - 2
+
+[docs] def get_list(self):
+ '''
+ Gets cloud instances list
+
+ No parameters
+
+ Return:
+ cloud instances list -- self.cloud_instances
+ '''
+
+ return self.cloud_instances
+
+[docs] def get_index(self, instance):
+ '''
+ Gets instance key from cloud instance list
+
+ Parameters:
+ instance -- an instance dictionary
+
+ Logic:
+ Finds the instance matches the id
+ then return the key
+
+ Return:
+ key -- instance index
+ '''
+ for key, element in self.cloud_instances.items():
+ # if element is an instance
+ if type(element) is dict:
+ if element['id'] == instance['id']:
+ return key
+
+[docs] def set_instance(self,
+ instance_id,
+ image_id,
+ instance_type,
+ instance_ip='',
+ index=None):
+ '''
+ Sets attributes of a given instance
+
+ Parameters:
+ instance_id -- instance id
+ image_id -- image id
+ instance_type -- instance type
+ instance_ip -- public IP associated with instance
+ default: ''
+
+ Logic:
+ Set id, image, type, IP for one instance
+ Using current cluster size as key
+ So first instance has key 0, second has key 1 ...
+
+ No returns
+ '''
+
+ instance = {}
+ instance['id'] = instance_id
+ instance['image'] = image_id
+ instance['type'] = instance_type
+ instance['ip'] = instance_ip
+ # instance key '0, 1, 2...'
+ # can use index to get instance
+ if index == None:
+ self.cloud_instances[self.get_cluster_size()] = instance
+ else:
+ self.cloud_instances[index] = instance
+
+[docs] def del_instance(self, instance):
+ '''
+ Deletes an instance from cloud instance list
+
+ Parameters:
+ instance -- an instance dictionary
+
+ Logic:
+ Find the instance matches id then delete
+ the instance using key
+
+ No returns
+ '''
+
+ for key, element in self.cloud_instances.items():
+ # if element is an instance
+ if type(element) is dict:
+ if element['id'] == instance['id']:
+ del self.cloud_instances[key]
+
+[docs] def set_ip_by_id(self, instance_id, instance_ip):
+ '''
+ Sets IP by given instance id
+
+ Parameters:
+ instance_id -- instance id
+ instance_ip -- public IP associated with instance
+
+ Logic:
+ Loop all values in cloud_instances, given an instance
+ is a dictionary, if current element is a dictionary,
+ check key 'id', and then set 'ip' if key 'id' matches
+
+ No returns
+ '''
+
+ for element in self.cloud_instances.values():
+ # if element is an instance
+ if type(element) is dict:
+ if element['id'] == instance_id:
+ element['ip'] = instance_ip
+
+[docs] def clear(self):
+ '''
+ Clears cloud instances list
+
+ No parameters
+ No returns
+ '''
+
+ self.cloud_instances = {}
+
+[docs] def get_by_id(self, cloud_id):
+ '''
+ Gets instance by index
+
+ Parameters:
+ cloud_id -- instance index, which is also the key for instance
+
+ Logic:
+ Given that we use current size of cluster as key, so we can use
+ this index to get one specific instance
+
+ Return:
+ Instance dictionary -- an instance which has the format of
+ {'id':'', 'image':'', 'type':'', 'ip':''}
+ '''
+
+ return self.cloud_instances[cloud_id]
+
+[docs] def save_instances(self):
+ '''
+ Writes current cloud instances list into backup file
+
+ No parameters
+
+ Logic:
+ Opens backup file to save cloud_instance
+ If file does not exist, then creates backup file, dumps
+ cloud_instances into the file
+ If file exists, then loads the content from the file,
+ combine all cloud instance into list, dump this list into
+ file. So the list has the format of [{cluster1},{cluster2},{}]
+ (may need to change it to dictionary)
+
+ No returns
+ '''
+
+ try:
+ # open file for reading
+ src_file = open(os.path.expanduser(self.backup_file), "r")
+ instance_list = pickle.load(src_file)
+ # combine existing clusters and current cluster into list
+ instance_list.insert(0, self.cloud_instances)
+ src_file = open(os.path.expanduser(self.backup_file), "w")
+ pickle.dump(instance_list, src_file)
+ src_file.close()
+
+ except IOError:
+ # if cannot find this file, then create the file
+ if not os.path.exists(os.path.expanduser
+ (os.path.split(self.backup_file)[0])):
+ os.makedirs(os.path.expanduser
+ (os.path.split(self.backup_file)[0]))
+ src_file = open(os.path.expanduser(self.backup_file), "w")
+ # directly save cluster into backup file
+ pickle.dump([self.cloud_instances], src_file)
+ src_file.close()
+
+[docs] def if_exist(self, name):
+ '''
+ Checks if a given cluster name exists
+
+ Parameters:
+ name -- cluster name
+
+ Logic:
+ Given all clusters stored in backup file is a list, so loop
+ this list, and check key 'name' in each cluster dictionary
+
+ Return:
+ true -- if find a cluster matches the cluster name given as
+ parameter
+ false -- if 1) file does not exist
+ 2) Cannot find a cluster matches the cluster
+ name given as parameter after checking
+ all clusters saved in the backup file
+ '''
+
+ try:
+ src_file = open(os.path.expanduser(self.backup_file), "r")
+ cloud_list = pickle.load(src_file)
+ for cloud in cloud_list:
+ if cloud['name'] == name:
+ return True
+ return False
+
+ except IOError:
+ return False
+
+[docs] def del_by_name(self, name):
+ '''
+ Deletes cloud instances list from backup file given name
+
+ Parameter:
+ name -- cluster name
+
+ Logic:
+ Loads content from backup file, given content is list, then
+ delete the cluster which matches name using remove, then
+ dump the new content into backup file
+
+ No returns
+ '''
+
+ src_file = open(os.path.expanduser(self.backup_file), "r")
+ cloud_list = pickle.load(src_file)
+ for cloud in cloud_list:
+ if cloud['name'] == name:
+ # remove cluster if game matches
+ cloud_list.remove(cloud)
+ src_file = open(os.path.expanduser(self.backup_file), "w")
+ pickle.dump(cloud_list, src_file)
+ src_file.close()
+ return
+
+[docs] def set_status(self, status):
+ '''
+ Gets status of virtual cluster
+
+ Parameters:
+ status -- cluster status (run, saved, terminated)
+
+ Logic:
+ Sets the status in cloud instance list
+
+ No returns
+ '''
+
+ self.cloud_instances['status'] = status
+
+[docs] def if_status(self, status):
+ '''
+ Checks status of virtual cluster
+
+ Parameters:
+ status -- cluster status (run, saved, terminated)
+
+ Logic:
+ checks if status in cluster instance list matches
+ status given as parameter
+
+ Return:
+ true -- if status of cluster instance list matches
+ the status given as parameter
+ false -- if status of cluster instance list does
+ not match the status given as parameter
+ '''
+
+ return self.cloud_instances['status'] == status
+
+[docs] def get_status(self):
+ '''
+ Gets cluster status
+
+ Parameters:
+ status -- cluster status (run, saved, terminated)
+
+ Logic:
+ Get the value of key 'status'
+
+ Return:
+ Returns the value of status
+ '''
+
+ return self.cloud_instances['status']
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/_modules/futuregrid/virtual/cluster/Shell.html b/doc/build/html/_modules/futuregrid/virtual/cluster/Shell.html
new file mode 100644
index 0000000..9492d9d
--- /dev/null
+++ b/doc/build/html/_modules/futuregrid/virtual/cluster/Shell.html
@@ -0,0 +1,273 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ Source code for futuregrid.virtual.cluster.FGCluster
+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+'''
+FGCluster command
+'''
+
+"""
+FGCluster.py (python)
+-------------------------
+
+Operations for managing virtual clusters
+
+Name fg-cluster
+
+Description
+
+usage:
+
+ fg-cluster <parameters>
+
+ -f -- futuregrid configuration file
+ which specifies the location of
+ necessary files to run program
+
+ --debug -- show debug message
+
+ --default-repository -- set if to use IU ubuntu repo
+ if specifyed, then not use IU
+ ubuntu repo
+
+ --create-key -- create euca key file
+
+ subcommands:
+
+ run: run a virtual cluster of given parameters
+
+ -a -- virtual cluster name
+ -n -- compute node number
+ -i -- image type
+ -t -- instance type
+
+ checkpoint: save the state of currently running cluster
+
+ -a -- virtual cluster name
+ -c -- control node bucket name
+ -t -- control node image name
+ -m -- compute node bucket name
+ -e -- compute node image name
+
+ restore: restore state of saved virtual cluster
+
+ -a -- virtual cluster name
+
+ terminate: terminate a virtual cluster of given name
+
+ -a -- virtual cluster name
+
+ status: show status of virtual cluster(s)
+
+ -a -- virtual cluster name
+
+ If no virtual cluster name is specified, the command will return
+ status of all virtual clusters recorded
+
+ list: list virtual clusters
+"""
+
+import argparse
+import sys
+import os
+import time
+import threading
+import ConfigParser
+import random
+import Queue
+import re
+import boto.ec2
+
+from boto.ec2.connection import EC2Connection
+from futuregrid.virtual.cluster.CloudInstances import CloudInstances
+#from CloudInstances import CloudInstances
+from subprocess import Popen, PIPE
+from ConfigParser import NoOptionError
+from ConfigParser import MissingSectionHeaderError
+from ConfigParser import NoSectionError
+
+
+[docs]class Cluster(object):
+ """
+ Class Cluster
+ -------------
+ Methods to
+ run virtual cluster
+ checkpoint virtual cluster
+ restore virtual cluster
+ terminate virtual cluster
+ show status of virtual cluster(s)
+ list virtual clusters
+ """
+
+ userkey = None
+ cloud_instances = None
+ backup_file = None
+ user = None
+ enrc = None
+ slurm = None
+ mutex = None
+ interface = None
+ cloud = None
+ ec2_conn = None
+ user_login = None
+
+ # debug switch
+ if_debug = False
+ # if false, using IU ubuntu repository
+ if_default = False
+ # if true, create userkey key
+ if_create_key = False
+
+ # repo file name
+ sources_list = 'sources.list'
+
+ def __init__(self):
+ '''
+ init method
+ '''
+ super(Cluster, self).__init__()
+ self.cloud_instances = CloudInstances()
+ self.mutex = threading.Lock()
+
+# ---------------------------------------------------------------------
+# METHODS TO PRINT HELP MESSAGES
+# ---------------------------------------------------------------------
+
+ @classmethod
+[docs] def msg(cls, message):
+ '''
+ Method for printing help messages
+
+ Parameters:
+ message -- message to print
+
+ Logic:
+ use print to print message
+
+ No returns
+ '''
+
+ print message
+
+[docs] def print_section(self, message):
+ '''
+ Prints section header
+ '''
+
+ self.msg('\n=========================')
+ self.msg(message)
+ self.msg('=========================')
+
+[docs] def debug(self, message):
+ '''
+ Method for printing debug message
+
+ Parameter:
+ message -- message to print
+
+ Logic:
+ Checks debug flag, if debug flag is set to
+ true, then prints debug message, if is set
+ to false, then does not print debug message
+
+ No returns
+ '''
+
+ if self.if_debug:
+ print message
+
+# ---------------------------------------------------------------------
+# METHODS TO SET PROGRAM FLAGS
+# ---------------------------------------------------------------------
+[docs] def set_interface(self, interface):
+ '''
+ Set interface the program to use
+
+ Parameter:
+ interface -- interface (boto or euca2ools)
+
+ command line argument --interface will
+ overwrite the configure in configuration file
+
+ No returns
+ '''
+
+ if interface:
+ self.interface = interface
+ if self.interface not in ('euca2ools', 'boto'):
+ self.msg('Interface should be boto or euca2ools')
+ sys.exit()
+ if self.interface == 'boto':
+ self.ec2_connect(self.cloud)
+
+[docs] def set_cloud(self, cloud):
+ '''
+ Set cloud
+
+ Parameter:
+ cloud -- cloud (nova or eucalyptus)
+
+ command line argument --cloud will
+ overwrite the configure in configuration file
+
+ No returns
+ '''
+
+ if cloud:
+ self.cloud = cloud
+ if self.cloud not in ('nova', 'eucalyptus'):
+ self.msg('Cloud should be boto or euca2ools')
+ sys.exit()
+ if self.cloud == 'nova':
+ self.user_login = 'ubuntu'
+ elif self.cloud == 'eucalyptus':
+ self.user_login = 'root'
+
+[docs] def set_flag(self, args):
+ '''
+ Sets control flags
+
+ Parameter:
+ args -- this method deals args.if_debug, if_default
+ if_create_key.
+
+ if_debug: default is false
+ true to print debug message
+ false not to print
+
+ if_default: default is false
+ true to use default ubuntu repository
+ false to use IU ubuntu repository
+
+ if_create_key: default is false
+ true to create key for users
+ false not to create
+ if_boto: default is false
+ true to use boto interface
+ false to use euca2ools
+
+ Logic:
+ Sets control flags
+
+ No returns
+ '''
+
+ self.if_debug = args.debug
+ self.if_default = args.default_repository
+ self.if_create_key = args.create_key
+
+# ---------------------------------------------------------------------
+# METHODS ABOUT KEYPAIRS
+# ---------------------------------------------------------------------
+[docs] def if_keypair_exits(self, name):
+ '''
+ Checks if key is already created
+
+ Parameters:
+ name -- key name
+
+ Logic:
+ Uses euca-describe-keypairs to check the keys created,
+ finds name in the command result
+
+ Return:
+ true -- if finds the key given name as parameter
+ false -- if does not fine the key given name as parameter
+ '''
+
+ command = 'euca-describe-keypairs'
+ for element in self.get_command_result(command).split('\t'):
+ if element.find(name) >= 0:
+ return True
+ return False
+
+ @classmethod
+[docs] def add_keypair(cls, name, key):
+ '''
+ Adds keypair
+
+ Parameter:
+ name -- userkey name
+ key -- location of userkey
+
+ Logic:
+ Uses euca-add-keypair to create key for user, key will
+ be created under the location referred by key
+
+ No returns
+ '''
+
+ os.system('euca-add-keypair %s > %s' % (name, key))
+ os.chmod(os.path.expanduser(key), 0600)
+
+# ---------------------------------------------------------------------
+# METHOD TO PARSE CONFIGURATION FILE
+# ---------------------------------------------------------------------
+
+[docs] def parse_conf(self, file_name='unspecified'):
+ """
+ Parse configuration file
+
+ Parameters:
+ file_name -- configuration file name
+ default: unspecified
+
+ Logic:
+ Parse configuration file if given, default location
+ '~/.futuregrid/futuregrid.cfg'. If no file is given,
+ finds configuration file by following order:
+ 1) finds current directory 'futuregrid.cfg'
+ 2) finds in default location
+
+ configuration file format:
+ [virtual-cluster]
+ # Backup file for saving and loading virtual cluster(s)
+ backup = ~/.futuregrid/virtual-cluster
+ # Slurm configuration input file
+ slurm = ~/.futuregrid/slurm.conf.in
+ # userkey pem file
+ userkey = ~/PUT-YOUR-USER-NAME.pem
+ # environment file
+ enrc = ~/novarc
+
+ Checks if all files specified in configuration
+ file are present.If create-key is set to true,
+ first check if key exists, if key does not exist,
+ then creates key under the location where is specified
+ by userkey in the configuration file
+
+ Due to different version of this tool, back-end
+ structure of backup file may change, so checks
+ if backup file has the correct format before the start
+
+ Checks all possible errors about configuration file
+
+ No returns
+ """
+
+ config = ConfigParser.ConfigParser()
+ try:
+ # default location ~/.futuregrid/futuregrid.cfg
+ # read[], reads begin from last to first
+ # so, first reads file_name, then current directory, then default
+ config.read([os.path.expanduser('~/.futuregrid/futuregrid.cfg'),
+ 'futuregrid.cfg',
+ os.path.expanduser(file_name)])
+
+ self.backup_file = config.get('virtual-cluster', 'backup')
+ self.debug('Backup file %s' % self.backup_file)
+ self.userkey = config.get('virtual-cluster', 'userkey')
+ self.debug('Userkey %s' % self.userkey)
+ self.user = os.path.splitext(self.userkey.split('/')[-1])[0]
+ self.enrc = config.get('virtual-cluster', 'enrc')
+ self.debug('enrc %s' % self.enrc)
+ self.slurm = config.get('virtual-cluster', 'slurm')
+ self.debug('SLURM configuration input file %s' % self.slurm)
+ self.interface = config.get('virtual-cluster', 'interface')
+ self.cloud = config.get('virtual-cluster', 'cloud')
+
+ # checking if all file are present
+ self.debug('Checking if all file are present')
+ if not os.path.exists(os.path.expanduser(self.userkey)):
+ if self.if_create_key:
+ # create key for user
+# self.add_keypair(self.user)
+ if self.if_keypair_exits(self.user):
+ self.msg('\nYou have userkey %s.pem, please correct'
+ ' its location in configuration file'
+ % self.user)
+ sys.exit()
+ else:
+ self.add_keypair(self.user, self.userkey)
+ else:
+ self.msg('ERROR: You must have userkey file')
+ sys.exit(1)
+ if not os.path.exists(os.path.expanduser(self.enrc)):
+ self.msg('ERROR: You must have novarc file')
+ sys.exit(1)
+ else:
+ self.debug('Reading environment')
+ key_dir = os.path.dirname(self.enrc)
+ if key_dir.strip() == "":
+ key_dir = "."
+ if self.cloud == 'nova':
+ os.environ["NOVA_KEY_DIR"] = key_dir
+ elif self.cloud =='eucalyptus':
+ os.environ['EUCA_KEY_DIR'] = key_dir
+ with open(os.path.expanduser(self.enrc)) as enrc_content:
+ for line in enrc_content:
+ if re.search("^export ", line):
+ line = line.split()[1]
+ parts = line.split("=")
+ value = ""
+ for i in range(1, len(parts)):
+ parts[i] = parts[i].strip()
+ tempvar = os.path.expandvars(parts[i])
+ parts[i] = os.path.expanduser(tempvar)
+ value += parts[i] + "="
+ value = value.rstrip("=")
+ value = value.strip('"')
+ value = value.strip("'")
+ if parts[0] == 'EC2_CERT' or \
+ parts[0] == 'EC2_PRIVATE_KEY' or \
+ parts[0] == 'EUCALYPTUS_CERT':
+ if not os.path.exists(value):
+ self.msg('%s does not exist' % value)
+ sys.exit()
+ os.environ[parts[0]] = value
+
+ if not os.path.exists(os.path.expanduser(self.slurm)):
+ self.msg('ERROR: You must have slurm.conf.in file')
+ sys.exit(1)
+
+ self.debug('Checking backup file')
+ self.backup_file += self.cloud
+ if not self.cloud_instances.set_backup_file(self.backup_file):
+ self.msg('\nBackup file is corrupted, or you are using an old'
+ ' version of this tool. Please delete this backup file'
+ ' and try again.')
+ sys.exit(1)
+ self.debug('Checking done')
+
+ except IOError:
+ self.msg('\nError in reading configuration file!'
+ ' configuration file not created?')
+ sys.exit()
+ except (MissingSectionHeaderError, NoSectionError):
+ self.msg('\nError in reading configuratin file!'
+ ' No section header?')
+ sys.exit()
+ except NoOptionError:
+ self.msg('\nError in reading configuration file!'
+ ' Correct configuration format?')
+ sys.exit()
+ except ValueError:
+ self.msg('\nError in reading configuration file!'
+ ' Correct python version?')
+ sys.exit()
+
+# ---------------------------------------------------------------------
+# METHOD TO INSTALL
+# ---------------------------------------------------------------------
+
+[docs] def check_avaliable(self, instance):
+ '''
+ Check if instance is available
+
+ Parameters:
+ instance -- instance dictionary
+
+ Return:
+ True -- if instance is available
+ False -- if instance is unavailable
+ '''
+
+ cmd = "ssh -i %s %s@%s uname" % (self.userkey,
+ self.user_login,
+ instance['ip'])
+
+ check_process = Popen(cmd, shell=True, stdout=PIPE, stderr=PIPE)
+ status = os.waitpid(check_process.pid, 0)[1]
+ if status == 0:
+ return True
+ else:
+ return False
+
+[docs] def euca_start_new_instance(self, instance, instance_index):
+ '''
+ Starts new instance given old instance info using euca2ools
+
+ Parameter:
+ instance -- old instance
+ instance_index -- instance index
+ Return:
+ new instance
+ '''
+ # try to create a new one
+ self.msg('Creating new instance')
+ self.euca_run_instance(self.user, 1, instance['image'],
+ instance['type'], instance_index)
+ self.debug('Getting free public IPs')
+ # get free IP list
+ ip_lists = self.euca_describe_addresses()
+ time.sleep(2)
+ new_instance = self.cloud_instances.get_by_id(instance_index)
+ self.msg('\nAssociating IP with %s' % new_instance['id'])
+ new_ip = ip_lists[random.randint(0, len(ip_lists) - 1)]
+ while not self.euca_associate_address(new_instance, new_ip):
+ self.msg('Error in associating IP %s with instance %s, '
+ 'trying again' % (new_ip, new_instance['id']))
+ return new_instance
+
+[docs] def boto_start_new_instance(self, instance, instance_index):
+ '''
+ Starts new instance given old instance info using euca2ools
+
+ Parameter:
+ instance -- old instance
+ instance_index -- instance index
+ Return:
+ new instance
+ '''
+
+ reservation = self.boto_run_instances(instance['image'],
+ 1,
+ instance['type'])
+ new_instance = reservation.instances[0]
+ arg1 = new_instance.id
+ arg2 = new_instance.image_id
+ arg3 = new_instance.instance_type
+ self.cloud_instances.set_instance(instance_id=arg1,
+ image_id=arg2,
+ instance_type=arg3,
+ index=instance_index)
+ ip_list = self.boto_describe_addresses()
+ free_public_ip = ip_list[random.randint(0, len(ip_list) - 1)]
+ self.boto_associate_address(new_instance.id, free_public_ip)
+ new_instance.update()
+ return self.cloud_instances.get_by_id(instance_index)
+
+[docs] def euca_change_ip(self, instance):
+ '''
+ Changes public IP address given instance using euca2ools
+ No returns
+ '''
+
+ ip_lists = self.euca_describe_addresses()
+ # disassociate current one
+ self.disassociate_address(instance['ip'])
+ # associate a new random free ip
+ self.debug('Associating new IP on %s' % instance['id'])
+ self.euca_associate_address(instance, ip_lists[random.randint(0,
+ len(ip_lists) - 1)])
+
+[docs] def boto_change_ip(self, instance):
+ '''
+ Changes public IP address given instance using boto
+ No returns
+ '''
+ ip_list = self.boto_describe_addresses()
+ free_public_ip = ip_list[random.randint(0, len(ip_list) - 1)]
+ self.ec2_conn.disassociate_address(instance['ip'])
+ self.get_instance_from_reservation(instance['id']).update()
+ self.boto_associate_address(instance['id'], free_public_ip)
+
+[docs] def installation(self, instance, max_retry, install=True):
+ '''
+ Checks if instances are ready to deploy and installs the
+ softwares on the instance which is ready
+
+ Parameters:
+ instance -- instance dictionary
+ max_retry -- max number of try for checking instance
+ install -- indicates if need to the installation
+ default: true
+
+ Logic:
+ ssh reomote host to check if can succeed
+ If install is set to true, installation of
+ SLURM and OpenMPI will start on VM which is ready
+
+ Each instances associates a count of times for trying
+ If exceeds the max try limit, then gets free public IPs,
+ and associates it with a random new one, reset count to 0
+
+ Break after all threads are done
+
+ No returns
+ '''
+
+ wait_count = 0
+
+ ip_change = False
+
+ # check if ssh port of all VMs are alive for listening
+ while True:
+ while self.if_status(instance['id'], 'pending'):
+ self.msg('Instance %s is pending' % instance['id'])
+ time.sleep(5)
+ if self.check_avaliable(instance):
+ self.mutex.acquire()
+ self.del_known_host(instance['ip'])
+ self.mutex.release()
+ if install:
+ self.deploy_services(instance)
+ break
+ elif self.cloud == 'nova':
+ self.debug('ssh in %s is closed' % instance['ip'])
+ self.msg('Checking %s (%s) availability, '
+ 'trying %d (max try %d)'
+ % (instance['id'], instance['ip'],
+ wait_count, max_retry))
+
+ wait_count += 1
+ if wait_count > max_retry:
+ if self.if_status(instance['id'], 'shutdown') or \
+ self.if_status(instance['id'], 'terminate') or \
+ self.if_status(instance['id'], 'terminated') or \
+ ip_change:
+ # get instance index
+ instance_index = \
+ self.cloud_instances.get_index(instance)
+ self.msg('Instance %s creation failed'
+ % instance['id'])
+ # delete this instance from cloud instance list
+ self.cloud_instances.del_instance(instance)
+ self.terminate_instance(instance['id'])
+ self.mutex.acquire()
+ if self.interface == 'euca2ools':
+ instance = \
+ self.euca_start_new_instance(instance,
+ instance_index)
+ elif self.interface == 'boto':
+ instance = \
+ self.boto_start_new_instance(instance,
+ instance_index)
+ self.mutex.release()
+ wait_count = 0
+ ip_change = False
+ elif self.if_status(instance['id'], 'running'):
+ self.msg('\nTrying different IP address on %s'
+ % instance['id'])
+ # get free ip addresses
+ self.mutex.acquire()
+ if self.interface == 'euca2ools':
+ self.euca_change_ip(instance)
+ elif self.interface == 'boto':
+ self.boto_change_ip(instance)
+ self.mutex.release()
+ self.debug('New IP is %s' % instance['ip'])
+ wait_count = 0
+ ip_change = True
+ time.sleep(1)
+
+# ---------------------------------------------------------------------
+# METHODS TO DO RPCs
+# ---------------------------------------------------------------------
+
+[docs] def get_command_result(self, command):
+ '''
+ Gets command output
+
+ Parameters:
+ command -- shell command
+
+ Return:
+ Command output
+ '''
+ self.debug(command)
+ return os.popen(command).read()
+
+[docs] def execute(self, instance, command):
+ '''
+ Executes a command on the instance
+
+ Parameters:
+ instance -- cloud instance
+ command -- shell command
+
+ No returns
+ '''
+
+ self.debug("ssh -i %s ubuntu@%s '%s'" % (self.userkey,
+ instance['ip'], command))
+ os.system("ssh -i %s %s@%s '%s'" % (self.userkey,
+ self.user_login,
+ instance['ip'],
+ command))
+
+[docs] def copyto(self, instance, filename):
+ '''
+ Copies the named file to the instance
+
+ Parameters:
+ instance -- cloud instance
+ filename -- the name of file to copy
+
+ No returns
+ '''
+
+ self.debug('scp -i %s %s ubuntu@%s:~/' % (self.userkey,
+ filename, instance['ip']))
+ os.system('scp -i %s %s %s@%s:~/' % (self.userkey,
+ filename,
+ self.user_login,
+ instance['ip']))
+
+[docs] def update(self, instance):
+ '''
+ Executes a software update on the specified instance
+
+ Parameters:
+ instance -- cloud instance
+
+ No returns
+ '''
+
+ self.execute(instance, 'sudo apt-get update')
+
+[docs] def install(self, instance, packagenames):
+ '''
+ Installs the package names that are specified
+ (blank separated on the given instance)
+
+ Parameter:
+ instance -- cloud instance
+ packagenames -- software names
+
+ No returns
+ '''
+
+ self.execute(instance, 'sudo apt-get install --yes '
+ + packagenames)
+
+# ---------------------------------------------------------------------
+# METHODS TO CREATE A VIRTUAL CLUSTER
+# ---------------------------------------------------------------------
+
+[docs] def ec2_connect(self, cloud):
+ '''
+ create connection
+ '''
+ if cloud.strip() == 'nova':
+ path = "/services/Cloud"
+ elif cloud.strip() == 'eucalyptus':
+ path = "/services/Eucalyptus"
+
+ endpoint = os.getenv('EC2_URL').lstrip("http://").split(":")[0]
+ try:
+ region = boto.ec2.regioninfo.RegionInfo(name=cloud,
+ endpoint=endpoint)
+ except:
+ self.msg('ERROR: error in getting region information')
+ sys.exit()
+
+ try:
+ self.ec2_conn = EC2Connection(os.getenv("EC2_ACCESS_KEY"),
+ os.getenv("EC2_SECRET_KEY"),
+ is_secure=False,
+ region=region,
+ port=8773, path=path)
+ except:
+ self.msg('ERROR: error in connecting to EC2')
+ sys.exit()
+
+[docs] def boto_describe_addresses(self):
+ '''
+ get all free ip addresses using boto
+ '''
+
+ ip_list = []
+ for address in \
+ self.ec2_conn.get_all_addresses(addresses=None):
+ address_public_ip = address.public_ip
+ address_instance_id = address.instance_id
+ if not address_instance_id:
+ ip_list.append(address_public_ip)
+ return ip_list
+
+[docs] def get_instance_from_reservation(self, instance_id):
+ '''
+ get instance from reservation
+ '''
+
+ r_in = self.ec2_conn.get_all_instances(instance_ids=[instance_id])
+ return r_in[0].instances[0]
+
+[docs] def boto_associate_address(self, instance_id, address_public_ip):
+ '''
+ associate ip address using boto
+ '''
+
+ while True:
+ try:
+ time.sleep(2)
+ self.ec2_conn.associate_address(instance_id, address_public_ip)
+ break
+ except:
+ self.msg('ERROR: Associating ip %s with instance %s failed, '
+ 'trying again' % (address_public_ip, instance_id))
+ self.msg('ADDRESS: %s %s' % (address_public_ip, instance_id))
+ self.cloud_instances.set_ip_by_id(instance_id, address_public_ip)
+
+[docs] def boto_run_instances(self, image, cluster_size, instance_type):
+ '''
+ run instance usign boto
+ '''
+
+ try:
+ return self.ec2_conn.run_instances(image_id=image,
+ min_count=cluster_size,
+ max_count=cluster_size,
+ key_name=self.user,
+ instance_type=instance_type)
+ except:
+ self.msg('ERROR: Error in lunching instances, please try again')
+ self.ec2_conn.close()
+ sys.exit()
+
+[docs] def euca_run_instance(
+ self,
+ userkey,
+ cluster_size,
+ image,
+ instance_type,
+ index=None
+ ):
+ '''
+ runs instances given parameters
+
+ Parameters:
+ userkey -- userkey name
+ cluster_size -- number of nodes
+ image -- image id
+ instance_type -- instance type
+ index -- instance key as index
+ default: None
+
+ Logic:
+ check if all instances are created
+ correctly. program will exit if
+ not all instances required are
+ created, and terminate those which
+ are running already
+
+ No returns
+ '''
+
+ instance_id_list = []
+
+ # get run instances store int instances list
+ self.debug('euca-run-instances -k %s -n %d -t %s %s'
+ % (userkey,
+ cluster_size,
+ instance_type,
+ image))
+ instances = [x for x in
+ self.get_command_result(
+ 'euca-run-instances -k %s'
+ ' -n %d -t %s %s'
+ % (userkey, cluster_size, instance_type,
+ image)).split()]
+ # find created instance, store into list
+ for instance in instances:
+ if instance.find('i-') == 0:
+ self.debug('%s is created' % instance)
+ instance_id_list.append(instance)
+
+ self.debug('Checking if all instances are created')
+ # check if all instances are created correctly
+ if not len(instance_id_list) == cluster_size:
+ self.msg('\nError in creating cluster, please check your input'
+ ' or instance limit exceeded?')
+ for created_instance_id in instance_id_list:
+ self.terminate_instance(created_instance_id)
+ sys.exit()
+ self.debug('Checking done')
+
+ self.debug('Adding instances into cloud instances list')
+ # add instance to instance list
+ for num in range(cluster_size):
+ try:
+ ip_addr = instance_id_list[num]
+ self.debug('Adding instance %s' % ip_addr)
+ self.cloud_instances.set_instance(instance_id=ip_addr,
+ image_id=image,
+ instance_type=instance_type,
+ index=index)
+ except:
+ self.msg('\nError in creating instances.'
+ ' Program will exit')
+ sys.exit()
+
+[docs] def euca_associate_address(self, instance, free_ip):
+ '''
+ Associates instance with ip
+
+ Parameters:
+ instance -- cloud instance
+ free_ip -- free public IP address
+
+ Logic:
+ Associates IP with a given instance, set IP in cloud_instances
+ delete host information in known_hosts if it has
+
+ Return:
+ false: if association failed
+ trueL: if association succeeded
+ '''
+
+ self.debug('euca-associate-address -i %s %s'
+ % (instance['id'], free_ip))
+ if self.get_command_result('euca-associate-address -i %s %s'
+ % (instance['id'],
+ free_ip)).find('ADDRESS') < 0:
+ return False
+ # set ip using instance id
+ self.msg('ADDRESS %s instance %s' % (free_ip, instance['id']))
+ self.cloud_instances.set_ip_by_id(instance['id'], free_ip)
+ # delete host from known_host file in case man-in-middle-attack
+ self.debug('Deleting %s from known host if it already existed'
+ % free_ip)
+ return True
+
+[docs] def disassociate_address(self, current_ip):
+ '''
+ Disassociates IP address
+
+ Parameters:
+ current_ip -- instance IP which is currently in use
+
+ Logic:
+ Uses euca-disassociates-address to disassociates a
+ given IP address
+
+ No returns
+ '''
+
+ self.msg('Disaasociating %s' % current_ip)
+ self.debug('euca-disassociate-address %s' % current_ip)
+ os.system('euca-disassociate-address %s' % current_ip)
+
+[docs] def euca_describe_addresses(self):
+ '''
+ Return list of free public IP addresses
+
+ No parameters
+
+ Logic:
+ Uses euca-describe-addresses to get all the IP addresses,
+ filter those which are currently in use
+
+ Return:
+ ip_list -- a list of free public IP addresses
+ '''
+
+ ip_list = []
+ ips = [x for x in self.get_command_result('euca-describe-addresses'
+ ).split('\n')]
+
+ # store free ip into ip list for return
+ for free_ip in ips:
+ if free_ip.find('i-') < 0 and len(free_ip) > 0:
+ ip_list.append(free_ip.split('\t')[1])
+ return ip_list
+
+[docs] def if_status(self, instance_id, status):
+ '''
+ Check if instance is running
+ '''
+
+ result = self.get_command_result('euca-describe-instances').split('\n')
+
+ for i in result:
+ if i.find(instance_id) >= 0:
+ if i.find(status) >= 0:
+ return True
+ else:
+ return False
+[docs] def euca_get_ip(self, instance_id):
+ '''
+ Get instance public given instance id
+ '''
+ result = self.get_command_result('euca-describe-instances').split('\n')
+ for i in result:
+ if i.find(instance_id) >= 0 and i.find('pending') >= 0:
+ return i.split('\t')[3]
+
+[docs] def create_cluster(self, args):
+ '''
+ Method for creating cluster
+
+ Parameters:
+ args -- this method deals
+ args.name: cluster name
+ args.number: cluster compute nodes number
+ args.typy: instance type
+ args,image: image id
+
+ Logic:
+ Check existence before creation of cluster
+ If cluster was created before, only start creating
+ if status is DOWN
+
+ Save each instance into cloud instance list
+ Associate IP with each instance
+ Create source list file if using IU ubuntu repository
+ Delete source list file after installtion and configuration
+ of SLURM and OpenMPI
+
+ No returns
+ '''
+
+ self.debug('Checking if %s is existed' % args.name)
+ if self.cloud_instances.if_exist(args.name):
+ # if exists, get cloud info by name
+ self.debug('%s is existed, gettting it from backup file'
+ % args.name)
+ self.cloud_instances.get_cloud_instances_by_name(args.name)
+ # if cluster is terminated, delete old info, start over
+ self.debug('Checking if %s is currently down' % args.name)
+ if self.cloud_instances.if_status(self.cloud_instances.DOWN):
+ self.debug('Deleting old info')
+ self.cloud_instances.del_by_name(args.name)
+ self.cloud_instances.clear()
+ else:
+ self.msg('\nError in creating cluster %s,name is in use'
+ % args.name)
+ sys.exit()
+
+ cluster_size = int(args.number) + 1
+ self.debug('Cluster size is (control node included): %d'
+ % cluster_size)
+ self.print_section('Creating virtual cluster')
+ self.msg('cluster name -- %s' % args.name)
+ self.msg('numbe of nodes -- %s' % cluster_size)
+ self.msg('instance type -- %s' % args.type)
+ self.msg('image id -- %s' % args.image)
+
+ self.debug('Creating new cloud instance %s' % args.name)
+ # set cloud instance list
+ self.cloud_instances.set_cloud_instances_by_name(args.name)
+
+ self.debug('Creating cluster')
+ # run instances given parameters
+ if self.interface == 'euca2ools':
+ self.euca_run_instance(self.user, cluster_size, args.image,
+ args.type)
+ # immediatly associate ip after run instance
+ # may lead to error, use sleep
+ time.sleep(3)
+
+ if self.cloud == 'nova':
+ self.debug('Getting free public IPs')
+ # get free IP list
+ ip_lists = self.euca_describe_addresses()
+
+ self.msg('\nAssociating public IP addresses')
+ for i in range(cluster_size):
+ instance = self.cloud_instances.get_by_id(i)
+ time.sleep(1)
+ while not self.euca_associate_address(instance, ip_lists[i]):
+ self.msg('Error in associating IP %s with instance %s, '
+ 'trying again' % (ip_lists[i], instance['id']))
+ # eucalyptus no need to associate ip
+ if self.cloud == 'eucalyptus':
+ for i in range(cluster_size):
+ instance = self.cloud_instances.get_by_id(i)
+ time.sleep(1)
+ self.cloud_instances.set_ip_by_id(instance['id'],
+ self.euca_get_ip(instance['id']))
+
+ elif self.interface == 'boto':
+ reservation = \
+ self.boto_run_instances(args.image, cluster_size, args.type)
+ self.msg('Associating public IPs')
+ ip_index = 0
+ for instance in reservation.instances:
+ instance.update()
+ self.cloud_instances.set_instance(instance.id,
+ instance.image_id,
+ instance.instance_type)
+ free_public_ip = self.boto_describe_addresses()[ip_index]
+ ip_index += 1
+ self.boto_associate_address(instance.id, free_public_ip)
+
+ self.debug('Creating IU ubunto repo source list')
+ # choose repo, by defalt, using IU ubuntu repo
+ self.define_repo()
+
+ self.debug('Checking alive instance for deploying')
+ # detect if VMs are ready for deploy
+
+ for instance in self.cloud_instances.get_list().values():
+ if type(instance) is dict:
+ threading.Thread(target=self.installation,
+ args=[instance, 30, True]).start()
+
+ while threading.activeCount() > 1:
+ time.sleep(1)
+
+ self.debug('Configuraing SLURM')
+ # config SLURM system
+ self.config_slurm()
+
+ self.debug('Cleaning up')
+ # clean repo file
+ self.clean_repo()
+ self.debug('Saving cloud instance into backup file')
+ # save cloud instance
+ self.cloud_instances.save_instances()
+ self.debug('Done creation of cluster')
+
+[docs] def clean_repo(self):
+ '''
+ Remove source list file
+
+ No parameters:
+
+ Logic:
+ If if_default flag is set to false, then delete
+ source list file
+
+ No returns
+ '''
+
+ if not self.if_default:
+ self.debug('Removing %s' % self.sources_list)
+ os.remove(self.sources_list)
+
+[docs] def config_slurm(self, create_key=True):
+ '''
+ Configures slurm
+
+ Parameters:
+ create_key -- indicates whether to create munge-key
+ default: true, creates key
+
+ Logic:
+ Reads slurm configuration input file, substitutes controlMachine
+ with control machine id. Append each computation node into
+ configuration file in COMPUTE NODES section
+ Generates munge-key on control node if flag is set to true
+ Does configuration on each node in parallel
+ After all threads are done, removes temp files
+
+ No returns
+ '''
+
+ slurm_conf_file = 'slurm.conf'
+ munge_key_file = 'munge.key'
+
+ self.debug('Opening %s' % self.slurm)
+ self.msg('\nConfiguring slurm.conf')
+ with open(os.path.expanduser(self.slurm)) as srcf:
+ input_content = srcf.readlines()
+ srcf.close()
+
+ self.debug('Getting control machie id')
+ # set control machine
+ controlMachine = self.cloud_instances.get_by_id(0)['id']
+ output = ''.join(input_content) % vars()
+
+ self.msg('\nControl node %s' % controlMachine)
+
+ self.debug('Writting into %s' % slurm_conf_file)
+ # write control machine into slurm.conf file
+ destf = open(slurm_conf_file, 'w')
+ print >> destf, output
+ destf.close()
+
+ self.debug('Openning %s for adding computation nodes'
+ % slurm_conf_file)
+ # add compute machines to slurm conf file
+ with open(slurm_conf_file, 'a') as conf:
+ for instance in self.cloud_instances.get_list().values():
+ if type(instance) is dict:
+ if not instance['id'] == controlMachine:
+ self.debug('Adding instance %s' % instance['id'])
+ conf.write('NodeName=%s Procs=1 State=UNKNOWN\n'
+ % instance['id'])
+ conf.write('PartitionName=debug Nodes=%s Default=YES'
+ ' MaxTime=INFINITE State=UP\n'
+ % instance['id'])
+ conf.close()
+
+ # if needs to create munge key
+ if create_key:
+ self.msg('\nGenerating munge-key')
+
+ # generate munge-key on control node
+
+ self.execute(self.cloud_instances.get_by_id(0),
+ 'sudo /usr/sbin/create-munge-key')
+ self.debug('Opening %s for writting munge-key' % munge_key_file)
+ munge_key = open(munge_key_file, 'w')
+ print >> munge_key, \
+ self.get_command_result("ssh -i %s %s@%s"
+ " 'sudo cat /etc/munge/munge.key'"
+ % (self.userkey,
+ self.user_login,
+ self.cloud_instances.get_by_id(0)['ip'
+ ]))
+ munge_key.close()
+
+ # copy SLURM conf file to every node
+ for instance in self.cloud_instances.get_list().values():
+ if type(instance) is dict:
+ self.debug('Starting SLURM on %s' % instance['id'])
+ process_thread = threading.Thread(target=self.start_slurm,
+ args=[instance,
+ create_key,
+ slurm_conf_file,
+ munge_key_file])
+ process_thread.start()
+
+ # wait all threads are done
+ self.debug('Waitting all threads are done')
+ while threading.activeCount() > 1:
+ time.sleep(1)
+
+ # clean temp files
+ if create_key:
+ self.debug('Removing %s' % munge_key_file)
+ os.remove(munge_key_file)
+ self.debug('Removing %s' % slurm_conf_file)
+ os.remove(slurm_conf_file)
+
+[docs] def start_slurm(self,
+ instance,
+ create_key,
+ slurm_conf_file,
+ munge_key_file):
+ '''
+ Copies SLURM configuration file and munge key file
+ to every node, and start slurm on each node
+
+ Parameters:
+ instance -- cloud instance
+ create_key -- flag indicates whether needs to create munge key
+ slurm_conf_file -- SLURM configuration file name
+ munge_key_file -- munge key file name
+
+ Logic:
+ Copies SLURM configuration file to each node in cluster.
+ If needs to create munge key, then creates munge key on
+ control node, copies it to each node in cluster. Starts
+ SLURM and munge after copy is done
+
+ No returns
+ '''
+
+ # copy slurm.conf
+ self.msg('\nCopying slurm.conf to node %s' % instance['id'])
+ self.copyto(instance, slurm_conf_file)
+ self.execute(instance, 'sudo cp slurm.conf /etc/slurm-llnl')
+
+ # copy munge key
+ if create_key:
+ self.msg('\nCopying munge-key to node %s' % instance['id'])
+ self.copyto(instance, munge_key_file)
+ self.execute(instance,
+ 'sudo cp munge.key /etc/munge/munge.key')
+ self.execute(instance,
+ 'sudo chown munge /etc/munge/munge.key')
+ self.execute(instance,
+ 'sudo chgrp munge /etc/munge/munge.key')
+ self.execute(instance,
+ 'sudo chmod 400 /etc/munge/munge.key')
+
+ # start slurm and munge daemon
+ self.msg('\nStarting slurm on node %s' % instance['id'])
+ self.execute(instance, 'sudo /etc/init.d/slurm-llnl start')
+ self.execute(instance, 'sudo /etc/init.d/munge start')
+
+[docs] def define_repo(self):
+ '''
+ Set ubuntu repo to IU ubuntu repository
+
+ No parameters
+
+ Source list file content:
+ -------------------------
+
+ deb http://ftp.ussg.iu.edu/linux/ubuntu/ natty-updates main
+ deb-src http://ftp.ussg.iu.edu/linux/ubuntu/ natty-updates main
+ deb http://ftp.ussg.iu.edu/linux/ubuntu/ natty universe
+ deb-src http://ftp.ussg.iu.edu/linux/ubuntu/ natty universe
+ deb http://ftp.ussg.iu.edu/linux/ubuntu/ natty-updates universe
+ deb-src http://ftp.ussg.iu.edu/linux/ubuntu/ natty-updates universe
+ deb http://ftp.ussg.iu.edu/linux/ubuntu/ natty main
+ deb-src http://ftp.ussg.iu.edu/linux/ubuntu/ natty main
+
+ Logic:
+ If uses IU ubuntu repository, creates sources_list file
+ which has the former content
+
+ No returns
+ '''
+
+ iu_repo = 'http://ftp.ussg.iu.edu/linux/ubuntu/'
+
+ if self.if_default:
+ self.msg('\nUsing default repository')
+ else:
+ self.msg('\nUsing IU ubuntu repository')
+ self.debug('Using %s' % iu_repo)
+ self.debug('Opening %s for writting' % self.sources_list)
+ with open(self.sources_list, 'w') as source:
+ source.write('deb ' + iu_repo + ' natty-updates main\n')
+ source.write('deb-src ' + iu_repo + ' natty-updates main\n')
+ source.write('deb ' + iu_repo + ' natty universe\n')
+ source.write('deb-src ' + iu_repo + ' natty universe\n')
+ source.write('deb ' + iu_repo + ' natty-updates universe\n')
+ source.write('deb-src ' + iu_repo +
+ ' natty-updates universe\n')
+ source.write('deb ' + iu_repo + ' natty main\n')
+ source.write('deb-src ' + iu_repo + ' natty main\n')
+ source.close()
+
+[docs] def deploy_services(self, instance):
+ '''
+ Deploies SLURM and OpenMPI services
+
+ Parameters:
+ instance -- cloud instance
+
+ Logic:
+ If using IU ubuntu repository, then copies source_list
+ to each instance in the cluster, installs SLURM and
+ OpenMPI on the instance
+
+ No returns
+ '''
+
+ self.msg('\nInstalling SLURM system and OpenMPI on %s\n'
+ % instance['ip'])
+
+ if not self.if_default:
+ self.debug('Copying %s to %s' % (self.sources_list,
+ instance['id']))
+ self.copyto(instance, self.sources_list)
+ self.execute(instance, 'sudo cp %s /etc/apt/'
+ % self.sources_list)
+
+ self.debug('Updating on %s' % instance['id'])
+ self.update(instance)
+ # install SLURM
+ self.debug('Installing slrum-llnl on %s' % instance['id'])
+ self.install(instance, 'slurm-llnl')
+ # install OpenMPI
+ self.debug('Installing openmpi on %s' % instance['id'])
+ self.install(instance, "openmpi-bin libopenmpi-dev")
+
+# ---------------------------------------------------------------------
+# METHODS TO SAVE RUNNING VIRTUAL CLUSTER
+# ---------------------------------------------------------------------
+
+[docs] def save_instance(
+ self,
+ kernel_id,
+ ramdisk_id,
+ instance_ip,
+ instance_name,
+ ):
+ '''
+ Saves one instance
+
+ Parameters:
+ kernel_id: image kernel id
+ ramdisk_id: image ramdisk id
+ instance_ip: instance public IP address
+ instance_name: image name
+
+ Logic:
+ Uses euca-bundle-vol to create a bundle of current image on
+ a remote host. Image has the size of 1GB, and stored under /mnt/
+
+ No returns
+ '''
+
+ if kernel_id == None:
+ self.debug("ssh -i %s %s@%s '. ~/.profile;"
+ " sudo euca-bundle-vol -c ${EC2_CERT}"
+ " -k ${EC2_PRIVATE_KEY} -u ${EC2_USER_ID}"
+ " --ec2cert ${EUCALYPTUS_CERT} --no-inherit"
+ " -p %s -s 1024 -d /mnt/'"
+ % (self.userkey,
+ self.user_login,
+ instance_ip,
+ instance_name))
+ return self.get_command_result("ssh -i %s %s@%s '. ~/.profile;"
+ " sudo euca-bundle-vol -c ${EC2_CERT}"
+ " -k ${EC2_PRIVATE_KEY} -u ${EC2_USER_ID}"
+ " --ec2cert ${EUCALYPTUS_CERT} --no-inherit"
+ " -p %s -s 1024 -d /mnt/'"
+ % (self.userkey,
+ self.user_login,
+ instance_ip,
+ instance_name))
+ elif ramdisk_id == None:
+ self.debug("ssh -i %s %s@%s '. ~/.profile;"
+ " sudo euca-bundle-vol -c ${EC2_CERT}"
+ " -k ${EC2_PRIVATE_KEY} -u ${EC2_USER_ID}"
+ " --ec2cert ${EUCALYPTUS_CERT} --no-inherit"
+ " -p %s -s 1024 -d /mnt/ --kernel %s'"
+ % (self.userkey,
+ self.user_login,
+ instance_ip,
+ instance_name,
+ kernel_id))
+ return self.get_command_result("ssh -i %s %s@%s '. ~/.profile;"
+ " sudo euca-bundle-vol -c ${EC2_CERT}"
+ " -k ${EC2_PRIVATE_KEY} -u ${EC2_USER_ID}"
+ " --ec2cert ${EUCALYPTUS_CERT} --no-inherit"
+ " -p %s -s 1024 -d /mnt/ --kernel %s'"
+ % (self.userkey,
+ self.user_login,
+ instance_ip,
+ instance_name,
+ kernel_id))
+ else:
+ self.debug("ssh -i %s %s@%s '. ~/.profile;"
+ " sudo euca-bundle-vol -c ${EC2_CERT}"
+ " -k ${EC2_PRIVATE_KEY} -u ${EC2_USER_ID}"
+ " --ec2cert ${EUCALYPTUS_CERT} --no-inherit"
+ " -p %s -s 1024 -d /mnt/ --kernel %s --ramdisk %s'"
+ % (self.userkey,
+ self.user_login,
+ instance_ip,
+ instance_name,
+ kernel_id,
+ ramdisk_id))
+ return self.get_command_result("ssh -i %s %s@%s '. ~/.profile;"
+ " sudo euca-bundle-vol -c ${EC2_CERT}"
+ " -k ${EC2_PRIVATE_KEY} -u ${EC2_USER_ID}"
+ " --ec2cert ${EUCALYPTUS_CERT} --no-inherit"
+ " -p %s -s 1024 -d /mnt/ --kernel %s --ramdisk %s'"
+ % (self.userkey,
+ self.user_login,
+ instance_ip,
+ instance_name,
+ kernel_id,
+ ramdisk_id))
+
+[docs] def upload_bundle(
+ self,
+ instance_ip,
+ bucket_name,
+ manifest,
+ ):
+ '''
+ Uploads bundle image given manifest
+
+ Parameters:
+ instance_ip -- instance public IP address
+ bucket_name -- bucket name for bundled image
+ manifest -- manifest for bundled image
+
+ Logic:
+ Uses euca-upload-bundle to upload bundle
+
+ No returns
+ '''
+
+ self.debug("ssh -i %s %s@%s '. ~/.profile;"
+ " euca-upload-bundle -b %s -m %s'"
+ % (self.userkey,
+ self.user_login,
+ instance_ip,
+ bucket_name,
+ manifest))
+ return self.get_command_result("ssh -i %s %s@%s '. ~/.profile;"
+ " euca-upload-bundle -b %s -m %s'"
+ % (self.userkey,
+ self.user_login,
+ instance_ip,
+ bucket_name,
+ manifest))
+
+[docs] def describe_images(self, image_id):
+ '''
+ Gets image infomation
+
+ Parameters:
+ image_id -- image id
+
+ Logic:
+ Uses euca-describe-images to get all the inforamtion of image
+ given image id
+
+ Return:
+ Command result of euca-describe-images of given image id
+ '''
+
+ self.debug('euca-describe-images %s' % image_id)
+ return self.get_command_result('euca-describe-images %s' % image_id)
+
+[docs] def get_kernel_id(self, image_id):
+ '''
+ Gets kernel id
+
+ Parameters:
+ image_id -- image id
+
+ Logic:
+ Parses the command results returned by describe image method.
+ Given that sometimes kernel id is empty for certain images,
+ checks if the command returens a kernel id before return it
+
+ Return:
+ kernal id -- kernal id is the 8th element of returned command
+ result
+ '''
+
+ command_result = [x for x in
+ self.describe_images(image_id).split()]
+ if len(command_result) >= 8:
+ self.debug('Kernel ID %s' % command_result[7])
+ return command_result[7]
+
+[docs] def get_ramdisk_id(self, image_id):
+ '''
+ Gets ramdisk id
+
+ Parameters:
+ image_id -- image id
+
+ Logic:
+ Parses the command results returned by describe image method.
+ Given that sometimes ramdisk id is empty for certain images,
+ checks if the command returens a ramdisk id before return it
+
+ Return:
+ ramdisk id -- ramdisk id is the 9th element of returned command
+ result
+ '''
+
+ command_result = [x for x in
+ self.describe_images(image_id).split()]
+ if len(command_result) == 9:
+ self.debug("Ramdisk ID %s" % command_result[8])
+ return command_result[8]
+
+[docs] def save_node(
+ self,
+ image_id,
+ instance_ip,
+ bucket_name,
+ image_name,
+ node_type,
+ out_queue
+ ):
+ '''
+ Saves nodem, upload and register
+
+ Parameters:
+ image_id -- image id of instance
+ instance_ip -- instance public IP address
+ bucket_name -- bucket name for bundled image
+ image_name -- image name for bundled image
+ node_type -- control node or compute node
+ out_queue -- queue in which puts output
+
+ Logic:
+ After gets all the necessary infomraion to save bundle a node,
+ then saves the node, uploads the bundle
+
+ Return:
+ bundled image id -- Parses the command result, gets the bundled
+ image id, then returns it
+ '''
+
+ bundled_image_id = {}
+
+ kernel_id = self.get_kernel_id(image_id)
+ self.debug('Kernel ID %s' % kernel_id)
+ ramdisk_id = self.get_ramdisk_id(image_id)
+ self.debug("Ramdisk ID %s" % ramdisk_id)
+ # get manifest from the last unit
+ manifest = [x for x in self.save_instance(kernel_id,
+ ramdisk_id, instance_ip, image_name).split()].pop()
+
+ self.msg('\nManifest generated: %s' % manifest)
+ self.msg('\nUploading bundle')
+
+ # upload image
+ image = [x for x in self.upload_bundle(instance_ip, bucket_name,
+ manifest).split()].pop()
+ self.debug(image)
+ self.msg('\nUploading done')
+ self.msg('\nRegistering image')
+
+ # register image, and return image id
+ bundled_image_id[node_type] = \
+ self.euca_register(image).split('\t')[1].strip()
+ out_queue.put(bundled_image_id)
+
+[docs] def euca_register(self, image):
+ '''register image'''
+
+ return self.get_command_result('euca-register %s' % image)
+
+[docs] def checkpoint_cluster(self, args):
+ '''
+ Method for saving virtual cluster
+
+ Parameters:
+ args -- this method deals with
+ args.name -- virtual cluster name
+ args.controlb -- control node bucket name
+ args.controln -- control node image name
+ args.computeb -- compute node bucket name
+ args.computen -- compute node image name
+
+ Logic;
+ Checks existence before saving virtual cluster
+ Only saves cluster which is running (Because saved and
+ terminated clusters are not currenly running)
+ Only saves control node and one compute node into images
+ Saves new control node image id and new compute node image id
+ into backup file for later restore. Change status to SAVED after
+ the saving. Then terminates cluster before deletes host information
+ after each termination
+
+ No returns
+ '''
+ if self.interface == 'boto':
+ self.msg('UNIMPLEMENTED')
+ sys.exit()
+ if self.cloud == 'eucalyptus':
+ self.msg('bugs')
+ sys.exit()
+
+ self.debug('Checking if %s is existed' % args.name)
+ # check if cluter is existed
+ if self.cloud_instances.if_exist(args.name):
+ # get cluter by name
+ self.debug('Getting cloud instance %s' % args.name)
+ self.cloud_instances.get_cloud_instances_by_name(args.name)
+ # if cluster is down, terminate the program
+ if not self.cloud_instances.if_status(self.cloud_instances.RUN):
+ self.msg('Error in locating virtual cluster %s, not running?'
+ % args.name)
+ sys.exit()
+ else:
+ self.msg('Error in locating virtual cluster %s, not created?'
+ % args.name)
+ sys.exit()
+
+ self.print_section('Saving virtual cluster')
+ self.msg('Virtual cluster name -- %s' % args.name)
+ self.msg('control node bucket -- %s' % args.controlb)
+ self.msg('control node name -- %s' % args.controln)
+ self.msg('compute node bucket -- %s' % args.computeb)
+ self.msg('compute node name -- %s' % args.computen)
+
+ control = self.cloud_instances.get_by_id(0)
+ compute = self.cloud_instances.get_by_id(1)
+ self.debug('Control node %s, compute node %s'
+ % (control, compute))
+
+ # copy necessary files to instances, and source profile
+ for instance in [control, compute]:
+ self.copyto(instance, os.environ['EC2_CERT'])
+ self.copyto(instance, os.environ['EC2_PRIVATE_KEY'])
+ self.copyto(instance, os.environ['EUCALYPTUS_CERT'])
+ self.copyto(instance, self.enrc)
+ self.execute(instance, 'cat %s >> ~/.profile'
+ % self.enrc.split('/')[-1])
+ self.execute(instance, 'source ~/.profile')
+
+ save_queue = Queue.Queue()
+ self.msg('\nSaving control node %s' % control['id'])
+ threading.Thread(target=self.save_node, args=[control['image'],
+ control['ip'],
+ args.controlb,
+ args.controln,
+ 'control',
+ save_queue
+ ]).start()
+ self.msg('\nSaving compute node %s' % compute['id'])
+ threading.Thread(target=self.save_node, args=[compute['image'],
+ compute['ip'],
+ args.computeb,
+ args.computen,
+ 'compute',
+ save_queue
+ ]).start()
+ while threading.activeCount() > 1:
+ time.sleep(1)
+
+ # return values has the format:
+ # {'control':'', 'compute':''}, but not sure the order
+ for bundled_image_id in [save_queue.get(), save_queue.get()]:
+ if 'control' in bundled_image_id:
+ control_node_id = bundled_image_id['control']
+ elif 'compute' in bundled_image_id:
+ compute_node_id = bundled_image_id['compute']
+
+ # get instance type
+ instance_type = control['type']
+ self.debug('Instance type %s' % instance_type)
+ # get compute node number
+ cluster_size = self.cloud_instances.get_cluster_size() - 1
+ self.debug('Number of computation nodes %d' % cluster_size)
+ # copy list for termination
+ temp_instance_list = list(self.cloud_instances.get_list().values())
+
+ # delete old info
+ self.debug('Deleting %s from backup file' % args.name)
+ self.cloud_instances.del_by_name(args.name)
+
+ self.debug('Setting save info')
+ # set saved cloud info, and change status to saved
+ self.cloud_instances.checkpoint_cloud_instances(args.name,
+ control_node_id,
+ compute_node_id,
+ instance_type,
+ cluster_size)
+ # save cluster
+ self.debug('Saving cluster into backup file')
+ self.cloud_instances.save_instances()
+
+ # terminate instances
+ for instance in temp_instance_list:
+ if type(instance) is dict:
+ self.terminate_instance(instance['id'])
+ self.del_known_host(instance['ip'])
+# ---------------------------------------------------------------------
+# METHODS TO RESTORE VIRTUAL CLUSTER
+# ---------------------------------------------------------------------
+
+[docs] def restore_cluster(self, args):
+ '''
+ Method for restoring cluster
+
+ Parameters:
+ args -- this method deals with
+ args.name -- virtual cluster name
+
+ Logic:
+ Loads control node id, compute node id, instance type,
+ cluster size from backup file.
+ Creates cluster of size one using control node id
+ Creates cluster of cluster size using compute node id
+ Associates IP addresses with all created instances
+ Granted cluster was saved before, so no need to install
+ softwares again, but need to configure SLURM accordingly
+ because all host names changed. No need to create mumge-key
+ because munge-key was also saved in each instance during the
+ saving process.
+
+ Change status to RUN in the end, and save this restored cluster
+ infomation into backup file
+
+ No returns
+ '''
+ if self.interface == 'boto':
+ self.msg('UNIMPLEMENTED')
+ sys.exit()
+ if self.cloud == 'eucalyptus':
+ self.msg('bugs')
+ sys.exit()
+
+ control_node_num = 1
+
+ # only restore cluster which is saved
+ self.debug('Checking if %s is existed' % args.name)
+ if self.cloud_instances.if_exist(args.name):
+ self.debug('Getting cloud %s' % args.name)
+ self.cloud_instances.get_cloud_instances_by_name(args.name)
+ if self.cloud_instances.if_status(self.cloud_instances.SAVED):
+ self.debug('Cloud status: %s' % self.cloud_instances.SAVED)
+ # if cluster is saved, delete old cluster, and create a
+ # new cloud instance for deploying
+ control_node_id = self.cloud_instances.get_list()['control']
+ self.debug('control node %s' % control_node_id)
+ compute_node_id = self.cloud_instances.get_list()['compute']
+ self.debug('compute node %s' % compute_node_id)
+ instance_type = self.cloud_instances.get_list()['type']
+ self.debug('instance type %s' % instance_type)
+ cluster_size = self.cloud_instances.get_list()['size']
+ self.debug('cluster size %s' % cluster_size)
+ self.debug('Creating new cloud instance list')
+ self.cloud_instances.clear()
+ self.debug('Deleting old info from backup file')
+ self.cloud_instances.set_cloud_instances_by_name(args.name)
+ else:
+ self.msg('Error in restoring virtual cluster %s, not saved?'
+ % args.name)
+ sys.exit()
+ else:
+ self.msg('Error in locating virtual cluster %s, not created?'
+ % args.name)
+ sys.exit()
+
+ cluster_size = int(cluster_size) + control_node_num
+ self.print_section('Restoring virtual cluster')
+ self.msg('cluster name -- %s' % args.name)
+ self.msg('number of nodes -- %s' % cluster_size)
+ self.msg('instance type -- %s' % instance_type)
+ self.msg('control image -- %s' % control_node_id)
+ self.msg('compute image -- %s' % compute_node_id)
+
+ # run control node
+ self.debug('Creating control node %s' % control_node_id)
+ self.euca_run_instance(self.user, control_node_num,
+ control_node_id, instance_type)
+ # run compute nodes given number
+ self.debug('Creating compute node %s' % compute_node_id)
+ self.euca_run_instance(self.user, cluster_size - control_node_num,
+ compute_node_id, instance_type)
+
+ # get free ip list
+ self.debug('Getting free IP list')
+ ip_lists = self.euca_describe_addresses()
+
+ time.sleep(5)
+
+ self.msg('\nAssociating IPs')
+ for i in range(cluster_size):
+ time.sleep(1)
+ self.debug('Getting cloud from index %d' % i)
+ instance = self.cloud_instances.get_by_id(i)
+ while not self.euca_associate_address(instance, ip_lists[i]):
+ self.msg('Error in associating IP %s with instance %s, '
+ 'trying again'
+ % (ip_lists[i], instance['id']))
+ time.sleep(1)
+
+ # check ssh port but not install
+ self.debug('Checking alive instance for deploying')
+ for instance in self.cloud_instances.get_list().values():
+ if type(instance) is dict:
+ threading.Thread(target=self.installation,
+ args=[instance, 60, False]).start()
+ while threading.activeCount() > 1:
+ time.sleep(1)
+ # cnfig SLURM but not generating munge-keys
+ self.debug('Configuating SLURM')
+ self.config_slurm(False)
+ # set status to run and save
+ self.debug('Setting status to %s' % self.cloud_instances.RUN)
+ self.cloud_instances.set_status(self.cloud_instances.RUN)
+ self.debug('Deleting old cloud instance info')
+ self.cloud_instances.del_by_name(args.name)
+ self.debug('Saving cloud instance info')
+ self.cloud_instances.save_instances()
+
+# ---------------------------------------------------------------------
+# METHODS TO TERMINATE NAD CLEANUP
+# ---------------------------------------------------------------------
+ @classmethod
+[docs] def del_known_host(cls, ip_addr):
+ '''
+ Deletes known host info from ~/.ssh/known_hosts
+
+ Parameter:
+ ip_addr -- IP address
+
+ Logic:
+ Deletes known host information from ~/.ssh/known_hosts
+ in case it shows man-in-middle-attack message when starts
+ different cluster but using the same IP addresses
+
+ No returns
+ '''
+
+ known_hosts = '~/.ssh/known_hosts'
+# self.msg('Deleting host info from known_hosts')
+
+ with open(os.path.expanduser(known_hosts)) as srcf:
+ host_list = srcf.readlines()
+ srcf.close()
+
+ with open(os.path.expanduser(known_hosts), 'w') as destf:
+ for host in host_list:
+ if host.find(ip_addr) < 0:
+ destf.write(host)
+ destf.close()
+
+[docs] def terminate_instance(self, instance_id):
+ '''terminate instance given instance id'''
+
+ self.msg('Terminating instance %s' % instance_id)
+ if self.interface == 'euca2ools':
+ os.system('euca-terminate-instances %s' % instance_id)
+ elif self.interface == 'boto':
+ try:
+ self.ec2_conn.terminate_instances([instance_id])
+ except:
+ self.msg('ERROR: terminat instance %s failed' % instance_id)
+
+[docs] def shut_down(self, args):
+ '''
+ Method for shutting down a cluster
+
+ Parameters:
+ args -- this method deals with
+ args.name -- virtual cluster name
+
+ Logic:
+ Only terminates cloud instance which is not terminated
+ Checking its existence before termination.
+ Delete host info from ~/.ssh/known_hosts after each termination
+ Change status according to following:
+ If current status is SAVED, then does not change status
+ If current status is RUN, then changes it to DOWN
+
+ No returns
+ '''
+
+ # only terminate cluster which is not terminated
+ self.debug('Checking if %s is existed' % args.name)
+ if self.cloud_instances.if_exist(args.name):
+ self.debug('Getting cloud instance %s' % args.name)
+ self.cloud_instances.get_cloud_instances_by_name(args.name)
+ self.debug('Checking cloud status')
+ # check if cloud instance is terminated
+ if self.cloud_instances.if_status(self.cloud_instances.DOWN):
+ self.msg('\nError in terminating cluster %s, already down?'
+ % args.name)
+ sys.exit()
+ else:
+ self.msg('\nError in finding virtual cluster %s, not created?'
+ % args.name)
+ sys.exit()
+
+ for instance in self.cloud_instances.get_list().values():
+ if type(instance) is dict:
+ self.terminate_instance(instance['id'])
+ self.del_known_host(instance['ip'])
+
+ # change status to terminated, and save
+ self.debug('If status is %s' % self.cloud_instances.RUN)
+ if self.cloud_instances.if_status(self.cloud_instances.RUN):
+ self.debug('Setting status to %s' % self.cloud_instances.DOWN)
+ self.cloud_instances.set_status(self.cloud_instances.DOWN)
+ self.debug('Deleting instance old info')
+ self.cloud_instances.del_by_name(args.name)
+ self.debug('Saving instance into backup file')
+ self.cloud_instances.save_instances()
+# ---------------------------------------------------------------------
+# METHODS TO SHOW VIRTUAL CLUSTER(S) STATUS
+# ---------------------------------------------------------------------
+
+[docs] def show_status(self, args):
+ '''
+ Show status of cluster(s)
+
+ Parameters:
+ args -- this method deals with
+ args.name -- virtual cluster name
+
+ Logic:
+ Reads from backup file to load cluster information given
+ cluster name. If cluster name is specified, then after
+ checks its existence, loads it to display. If no cluster
+ name is specified, then loads all cluster instances to
+ display. If no clusters are saved in the backup file, after
+ prints help message, then quits the program
+
+ No returns
+ '''
+
+ if not args.name:
+ # get all cloud intances
+ cloud_set = self.cloud_instances.get_all_cloud_instances()
+ # if no cloud instances created, then prints msg and quits
+ if len(cloud_set) == 0:
+ self.msg('\nYou do not have any virtual clusters')
+ sys.exit()
+ for cloud in cloud_set:
+ self.msg('\n====================================')
+ self.msg('Virtual Cluster %s (status: %s)'
+ % (cloud['name'], cloud['status']))
+ self.msg('====================================')
+ if cloud['status'] == self.cloud_instances.SAVED:
+ self.msg('Control node -- %s, '
+ 'Compute node -- %s, '
+ 'Instance type -- %s, '
+ 'Cluster size -- %s'
+ % (cloud['control'], cloud['compute'],
+ cloud['type'], cloud['size']))
+ else:
+ for index in range(self.cloud_instances.get_cluster_size(
+ cloud)):
+ if index == 0:
+ node_type = 'control node'
+ else:
+ node_type = 'compute node'
+ self.msg('%s:\t%s\t%s\t%s\t%s'
+ % (cloud[index]['id'], node_type,
+ cloud[index]['ip'], cloud[index]['image'],
+ cloud[index]['type']))
+ else:
+ if not self.cloud_instances.if_exist(args.name):
+ self.msg('Error in finding virtual cluster %s, not created.'
+ % args.name)
+ sys.exit()
+ self.cloud_instances.get_cloud_instances_by_name(args.name)
+ self.msg('\n====================================')
+ self.msg('Virtual Cluster %s (status: %s)'
+ % (args.name, self.cloud_instances.get_status()))
+ self.msg('====================================')
+ if self.cloud_instances.get_list()['status'] == \
+ self.cloud_instances.SAVED:
+ self.msg('Control node -- %s, '
+ 'Compute node -- %s, '
+ 'Instance type -- %s, '
+ 'Cluster size -- %s'
+ % (self.cloud_instances.get_list()['control'],
+ self.cloud_instances.get_list()['compute'],
+ self.cloud_instances.get_list()['type'],
+ self.cloud_instances.get_list()['size']))
+ else:
+ for index in range(self.cloud_instances.get_cluster_size()):
+ cloud = self.cloud_instances.get_by_id(index)
+ if index == 0:
+ node_type = 'control node'
+ else:
+ node_type = 'compute node'
+ self.msg('%s:\t%s\t%s\t%s\t%s'
+ % (cloud['id'], node_type,
+ cloud['ip'], cloud['image'],
+ cloud['type']))
+
+# ---------------------------------------------------------------------
+# METHODS TO SHOW VIRTUAL CLUSTER LIST
+# ---------------------------------------------------------------------
+
+[docs] def get_list(self, _args):
+ '''
+ lists all virtual clusters and status
+
+ Parameters:
+ args -- this method deals with
+ None
+
+ Logic:
+ Reads from backup file to load all virtual cluster information
+ If no cloud instance saved, after shows help message, then quits
+ the program
+ '''
+
+ # get all cloud instances lists
+ cloud_set = self.cloud_instances.get_all_cloud_instances()
+ # if no cloud created, then prints msg and quits
+ if len(cloud_set) == 0:
+ self.msg('\nYou do not have any virtual clusters')
+ sys.exit()
+
+ self.msg('\n===============================')
+ self.msg('Virtual Cluster list')
+ self.msg('================================')
+ for cloud in cloud_set:
+ self.msg('%s: %d compute nodes, 1 control node; status: %s'
+ % (cloud['name'],
+ self.cloud_instances.get_cluster_size(cloud) - 1,
+ cloud['status']))
+
+######################################################################
+# MAIN
+######################################################################
+
+
+[docs]def commandline_parser():
+ '''
+ Parses commandline
+
+ Checks if python version is above 2.7
+ '''
+
+ # Check pyhon version
+ if sys.version_info < (2, 7):
+ print "ERROR: you must use python 2.7 or greater"
+ sys.exit(1)
+
+ virtual_cluster = Cluster()
+
+ parser = \
+ argparse.ArgumentParser(description='Virtual'
+ ' cluster management operations',
+ version='0.2.0')
+ parser.add_argument('--file', action='store',
+ help='Specify futuregrid configure file')
+ parser.add_argument('--debug', action='store_true',
+ help='print debug message')
+ parser.add_argument('--default-repository', action='store_true',
+ help='using default software repository')
+ parser.add_argument('--create-key', action='store_true',
+ help='create userkey')
+ parser.add_argument('--interface', action='store',
+ help='choose interface to use')
+ parser.add_argument('--cloud', action='store',
+ help='choose cloud')
+ subparsers = parser.add_subparsers(help='commands')
+
+ # status command
+ status_parser = subparsers.add_parser('status',
+ help='Show virtual cluster status')
+ status_parser.add_argument('-a', '--name', action='store',
+ help='Show status of '
+ 'virtual cluster of given name')
+ status_parser.set_defaults(func=virtual_cluster.show_status)
+
+ # list command
+ list_parser = subparsers.add_parser('list',
+ help='List virtual cluster and status')
+ list_parser.set_defaults(func=virtual_cluster.get_list)
+
+ # run command
+ run_parser = subparsers.add_parser('run',
+ help='Create a virtual cluster')
+ run_parser.add_argument('-a', '--name', action='store',
+ required=True, help='Virtual cluster name')
+ run_parser.add_argument('-n', '--number', action='store',
+ required=True, help='Numbe of compute nodes'
+ )
+ run_parser.add_argument('-t', '--type', action='store',
+ required=True, help='Instance type')
+ run_parser.add_argument('-i', '--image', action='store',
+ required=True, help='Image id')
+ run_parser.set_defaults(func=virtual_cluster.create_cluster)
+
+ # terminate command
+ terminate_parser = subparsers.add_parser('terminate',
+ help='Terminate virtual cluster')
+ terminate_parser.add_argument('-a', '--name', action='store',
+ required=True,
+ help='Virtual cluster name')
+ terminate_parser.set_defaults(func=virtual_cluster.shut_down)
+
+ # checkpoint command
+ checkpoint_parser = subparsers.add_parser('checkpoint',
+ help='Save virtual cluster')
+ checkpoint_parser.add_argument('-a', '--name', action='store',
+ required=True,
+ help='Virtual cluster name')
+ checkpoint_parser.add_argument('-c', '--controlb', action='store',
+ required=True,
+ help='Control node bucket name')
+ checkpoint_parser.add_argument('-t', '--controln', action='store',
+ required=True,
+ help='Control node image name')
+ checkpoint_parser.add_argument('-m', '--computeb', action='store',
+ required=True,
+ help='Compute node bucket name')
+ checkpoint_parser.add_argument('-e', '--computen', action='store',
+ required=True,
+ help='Compute node image name')
+ checkpoint_parser.set_defaults(func=virtual_cluster.checkpoint_cluster)
+
+ # restore command
+ restore_parser = subparsers.add_parser('restore',
+ help='Restore saved virtual cluster')
+ restore_parser.add_argument('-a', '--name', action='store',
+ required=True,
+ help='Virtual cluster name')
+ restore_parser.set_defaults(func=virtual_cluster.restore_cluster)
+
+ args = parser.parse_args()
+
+ # parse config file, if config file is not specified,
+ # then use default file which is ~/.futuregrid/futuregrid.cfg
+ if args.file:
+ virtual_cluster.parse_conf(args.file)
+ else:
+ virtual_cluster.parse_conf()
+
+ # set flags
+ virtual_cluster.set_flag(args)
+ # choose interface
+ virtual_cluster.set_cloud(args.cloud)
+ virtual_cluster.set_interface(args.interface)
+ args.func(args)
+
+if __name__ == '__main__':
+ commandline_parser()
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/_modules/index.html b/doc/build/html/_modules/index.html
new file mode 100644
index 0000000..1f22165
--- /dev/null
+++ b/doc/build/html/_modules/index.html
@@ -0,0 +1,137 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ Source code for futuregrid.virtual.cluster.Shell
+#!/usr/bin/env python
+'''Virtual cluster shell'''
+
+#import argparse
+#import os
+import pprint
+import optparse
+from cmd2 import Cmd
+from cmd2 import make_option
+from cmd2 import options
+#from cmd2 import Cmd2TestCase
+from futuregrid.virtual.cluster.FGCluster import Cluster
+#from FGCluster import Cluster
+import unittest
+import sys
+
+
+[docs]class Shell(Cmd):
+ '''Virtual cluster shell command'''
+ #multilineCommands = ['None']
+ #Cmd.shortcuts.update({'&': 'speak'})
+ #maxrepeats = 3
+ #Cmd.settable.append('maxrepeats')
+
+ pp = pprint.PrettyPrinter(indent=0)
+
+ echo = True
+ timing = True
+ cluster = None
+
+ prompt = "fg> "
+
+ logo = """
+ _____ _ ____ _ _
+ | ___|_ _| |_ _ _ _ __ ___ / ___|_ __(_) __| |
+ | |_ | | | | __| | | | '__/ _ \ | _| '__| |/ _` |
+ | _| | |_| | |_| |_| | | | __/ |_| | | | | (_| |
+ |_| \__,_|\__|\__,_|_| \___|\____|_| |_|\__,_|
+----------------------------------------------------
+ """
+
+ def preloop(self):
+ self.cluster = Cluster()
+ print self.logo
+
+ def postloop(self):
+ print "BYE FORM GREGOR"
+
+ @options([
+ make_option('-f', '--file', type="string",
+ help="cluster name"),
+ make_option('-i', '--interface', type="string",
+ help="interface")
+ ])
+[docs] def do_config(self, args, opts):
+ '''config - add the configuration from a file'''
+ if not opts.file:
+ self.cluster.parse_conf()
+ else:
+ self.cluster.parse_conf()
+ self.cluster.set_interface(opts.interface)
+
+ def do_debug(self, debug=True):
+ print "to do debug"
+
+ @options([
+ make_option('-a', '--name', type="string",
+ help="cluster name"),
+ make_option('-n', '--number', type="int",
+ help="number of computation nodes"),
+ make_option('-t', '--type', type="string",
+ help="instance type"),
+ make_option('-i', '--image', type="string",
+ help="image id"),
+ ])
+
+ @options([
+ make_option('-a', '--name', type="string",
+ help="cluster name"),
+ make_option('-c', '--controlb', type="string",
+ help="control node bucket"),
+ make_option('-t', '--controln', type="string",
+ help="control node image"),
+ make_option('-m', '--computeb', type="string",
+ help="compute node bucket"),
+ make_option('-e', '--computen', type="string",
+ default='not specified',
+ help="compute node image"),
+ ])
+
+ @options([
+ make_option('-a', '--name', type="string",
+ help="cluster name"),
+ ])
+
+ @options([
+ make_option('-a', '--name', type="string",
+ help="cluster name"),
+ ])
+
+ @options([
+ make_option('-a', '--name', type="string",
+ help="cluster name"),
+ ])
+
+ def do_list(self, args):
+ self.cluster.get_list(args)
+
+
+def main():
+ parser = optparse.OptionParser()
+ parser.add_option('-t', '--test',
+ dest='unittests',
+ action='store_true',
+ default=False,
+ help='Run unit test suite')
+ (callopts, callargs) = parser.parse_args()
+ if callopts.unittests:
+ sys.argv = [sys.argv[0]] # the --test argument upsets unittest.main()
+ unittest.main()
+ else:
+ app = Shell()
+ app.cmdloop()
+
+if __name__ == '__main__':
+ main()
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/_sources/index.txt b/doc/build/html/_sources/index.txt
new file mode 100644
index 0000000..cfaf523
--- /dev/null
+++ b/doc/build/html/_sources/index.txt
@@ -0,0 +1,24 @@
+.. Cloud Metrics documentation master file, created by
+ sphinx-quickstart on Tue Apr 10 10:11:26 2012.
+ You can adapt this file completely to your liking, but it should at least
+ contain the root `toctree` directive.
+
+Welcome to Virtual Cluster documentation!
+=========================================
+
+Contents:
+
+.. toctree::
+ :maxdepth: 2
+
+ intro
+ modules
+ futuregrid
+
+Indices and tables
+==================
+
+* :ref:`genindex`
+* :ref:`modindex`
+* :ref:`search`
+
diff --git a/doc/build/html/_sources/intro.txt b/doc/build/html/_sources/intro.txt
new file mode 100644
index 0000000..8abef7d
--- /dev/null
+++ b/doc/build/html/_sources/intro.txt
@@ -0,0 +1,470 @@
+INTRODUCTION
+============
+
+System requirements
+------------------
+
+* euca2ools: verison 1.2
+* python: version 2.7
+* virtualenv (no admin rights)
+
+Introduction
+------------
+
+Project futuregrid.virtual.cluster is a virtual cluster management
+software utilizing openstack resources on FutureGrid. The software
+contains several parts that could help users easily manage their
+operations to virtual clusters on FutureGrid.
+
+Managment operations include:
+
+* Create virtual cluster(s)
+* Save currently running virtual cluster(s)
+* Restore saved virtual cluster(s)
+* Show status of virtual cluster(s)
+* Terminate virtual cluster(s)
+
+
+HOW TO INSTALL
+==============
+
+Installation
+------------
+
+We assume that yo do not have super user priviledges on the computer
+where you like to install our tool.
+
+### Step 0: Prerequisites (not using india.futuregrid.org machines):
+
+In order to make the installation process work smoothly, please make
+sure that the computer you like to install our tool has already
+installed euca2ools (version 1.2) and Python (version 2.7). If you do
+not have those tools or correct version installed. you may install
+euca2ools (version 1.2) from
+
+* http://eucalyptussoftware.com/downloads/releases/
+
+and install python (version 2.7) from
+
+* http://python.org
+
+After you check all the tools with version are correctly installed,
+you may proceed with following steps to start installation.
+
+
+### Step 0: Prerequisites (using india.futuregrid.org machines):
+
+india.futuregrid.org has installed every tools you may need to finish this
+installation, so to activate euca2ools (version 1.2) after you login into
+india futuregrid machines, just simply do:
+
+ $ module load euca2ools
+
+To activate python 2.7, simply do:
+
+ $ module load python
+
+Those commands will help you load tools with correct version you need
+to finish installation. So now you may proceed with following
+installation steps.
+
+### Step 1: Download virtualenv
+
+Since you do not have super user priviledges, you need virtualenv in
+order to finish the installtion. You may download virtualenv.py by
+following command:
+
+ $ wget https://raw.github.com/pypa/virtualenv/master/virtualenv.py
+
+### Step 2: Install virtualenv
+
+After you downloaded virtualenv, you can install it by following
+command:
+
+ $ python virtualenv.py --system-site-packages ENV
+
+### Step 3: Activate virtualenv
+
+After installation of virtualenv, you can activate virtualenv by
+following command:
+
+ $ source ENV/bin/activate
+
+### Step 4: install the virtual cluster with pip
+
+Once virtualenv is activated, you can use pip to install our tool by
+following command:
+
+ $ ENV/bin/pip install futuregrid.virtual.cluster
+
+If you already had our tool installed, and you want to upgrade to
+newest version, you can do it by following command:
+
+ $ ENV/bin/pip install --upgrade futuregrid.virtual.cluster
+
+NOTE: For more information about virtualenv, you may see documentation
+of virtualenv on
+
+* http://www.virtualenv.org/en/latest/index.html
+
+FutureGrid Specific Instalation
+-------------------------------
+
+### Install without admin rights on india futuregrid
+
+Go to the futuregrid portal
+
+* https://portal.futuregrid.org/
+
+If you do not have a futuregrid account, please apply for one at link:
+
+* https://portal.futuregrid.org/user/register
+
+In order for you to get access to FutureGrid resources, you need to
+apply for a portal account and create a Futuregrid project. This is in
+detail explained at
+
+* https://portal.futuregrid.org/gettingstarted
+
+Do not forget to upload your public key. (see also
+https://portal.futuregrid.org/generating-ssh-keys-futuregrid-access)
+
+Once you have a vaild portal account and a valid project, you can go
+ahead and use FutureGrid
+
+Our virtual cluster is best executed on our machine called
+india.futuregrid.org
+
+Please log into this machine and follow the steps that we have outlined
+in the previous section to install the software and then run it while
+following the instaructions from the next section
+
+
+HOW TO RUN
+==========
+
+Prerequisites
+-------------
+
+In order to use our tool, you need to obatin nova credentials and
+configuration files for FutureGrid system, you can obtain your nova
+credentials and configuration files for the FutureGrid systems. These
+should have been placed in your home directory on the INDIA
+system. Log in with your FutureGrid username (and SSH public key) and
+look for a file called 'username'-nova.zip. If you do not have a
+portal and HPC account please create one. The credential zip file
+(username-nova.zip)contains the user keys and rc file .Unzip this
+file in your hom e directory. The novarc file contains the necessary
+environment variables.
+
+Create your private key by (Recommended: Use your user name as your
+private key name):
+
+ $ euca-add-keypair youruserkey > youruserkey.pem
+ $ chmod 0600 youruserkey.pem
+
+You can also use our tool to create a userkey for you, but you need
+to specify the userkey name in confifuration file which is listed
+below. The key can be created if you use --create-key argument before
+any subcommands when you first run our tool. For more help,
+
+ $ fg-cluster --help
+
+NOTE: For more information about nova credentials, you can refer
+to tutorial at:
+
+* https://portal.futuregrid.org/tutorials/openstack
+
+
+Create configuration file
+-------------------------
+
+You need to create a configuration file which needs to be passed to
+this tool for locating necessary files in order to run. You can create
+configuration file using text editor, or using an example we provide
+to you
+
+* https://github.com/futuregrid/virtual-cluster/blob/master/etc/futuregrid.cfg
+
+It has the following format:
+
+ [virtual-cluster]
+ # Backup file for saving and loading virtual cluster(s)
+ backup = ~/.futuregrid/virtual-cluster
+ # Slurm configuration input file
+ slurm = ~/.futuregrid/slurm.conf.in
+ # userkey pem file
+ userkey = ~/PUT-YOUR-USER-NAME.pem
+ # environment file
+ enrc = ~/novarc
+ # program interface
+ interface = euca2ools
+
+You will have to modify the
+
+
+
+
+
+
+
+
+ All modules for which code is available
+ + +' + _('Hide Search Matches') + '
') + .appendTo($('#searchbox')); + } + }, + + /** + * init the domain index toggle buttons + */ + initIndexTable : function() { + var togglers = $('img.toggler').click(function() { + var src = $(this).attr('src'); + var idnum = $(this).attr('id').substr(7); + $('tr.cg-' + idnum).toggle(); + if (src.substr(-9) == 'minus.png') + $(this).attr('src', src.substr(0, src.length-9) + 'plus.png'); + else + $(this).attr('src', src.substr(0, src.length-8) + 'minus.png'); + }).css('display', ''); + if (DOCUMENTATION_OPTIONS.COLLAPSE_INDEX) { + togglers.click(); + } + }, + + /** + * helper function to hide the search marks again + */ + hideSearchWords : function() { + $('#searchbox .highlight-link').fadeOut(300); + $('span.highlighted').removeClass('highlighted'); + }, + + /** + * make the url absolute + */ + makeURL : function(relativeURL) { + return DOCUMENTATION_OPTIONS.URL_ROOT + '/' + relativeURL; + }, + + /** + * get the current relative url + */ + getCurrentURL : function() { + var path = document.location.pathname; + var parts = path.split(/\//); + $.each(DOCUMENTATION_OPTIONS.URL_ROOT.split(/\//), function() { + if (this == '..') + parts.pop(); + }); + var url = parts.join('/'); + return path.substring(url.lastIndexOf('/') + 1, path.length - 1); + } +}; + +// quick alias for translations +_ = Documentation.gettext; + +$(document).ready(function() { + Documentation.init(); +}); diff --git a/doc/build/html/_static/down-pressed.png b/doc/build/html/_static/down-pressed.png new file mode 100644 index 0000000..6f7ad78 Binary files /dev/null and b/doc/build/html/_static/down-pressed.png differ diff --git a/doc/build/html/_static/down.png b/doc/build/html/_static/down.png new file mode 100644 index 0000000..3003a88 Binary files /dev/null and b/doc/build/html/_static/down.png differ diff --git a/doc/build/html/_static/file.png b/doc/build/html/_static/file.png new file mode 100644 index 0000000..d18082e Binary files /dev/null and b/doc/build/html/_static/file.png differ diff --git a/doc/source/old/_static/header_bg.png b/doc/build/html/_static/header_bg.png similarity index 100% rename from doc/source/old/_static/header_bg.png rename to doc/build/html/_static/header_bg.png diff --git a/doc/build/html/_static/jquery.js b/doc/build/html/_static/jquery.js new file mode 100644 index 0000000..7c24308 --- /dev/null +++ b/doc/build/html/_static/jquery.js @@ -0,0 +1,154 @@ +/*! + * jQuery JavaScript Library v1.4.2 + * http://jquery.com/ + * + * Copyright 2010, John Resig + * Dual licensed under the MIT or GPL Version 2 licenses. + * http://jquery.org/license + * + * Includes Sizzle.js + * http://sizzlejs.com/ + * Copyright 2010, The Dojo Foundation + * Released under the MIT, BSD, and GPL Licenses. + * + * Date: Sat Feb 13 22:33:48 2010 -0500 + */ +(function(A,w){function ma(){if(!c.isReady){try{s.documentElement.doScroll("left")}catch(a){setTimeout(ma,1);return}c.ready()}}function Qa(a,b){b.src?c.ajax({url:b.src,async:false,dataType:"script"}):c.globalEval(b.text||b.textContent||b.innerHTML||"");b.parentNode&&b.parentNode.removeChild(b)}function X(a,b,d,f,e,j){var i=a.length;if(typeof b==="object"){for(var o in b)X(a,o,b[o],f,e,d);return a}if(d!==w){f=!j&&f&&c.isFunction(d);for(o=0;o","
"];c.fn.extend({text:function(a){if(c.isFunction(a))return this.each(function(b){var d=
+c(this);d.text(a.call(this,b,d.text()))});if(typeof a!=="object"&&a!==w)return this.empty().append((this[0]&&this[0].ownerDocument||s).createTextNode(a));return c.text(this)},wrapAll:function(a){if(c.isFunction(a))return this.each(function(d){c(this).wrapAll(a.call(this,d))});if(this[0]){var b=c(a,this[0].ownerDocument).eq(0).clone(true);this[0].parentNode&&b.insertBefore(this[0]);b.map(function(){for(var d=this;d.firstChild&&d.firstChild.nodeType===1;)d=d.firstChild;return d}).append(this)}return this},
+wrapInner:function(a){if(c.isFunction(a))return this.each(function(b){c(this).wrapInner(a.call(this,b))});return this.each(function(){var b=c(this),d=b.contents();d.length?d.wrapAll(a):b.append(a)})},wrap:function(a){return this.each(function(){c(this).wrapAll(a)})},unwrap:function(){return this.parent().each(function(){c.nodeName(this,"body")||c(this).replaceWith(this.childNodes)}).end()},append:function(){return this.domManip(arguments,true,function(a){this.nodeType===1&&this.appendChild(a)})},
+prepend:function(){return this.domManip(arguments,true,function(a){this.nodeType===1&&this.insertBefore(a,this.firstChild)})},before:function(){if(this[0]&&this[0].parentNode)return this.domManip(arguments,false,function(b){this.parentNode.insertBefore(b,this)});else if(arguments.length){var a=c(arguments[0]);a.push.apply(a,this.toArray());return this.pushStack(a,"before",arguments)}},after:function(){if(this[0]&&this[0].parentNode)return this.domManip(arguments,false,function(b){this.parentNode.insertBefore(b,
+this.nextSibling)});else if(arguments.length){var a=this.pushStack(this,"after",arguments);a.push.apply(a,c(arguments[0]).toArray());return a}},remove:function(a,b){for(var d=0,f;(f=this[d])!=null;d++)if(!a||c.filter(a,[f]).length){if(!b&&f.nodeType===1){c.cleanData(f.getElementsByTagName("*"));c.cleanData([f])}f.parentNode&&f.parentNode.removeChild(f)}return this},empty:function(){for(var a=0,b;(b=this[a])!=null;a++)for(b.nodeType===1&&c.cleanData(b.getElementsByTagName("*"));b.firstChild;)b.removeChild(b.firstChild);
+return this},clone:function(a){var b=this.map(function(){if(!c.support.noCloneEvent&&!c.isXMLDoc(this)){var d=this.outerHTML,f=this.ownerDocument;if(!d){d=f.createElement("div");d.appendChild(this.cloneNode(true));d=d.innerHTML}return c.clean([d.replace(Ja,"").replace(/=([^="'>\s]+\/)>/g,'="$1">').replace(V,"")],f)[0]}else return this.cloneNode(true)});if(a===true){ra(this,b);ra(this.find("*"),b.find("*"))}return b},html:function(a){if(a===w)return this[0]&&this[0].nodeType===1?this[0].innerHTML.replace(Ja,
+""):null;else if(typeof a==="string"&&!ta.test(a)&&(c.support.leadingWhitespace||!V.test(a))&&!F[(La.exec(a)||["",""])[1].toLowerCase()]){a=a.replace(Ka,Ma);try{for(var b=0,d=this.length;b' + _('Searching') + '
').appendTo(this.out); + this.dots = $('').appendTo(this.title); + this.status = $('').appendTo(this.out); + this.output = $('\
+ \
+
';
+
+ var commentTemplate = '\
+ Comments
\
+ \
+ loading comments... 
\
+ \
+
\
+ Add a comment\ + (markup):
\ +\
+ reStructured text markup: *emph*, **strong**, \
+ ``code``, \
+ code blocks: :: and an indented block after blank line
\
+ \
+ \
+ \
+
\
+ ';
+
+ var replyTemplate = '\
+ \
+
\
+ \
+ \ + <%username%>\ + \ + <%time.delta%>\ +
\ +<#text#>
\
+ \ + \ + reply ▿\ + proposal ▹\ + \ + \ + \ +
\ +\ +<#proposal_diff#>\ +\ +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/index.html b/doc/build/html/index.html
new file mode 100644
index 0000000..645cdab
--- /dev/null
+++ b/doc/build/html/index.html
@@ -0,0 +1,202 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ Index
+ + +A
+| + |
B
+| + | + |
C
+D
+E
+F
+| + | + |
G
+I
+| + | + |
M
+| + |
P
+| + | + |
R
+| + |
S
+T
+| + |
U
+| + | + |
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/intro.html b/doc/build/html/intro.html
new file mode 100644
index 0000000..6c96b4b
--- /dev/null
+++ b/doc/build/html/intro.html
@@ -0,0 +1,586 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+Welcome to Virtual Cluster documentation!¶
+Contents:
+
+
+
+
+
+
+
+ Indices and tables¶
+-
+
- Index +
- Module Index +
- Search Page +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/modules.html b/doc/build/html/modules.html
new file mode 100644
index 0000000..72f636f
--- /dev/null
+++ b/doc/build/html/modules.html
@@ -0,0 +1,1473 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+INTRODUCTION¶
+
+
+System requirements¶
+-
+
- euca2ools: verison 1.2 +
- python: version 2.7 +
- virtualenv (no admin rights) +
+
+Introduction¶
+Project futuregrid.virtual.cluster is a virtual cluster management +software utilizing openstack resources on FutureGrid. The software +contains several parts that could help users easily manage their +operations to virtual clusters on FutureGrid.
+Managment operations include:
+-
+
- Create virtual cluster(s) +
- Save currently running virtual cluster(s) +
- Restore saved virtual cluster(s) +
- Show status of virtual cluster(s) +
- Terminate virtual cluster(s) +
+
+HOW TO INSTALL¶
+
+
+Installation¶
+We assume that yo do not have super user priviledges on the computer +where you like to install our tool.
+### Step 0: Prerequisites (not using india.futuregrid.org machines):
+In order to make the installation process work smoothly, please make +sure that the computer you like to install our tool has already +installed euca2ools (version 1.2) and Python (version 2.7). If you do +not have those tools or correct version installed. you may install +euca2ools (version 1.2) from
+ +and install python (version 2.7) from
+ +After you check all the tools with version are correctly installed, +you may proceed with following steps to start installation.
+### Step 0: Prerequisites (using india.futuregrid.org machines):
+india.futuregrid.org has installed every tools you may need to finish this +installation, so to activate euca2ools (version 1.2) after you login into +india futuregrid machines, just simply do:
+++$ module load euca2ools
To activate python 2.7, simply do:
+++$ module load python
Those commands will help you load tools with correct version you need +to finish installation. So now you may proceed with following +installation steps.
+### Step 1: Download virtualenv
+Since you do not have super user priviledges, you need virtualenv in +order to finish the installtion. You may download virtualenv.py by +following command:
+++
### Step 2: Install virtualenv
+After you downloaded virtualenv, you can install it by following +command:
+++$ python virtualenv.py –system-site-packages ENV
### Step 3: Activate virtualenv
+After installation of virtualenv, you can activate virtualenv by +following command:
+++$ source ENV/bin/activate
### Step 4: install the virtual cluster with pip
+Once virtualenv is activated, you can use pip to install our tool by +following command:
+++$ ENV/bin/pip install futuregrid.virtual.cluster
If you already had our tool installed, and you want to upgrade to +newest version, you can do it by following command:
+++$ ENV/bin/pip install –upgrade futuregrid.virtual.cluster
NOTE: For more information about virtualenv, you may see documentation +of virtualenv on
+ +
+
+FutureGrid Specific Instalation¶
+### Install without admin rights on india futuregrid
+Go to the futuregrid portal
+ +If you do not have a futuregrid account, please apply for one at link:
+ +In order for you to get access to FutureGrid resources, you need to +apply for a portal account and create a Futuregrid project. This is in +detail explained at
+ +Do not forget to upload your public key. (see also +https://portal.futuregrid.org/generating-ssh-keys-futuregrid-access)
+Once you have a vaild portal account and a valid project, you can go +ahead and use FutureGrid
+Our virtual cluster is best executed on our machine called +india.futuregrid.org
+Please log into this machine and follow the steps that we have outlined +in the previous section to install the software and then run it while +following the instaructions from the next section
+
+
+HOW TO RUN¶
+
+
+Prerequisites¶
+In order to use our tool, you need to obatin nova credentials and +configuration files for FutureGrid system, you can obtain your nova +credentials and configuration files for the FutureGrid systems. These +should have been placed in your home directory on the INDIA +system. Log in with your FutureGrid username (and SSH public key) and +look for a file called ‘username’-nova.zip. If you do not have a +portal and HPC account please create one. The credential zip file +(username-nova.zip)contains the user keys and rc file .Unzip this +file in your hom e directory. The novarc file contains the necessary +environment variables.
+Create your private key by (Recommended: Use your user name as your +private key name):
+++$ euca-add-keypair youruserkey > youruserkey.pem +$ chmod 0600 youruserkey.pem
You can also use our tool to create a userkey for you, but you need +to specify the userkey name in confifuration file which is listed +below. The key can be created if you use –create-key argument before +any subcommands when you first run our tool. For more help,
+++$ fg-cluster –help
NOTE: For more information about nova credentials, you can refer +to tutorial at:
+ +
+
+Create configuration file¶
+You need to create a configuration file which needs to be passed to +this tool for locating necessary files in order to run. You can create +configuration file using text editor, or using an example we provide +to you
+ +It has the following format:
+++[virtual-cluster] +# Backup file for saving and loading virtual cluster(s) +backup = ~/.futuregrid/virtual-cluster +# Slurm configuration input file +slurm = ~/.futuregrid/slurm.conf.in +# userkey pem file +userkey = ~/PUT-YOUR-USER-NAME.pem +# environment file +enrc = ~/novarc +# program interface +interface = euca2ools
You will have to modify the <PUT-YOUR-USER-NAME-HERE> occurrence +within the file with the name that you use to associate your key. The +file is to be placed at ~/.futuregrid/futuregrid.cfg or concatenated +to an already existing futuregrid.cfg file. In order to use different +interface (euca2ools/boto) to run this tool, you can change interface +parameter in the configuration file to achieve that.
+NOTE: Please make sure all the files are placed under the location as +you specified in configuration file. You can also find an example of +slurm.conf.in file which is used by configuring SLURM system from
+ +You can modify SLURM configuration parameters for your customized +SLURM. But please leave “controlMachine” and “COMPUTE NODES” +untouched.
+After you finished all steps above, you can use our tool to manage +your virtual clusters
+
+
+Create a virtual cluster¶
+Run following command will create a virtual cluster of given +parameters (command example is given below):
+++$ fg-cluster -f <config-file> run -n <number-of-computation-nodes> -t <instance-type> -i <image-id> -a <cluster-name>
Parameters:
+-
+
- -f: Futuregrid configuration file named futuregrid.cfg. +
- -n: Number of computation nodes. This number of computation nodes does not include control node, so the +
actual number for virtual cluster nodes is the number of computations node plus one control node. +* -s: Instance type. Instance type includes: m1.tiny, m1.small and m1.large. +* -i: Image id. You can obtain image id by following command:
+++$ euca-describe-images
-
+
- -a: cluster name. The virtual cluster name which uniquely identifies your cluster. +
For example:
+++$ fg-cluster -f futuregrid.cfg run -n 2 -t m1.small -i ami-0000001d -a mycluster1
Virtual cluster info will be saved in backup file specified in +futuregrid configuration file. Note: Cluster name should be different +as other virtual clusters which is running if you want to run multiple +virtual clusters. If you want to use default configure file, you +should put this file at ~/.futuregrid/futuregrid.cfg, then argument -f +can be omitted
+
+
+Save a virtual cluster¶
+Run following command will save a currently running virtual cluster into one +control image and compute image for later resotre. (Installed softwares and +unfinished jobs will also be saved)
+++$ fg-cluster -f <config-file> checkpoint -c <control-node-bucket> -t <control-node-name> -m <compute-bucket> -e <compute-name> -a <cluster-name>
Parameters:
+-
+
- -f: Futuregrid configuration file +
- -c: Control node bucket name. Bucket name which you can identify control image +
- -t: Control node image name. Image name which you can use to identify your control image +
- -m: Compute node bucket name. Bucket name which you can identify your compute image +
- -e: compute node image name. Image name which you can use to identify your compute image +
- -a: Virtual cluster name +
For example:
+++$ fg-cluster -f futuregrid.cfg checkpoint -c myname -t c1.img -m myname -e c2.img -a mycluster1
If you successfully upload your control image and compute image, you +can find them in openstack image repository according to the bucker +name and image name you give to them by command:
+++$ euca-describe-images
Note: Cluster name should be a name of cluster which is +currently running. Generated image ids (including one control +node image id and one compute image id) will be registered which +are used for later restore.
+
+
+Restore a virtual cluster¶
+Run following command will restore a virtual cluster state including +installed softwares, unfinished jobs which was saved before, so that +you can continue your work from that saved point.
+++$ fg-cluster -f <config-file> restore -a <cluster-name>
Parameters;
+-
+
- -a: Cluster name. The virtual cluster name which uniquely identifies your cluster. +
For example:
+++$ fg-cluster -f futuregrid.cfg restore -a mycluster2
Note: Cluster name should be the name of cluster which had been saved +before. You can check the images you saved, the images you saved will +have the bucket name and image name you specified from checkpoint command, +and which can be shown by following command:
+++$ euca-describe-images
+
+Shutdown a virtual cluster¶
+Run following command will terminate a virtual cluster
+++$ fg-cluster -f <config-file> terminate -a <cluster-name>
Parameters:
+-
+
- -f: Futuregrid configuration file +
- -a: Virtual cluster name +
For example:
+++$ fg-cluster -f futuregrid.cfg terminate -a mycluster2
Note: Cluster name should be a name of cluster which is currently +running. After executing this command, cluster info will be removed +from backup file which is specified by configuration file
+
+
+Show status of virtual cluster(s)¶
+Run following command will show status of currently running +virtual cluster(s) including cluster size, image id, instance id, ip
+++$ fg-cluster -f <config-file> status -a <cluster-name>
Parameters:
+-
+
- -f: Futuregrid configuration file +
- -a: Virtual cluster name +
For example:
+Show status of one specific cluster given cluster name
++++
+- fg-cluster -f futuregrid.cfg status -a mycluster1
+
Show status of all currently running clusters
+++fg-cluster -f futuregrid.cfg status
Note: If argument -a is specified, then name of cluster should be +a cluster that is currently running
+
+
+List the virtual clusters¶
+Run following command will give you a list of virtual clusters and their status
+++$ fg-cluster -f <config-file> list
For example:
+++$ fg-cluster -f futuregrid.cfg list
+
+Run a simple MPI program on virtual cluster¶
+A simple MPI version of helloworld can be found at:
+ +You may use this for test purpose.
+We assume that you are using helloworld.c from above link. So in order to run this MPI program +on the cluster you created using SLURM system, you can
+### Step 1: Copy helloworld.c to HOME directory on each node in virtual cluster
+++$ scp -i <your-userkey-pem-file> helloworld.c ubuntu@<instance-ip>:~/
### Step 2: Login to instances, complie helloworld.c on each node, run:
+++$ ssh -i <your-userkey-pem-file> ubuntu@<instance-ip> +$ mpicc hellowrld.c -o helloworld
### Step 3: run MPI program, you need to login into control node
+Option 1: Using salloc command:
+++$ salloc -N 2 mpirun helloworld
where -N is the number of computation nodes you want to run with. And +should not be larger than the actual number of computation nodes
+Option 2: Using sbatch command by submitting a job script:
+++$ sbatch helloworld.sh
You can find example helloworld.sh at
+ +Execution result:
+++Running program helloworld +salloc: Granted job allocation 2 +Hello world from processor i-000023c8, rank 0 out of 2 processors +Hello world from processor i-000023c9, rank 1 out of 2 processors +salloc: Relinquishing job allocation 2
+
+Using FGClusterRunprogram¶
+A program which could help you to run a simple MPI program can be found at
+ +So you can simply run command:
+++# python FGClusterRunprogram.py -f futuregrid.cfg -p helloworld.c -n 2 -a mycluster1
Parameters:
+-
+
- -f: Futuregrid configuration file +
- -p: Program source code file +
- -n: Number of computaion nodes you want to run with. Make sure that the number you input is no larger +
than the acutal number of computaion node you created. +* -a: Name of virtual cluster you want to run program on
+Note: Virtual cluster name should be a name of cluster which is +currently running
+
+
+
+
+ FOR DEVELOPERS ONLY¶
+
+
+Generating the Distribution¶
+Assume that you have git correctly installed and configured on your +computer.
+### Step 1: You can pull source code from github by:
+++git clone git@github.com:futuregrid/virtual-cluster.git
### Step 2: Create tar file for installation
+++make pip
This creates the tar file that you can install via pip in ./dist
+### Step 3: Install
+++sudo pip install –upgrade dist/*.tar.gz
This wil install the files by default into /usr/local/bin/fg-cluster
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/objects.inv b/doc/build/html/objects.inv
new file mode 100644
index 0000000..ac21ec7
Binary files /dev/null and b/doc/build/html/objects.inv differ
diff --git a/doc/build/html/py-modindex.html b/doc/build/html/py-modindex.html
new file mode 100644
index 0000000..e7b35e3
--- /dev/null
+++ b/doc/build/html/py-modindex.html
@@ -0,0 +1,169 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ Modules¶
+CloudInstances.py (python)
+-
+
- +class futuregrid.virtual.cluster.CloudInstances.CloudInstances[source]¶ +
Class CloudInstances
+-
+
- +checkpoint_cloud_instances(cluster_name, control_node_id, compute_node_id, instance_type, cluster_size)[source]¶ +
Sets cluster saving information after actually saves it
+-
+
- Parameters: +
- cluster_name – virtual cluster name +control_node_id – control node image id +compute_node_id – compute node image id +instance_type – instance type +cluster_size – size of cluster (control node not included) +
- Logic: +
Sets all possible values that could be used to restore cluster +in the future before saves it into backup file
+Sets cluster status to SAVED
+
+
No returns
+
-
+
- +del_by_name(name)[source]¶ +
Deletes cloud instances list from backup file given name
+-
+
- Parameter: +
- name – cluster name +
- Logic: +
- Loads content from backup file, given content is list, then +delete the cluster which matches name using remove, then +dump the new content into backup file +
No returns
+
-
+
- +del_instance(instance)[source]¶ +
Deletes an instance from cloud instance list
+-
+
- Parameters: +
- instance – an instance dictionary +
- Logic: +
- Find the instance matches id then delete +the instance using key +
No returns
+
-
+
- +get_all_cloud_instances()[source]¶ +
Gets all cloud instances lists
+No parameters
+-
+
- Logic: +
- Loads all cloud instances from backup file. +
- Return: +
-
+
- cloud_list – if file exists, return all the cloud lists +
- saved +
[] – if file does not exist
+
+
-
+
- +get_by_id(cloud_id)[source]¶ +
Gets instance by index
+-
+
- Parameters: +
- cloud_id – instance index, which is also the key for instance +
- Logic: +
- Given that we use current size of cluster as key, so we can use +this index to get one specific instance +
- Return: +
-
+
- Instance dictionary – an instance which has the format of +
- {‘id’:’‘, ‘image’:’‘, ‘type’:’‘, ‘ip’:’‘} +
+
-
+
- +get_cloud_instances_by_name(name)[source]¶ +
Gets cloud instance list by cluster name
+-
+
- Parameters: +
- name – virtual cluster name +
- Logic: +
Granted we check cluster existence every time before +each operation (create, save, restore, terminate, ...) +so here no need to check if backup file is existed
+Opens backup file for reading, loads cluster lists, +find cluster instances list given cluster name, then +sets cloud_instances to the cluster loaded
+
+
No returns
+
-
+
- +get_cluster_size(cluster=None)[source]¶ +
Gets virtual cluster size
+-
+
- Parameters: +
- cluster – virtual cluster name +default: None +
- Logic: +
Default cloud instance is the dictionary self.cloud_instances, +if cluster is specified, then returns the size of cluster given +name.
+Given cluster_instance has the format of: +{‘name’:’‘, ‘status’:’‘, ‘0’:’‘, ‘1’:’‘,...} +The size of cluster is the length of this dictionary subtracts +length of name and status which is 2
+
+- Return: +
- length of self.cloud_instances – if cluster is set to default +length of cluster – if cluster is set to any other clusters +
-
+
- +get_index(instance)[source]¶ +
Gets instance key from cloud instance list
+-
+
- Parameters: +
- instance – an instance dictionary +
- Logic: +
- Finds the instance matches the id +then return the key +
- Return: +
- key – instance index +
-
+
- +get_list()[source]¶ +
Gets cloud instances list
+No parameters
+-
+
- Return: +
- cloud instances list – self.cloud_instances +
-
+
- +get_status()[source]¶ +
Gets cluster status
+-
+
- Parameters: +
- status – cluster status (run, saved, terminated) +
- Logic: +
- Get the value of key ‘status’ +
- Return: +
- Returns the value of status +
-
+
- +if_exist(name)[source]¶ +
Checks if a given cluster name exists
+-
+
- Parameters: +
- name – cluster name +
- Logic: +
- Given all clusters stored in backup file is a list, so loop +this list, and check key ‘name’ in each cluster dictionary +
- Return: +
-
+
- true – if find a cluster matches the cluster name given as +
- parameter +
- false – if 1) file does not exist +
-
+
- Cannot find a cluster matches the cluster +name given as parameter after checking +all clusters saved in the backup file +
+
+
-
+
- +if_status(status)[source]¶ +
Checks status of virtual cluster
+-
+
- Parameters: +
- status – cluster status (run, saved, terminated) +
- Logic: +
- checks if status in cluster instance list matches +status given as parameter +
- Return: +
-
+
- true – if status of cluster instance list matches +
- the status given as parameter +
- false – if status of cluster instance list does +
- not match the status given as parameter +
+
-
+
- +save_instances()[source]¶ +
Writes current cloud instances list into backup file
+No parameters
+-
+
- Logic: +
- Opens backup file to save cloud_instance +If file does not exist, then creates backup file, dumps +cloud_instances into the file +If file exists, then loads the content from the file, +combine all cloud instance into list, dump this list into +file. So the list has the format of [{cluster1},{cluster2},{}] +(may need to change it to dictionary) +
No returns
+
-
+
- +set_backup_file(filename)[source]¶ +
Sets backup file for reading and writing
+-
+
- Parameters: +
- filename – backup file name +
- Logic: +
- Checks if the backup file has the correct format by +checking if all keys are present in each dictionary. +
- Return: +
-
+
- true – If can not find backup file, so the file has not +
- been created. +
- false – If file exists, but can not find all the keys +
- which are needed. +
+
-
+
- +set_cloud_instances_by_name(name)[source]¶ +
Sets a new cloud instance
+-
+
- Parameters: +
- name – virtual cluster name +
- Logic: +
- Granted this cluster is new and about to be created, +sets cluster name to name, and sets status to run. +
No returns
+
-
+
- +set_instance(instance_id, image_id, instance_type, instance_ip='', index=None)[source]¶ +
Sets attributes of a given instance
+-
+
- Parameters: +
- instance_id – instance id +image_id – image id +instance_type – instance type +instance_ip – public IP associated with instance +default: ‘’ +
- Logic: +
- Set id, image, type, IP for one instance +Using current cluster size as key +So first instance has key 0, second has key 1 ... +
No returns
+
-
+
- +set_ip_by_id(instance_id, instance_ip)[source]¶ +
Sets IP by given instance id
+-
+
- Parameters: +
- instance_id – instance id +instance_ip – public IP associated with instance +
- Logic: +
- Loop all values in cloud_instances, given an instance +is a dictionary, if current element is a dictionary, +check key ‘id’, and then set ‘ip’ if key ‘id’ matches +
No returns
+
FGCluster command
+-
+
- +class futuregrid.virtual.cluster.FGCluster.Cluster[source]¶ +
-
+
- Methods to +
- run virtual cluster +checkpoint virtual cluster +restore virtual cluster +terminate virtual cluster +show status of virtual cluster(s) +list virtual clusters +
-
+
- +classmethod add_keypair(name, key)[source]¶ +
Adds keypair
+-
+
- Parameter: +
- name – userkey name +key – location of userkey +
- Logic: +
- Uses euca-add-keypair to create key for user, key will +be created under the location referred by key +
No returns
+
-
+
- +boto_associate_address(instance_id, address_public_ip)[source]¶ +
associate ip address using boto
+
-
+
- +boto_change_ip(instance)[source]¶ +
Changes public IP address given instance using boto +No returns
+
-
+
- +boto_start_new_instance(instance, instance_index)[source]¶ +
Starts new instance given old instance info using euca2ools
+-
+
- Parameter: +
- instance – old instance +instance_index – instance index +
- Return: +
- new instance +
-
+
- +check_avaliable(instance)[source]¶ +
Check if instance is available
+-
+
- Parameters: +
- instance – instance dictionary +
- Return: +
- True – if instance is available +False – if instance is unavailable +
-
+
- +checkpoint_cluster(args)[source]¶ +
Method for saving virtual cluster
+-
+
- Parameters: +
-
+
- args – this method deals with +
- args.name – virtual cluster name +args.controlb – control node bucket name +args.controln – control node image name +args.computeb – compute node bucket name +args.computen – compute node image name +
+- Logic; +
- Checks existence before saving virtual cluster +Only saves cluster which is running (Because saved and +terminated clusters are not currenly running) +Only saves control node and one compute node into images +Saves new control node image id and new compute node image id +into backup file for later restore. Change status to SAVED after +the saving. Then terminates cluster before deletes host information +after each termination +
No returns
+
-
+
- +clean_repo()[source]¶ +
Remove source list file
+No parameters:
+-
+
- Logic: +
- If if_default flag is set to false, then delete +source list file +
No returns
+
-
+
- +config_slurm(create_key=True)[source]¶ +
Configures slurm
+-
+
- Parameters: +
- create_key – indicates whether to create munge-key +default: true, creates key +
- Logic: +
- Reads slurm configuration input file, substitutes controlMachine +with control machine id. Append each computation node into +configuration file in COMPUTE NODES section +Generates munge-key on control node if flag is set to true +Does configuration on each node in parallel +After all threads are done, removes temp files +
No returns
+
-
+
- +copyto(instance, filename)[source]¶ +
Copies the named file to the instance
+-
+
- Parameters: +
- instance – cloud instance +filename – the name of file to copy +
No returns
+
-
+
- +create_cluster(args)[source]¶ +
Method for creating cluster
+-
+
- Parameters: +
-
+
- args – this method deals +
- args.name: cluster name +args.number: cluster compute nodes number +args.typy: instance type +args,image: image id +
+- Logic: +
Check existence before creation of cluster +If cluster was created before, only start creating +if status is DOWN
+Save each instance into cloud instance list +Associate IP with each instance +Create source list file if using IU ubuntu repository +Delete source list file after installtion and configuration +of SLURM and OpenMPI
+
+
No returns
+
-
+
- +debug(message)[source]¶ +
Method for printing debug message
+-
+
- Parameter: +
- message – message to print +
- Logic: +
- Checks debug flag, if debug flag is set to +true, then prints debug message, if is set +to false, then does not print debug message +
No returns
+
-
+
- +define_repo()[source]¶ +
Set ubuntu repo to IU ubuntu repository
+No parameters
+deb http://ftp.ussg.iu.edu/linux/ubuntu/ natty-updates main +deb-src http://ftp.ussg.iu.edu/linux/ubuntu/ natty-updates main +deb http://ftp.ussg.iu.edu/linux/ubuntu/ natty universe +deb-src http://ftp.ussg.iu.edu/linux/ubuntu/ natty universe +deb http://ftp.ussg.iu.edu/linux/ubuntu/ natty-updates universe +deb-src http://ftp.ussg.iu.edu/linux/ubuntu/ natty-updates universe +deb http://ftp.ussg.iu.edu/linux/ubuntu/ natty main +deb-src http://ftp.ussg.iu.edu/linux/ubuntu/ natty main
+-
+
- Logic: +
- If uses IU ubuntu repository, creates sources_list file +which has the former content +
No returns
+
-
+
- +classmethod del_known_host(ip_addr)[source]¶ +
Deletes known host info from ~/.ssh/known_hosts
+-
+
- Parameter: +
- ip_addr – IP address +
- Logic: +
- Deletes known host information from ~/.ssh/known_hosts +in case it shows man-in-middle-attack message when starts +different cluster but using the same IP addresses +
No returns
+
-
+
- +deploy_services(instance)[source]¶ +
Deploies SLURM and OpenMPI services
+-
+
- Parameters: +
- instance – cloud instance +
- Logic: +
- If using IU ubuntu repository, then copies source_list +to each instance in the cluster, installs SLURM and +OpenMPI on the instance +
No returns
+
-
+
- +describe_images(image_id)[source]¶ +
Gets image infomation
+-
+
- Parameters: +
- image_id – image id +
- Logic: +
- Uses euca-describe-images to get all the inforamtion of image +given image id +
- Return: +
- Command result of euca-describe-images of given image id +
-
+
- +disassociate_address(current_ip)[source]¶ +
Disassociates IP address
+-
+
- Parameters: +
- current_ip – instance IP which is currently in use +
- Logic: +
- Uses euca-disassociates-address to disassociates a +given IP address +
No returns
+
-
+
- +euca_associate_address(instance, free_ip)[source]¶ +
Associates instance with ip
+-
+
- Parameters: +
- instance – cloud instance +free_ip – free public IP address +
- Logic: +
- Associates IP with a given instance, set IP in cloud_instances +delete host information in known_hosts if it has +
- Return: +
- false: if association failed +trueL: if association succeeded +
-
+
- +euca_change_ip(instance)[source]¶ +
Changes public IP address given instance using euca2ools +No returns
+
-
+
- +euca_describe_addresses()[source]¶ +
Return list of free public IP addresses
+No parameters
+-
+
- Logic: +
- Uses euca-describe-addresses to get all the IP addresses, +filter those which are currently in use +
- Return: +
- ip_list – a list of free public IP addresses +
-
+
- +euca_run_instance(userkey, cluster_size, image, instance_type, index=None)[source]¶ +
runs instances given parameters
+-
+
- Parameters: +
- userkey – userkey name +cluster_size – number of nodes +image – image id +instance_type – instance type +index – instance key as index +default: None +
- Logic: +
- check if all instances are created +correctly. program will exit if +not all instances required are +created, and terminate those which +are running already +
No returns
+
-
+
- +euca_start_new_instance(instance, instance_index)[source]¶ +
Starts new instance given old instance info using euca2ools
+-
+
- Parameter: +
- instance – old instance +instance_index – instance index +
- Return: +
- new instance +
-
+
- +execute(instance, command)[source]¶ +
Executes a command on the instance
+-
+
- Parameters: +
- instance – cloud instance +command – shell command +
No returns
+
-
+
- +get_command_result(command)[source]¶ +
Gets command output
+-
+
- Parameters: +
- command – shell command +
- Return: +
- Command output +
-
+
- +get_kernel_id(image_id)[source]¶ +
Gets kernel id
+-
+
- Parameters: +
- image_id – image id +
- Logic: +
- Parses the command results returned by describe image method. +Given that sometimes kernel id is empty for certain images, +checks if the command returens a kernel id before return it +
- Return: +
-
+
- kernal id – kernal id is the 8th element of returned command +
- result +
+
-
+
- +get_list(_args)[source]¶ +
lists all virtual clusters and status
+-
+
- Parameters: +
-
+
- args – this method deals with +
- None +
+- Logic: +
- Reads from backup file to load all virtual cluster information +If no cloud instance saved, after shows help message, then quits +the program +
-
+
- +get_ramdisk_id(image_id)[source]¶ +
Gets ramdisk id
+-
+
- Parameters: +
- image_id – image id +
- Logic: +
- Parses the command results returned by describe image method. +Given that sometimes ramdisk id is empty for certain images, +checks if the command returens a ramdisk id before return it +
- Return: +
-
+
- ramdisk id – ramdisk id is the 9th element of returned command +
- result +
+
-
+
- +if_keypair_exits(name)[source]¶ +
Checks if key is already created
+-
+
- Parameters: +
- name – key name +
- Logic: +
- Uses euca-describe-keypairs to check the keys created, +finds name in the command result +
- Return: +
- true – if finds the key given name as parameter +false – if does not fine the key given name as parameter +
-
+
- +install(instance, packagenames)[source]¶ +
Installs the package names that are specified +(blank separated on the given instance)
+-
+
- Parameter: +
- instance – cloud instance +packagenames – software names +
No returns
+
-
+
- +installation(instance, max_retry, install=True)[source]¶ +
Checks if instances are ready to deploy and installs the +softwares on the instance which is ready
+-
+
- Parameters: +
- instance – instance dictionary +max_retry – max number of try for checking instance +install – indicates if need to the installation +default: true +
- Logic: +
ssh reomote host to check if can succeed +If install is set to true, installation of +SLURM and OpenMPI will start on VM which is ready
+Each instances associates a count of times for trying +If exceeds the max try limit, then gets free public IPs, +and associates it with a random new one, reset count to 0
+Break after all threads are done
+
+
No returns
+
-
+
- +classmethod msg(message)[source]¶ +
Method for printing help messages
+-
+
- Parameters: +
- message – message to print +
- Logic: +
- use print to print message +
No returns
+
-
+
- +parse_conf(file_name='unspecified')[source]¶ +
Parse configuration file
+-
+
- Parameters: +
- file_name – configuration file name +default: unspecified +
- Logic: +
Parse configuration file if given, default location +‘~/.futuregrid/futuregrid.cfg’. If no file is given, +finds configuration file by following order:
++
+-
+
- finds current directory ‘futuregrid.cfg’ +
- finds in default location +
configuration file format: +[virtual-cluster] +# Backup file for saving and loading virtual cluster(s) +backup = ~/.futuregrid/virtual-cluster +# Slurm configuration input file +slurm = ~/.futuregrid/slurm.conf.in +# userkey pem file +userkey = ~/PUT-YOUR-USER-NAME.pem +# environment file +enrc = ~/novarc
+Checks if all files specified in configuration +file are present.If create-key is set to true, +first check if key exists, if key does not exist, +then creates key under the location where is specified +by userkey in the configuration file
+Due to different version of this tool, back-end +structure of backup file may change, so checks +if backup file has the correct format before the start
+Checks all possible errors about configuration file
+
+
No returns
+
-
+
- +restore_cluster(args)[source]¶ +
Method for restoring cluster
+-
+
- Parameters: +
-
+
- args – this method deals with +
- args.name – virtual cluster name +
+- Logic: +
Loads control node id, compute node id, instance type, +cluster size from backup file. +Creates cluster of size one using control node id +Creates cluster of cluster size using compute node id +Associates IP addresses with all created instances +Granted cluster was saved before, so no need to install +softwares again, but need to configure SLURM accordingly +because all host names changed. No need to create mumge-key +because munge-key was also saved in each instance during the +saving process.
+Change status to RUN in the end, and save this restored cluster +infomation into backup file
+
+
No returns
+
-
+
- +save_instance(kernel_id, ramdisk_id, instance_ip, instance_name)[source]¶ +
Saves one instance
+-
+
- Parameters: +
- kernel_id: image kernel id +ramdisk_id: image ramdisk id +instance_ip: instance public IP address +instance_name: image name +
- Logic: +
- Uses euca-bundle-vol to create a bundle of current image on +a remote host. Image has the size of 1GB, and stored under /mnt/ +
No returns
+
-
+
- +save_node(image_id, instance_ip, bucket_name, image_name, node_type, out_queue)[source]¶ +
Saves nodem, upload and register
+-
+
- Parameters: +
- image_id – image id of instance +instance_ip – instance public IP address +bucket_name – bucket name for bundled image +image_name – image name for bundled image +node_type – control node or compute node +out_queue – queue in which puts output +
- Logic: +
- After gets all the necessary infomraion to save bundle a node, +then saves the node, uploads the bundle +
- Return: +
-
+
- bundled image id – Parses the command result, gets the bundled +
- image id, then returns it +
+
-
+
- +set_cloud(cloud)[source]¶ +
Set cloud
+-
+
- Parameter: +
- cloud – cloud (nova or eucalyptus) +
command line argument –cloud will +overwrite the configure in configuration file
+No returns
+
-
+
- +set_flag(args)[source]¶ +
Sets control flags
+-
+
- Parameter: +
-
+
- args – this method deals args.if_debug, if_default +
if_create_key.
+-
+
- if_debug: default is false +
- true to print debug message +false not to print +
- if_default: default is false +
- true to use default ubuntu repository +false to use IU ubuntu repository +
- if_create_key: default is false +
- true to create key for users +false not to create +
- if_boto: default is false +
- true to use boto interface +false to use euca2ools +
+
+- Logic: +
- Sets control flags +
No returns
+
-
+
- +set_interface(interface)[source]¶ +
Set interface the program to use
+-
+
- Parameter: +
- interface – interface (boto or euca2ools) +
command line argument –interface will +overwrite the configure in configuration file
+No returns
+
-
+
- +show_status(args)[source]¶ +
Show status of cluster(s)
+-
+
- Parameters: +
-
+
- args – this method deals with +
- args.name – virtual cluster name +
+- Logic: +
- Reads from backup file to load cluster information given +cluster name. If cluster name is specified, then after +checks its existence, loads it to display. If no cluster +name is specified, then loads all cluster instances to +display. If no clusters are saved in the backup file, after +prints help message, then quits the program +
No returns
+
-
+
- +shut_down(args)[source]¶ +
Method for shutting down a cluster
+-
+
- Parameters: +
-
+
- args – this method deals with +
- args.name – virtual cluster name +
+- Logic: +
Only terminates cloud instance which is not terminated +Checking its existence before termination. +Delete host info from ~/.ssh/known_hosts after each termination +Change status according to following:
++
+If current status is SAVED, then does not change status +If current status is RUN, then changes it to DOWN
+
No returns
+
-
+
- +start_slurm(instance, create_key, slurm_conf_file, munge_key_file)[source]¶ +
Copies SLURM configuration file and munge key file +to every node, and start slurm on each node
+-
+
- Parameters: +
- instance – cloud instance +create_key – flag indicates whether needs to create munge key +slurm_conf_file – SLURM configuration file name +munge_key_file – munge key file name +
- Logic: +
- Copies SLURM configuration file to each node in cluster. +If needs to create munge key, then creates munge key on +control node, copies it to each node in cluster. Starts +SLURM and munge after copy is done +
No returns
+
-
+
- +update(instance)[source]¶ +
Executes a software update on the specified instance
+-
+
- Parameters: +
- instance – cloud instance +
No returns
+
-
+
- +upload_bundle(instance_ip, bucket_name, manifest)[source]¶ +
Uploads bundle image given manifest
+-
+
- Parameters: +
- instance_ip – instance public IP address +bucket_name – bucket name for bundled image +manifest – manifest for bundled image +
- Logic: +
- Uses euca-upload-bundle to upload bundle +
No returns
+
-
+
- +futuregrid.virtual.cluster.FGCluster.commandline_parser()[source]¶ +
Parses commandline
+Checks if python version is above 2.7
+
Virtual cluster shell
+-
+
- +class futuregrid.virtual.cluster.Shell.Shell(*args, **kwargs)[source]¶ +
Virtual cluster shell command
+-
+
- +do_checkpoint(instance, arg)[source]¶ +
None +Usage: checkpoint [options] arg
+-
+
- Options: +
+
++ + + +-h, --help +show this help message and exit +-a NAME, --name=NAME +cluster name +-c CONTROLB, --controlb=CONTROLB +control node bucket +-t CONTROLN, --controln=CONTROLN +control node image +-m COMPUTEB, --computeb=COMPUTEB +compute node bucket +-e COMPUTEN, --computen=COMPUTEN +compute node image
+
-
+
- +do_config(instance, arg)[source]¶ +
config - add the configuration from a file +Usage: config [options] arg
+-
+
- Options: +
+
++ + + +-h, --help +show this help message and exit +-f FILE, --file=FILE +cluster name +-i INTERFACE, --interface=INTERFACE +interface
+
-
+
- +do_restore(instance, arg)[source]¶ +
None +Usage: restore [options] arg
+-
+
- Options: +
+
++ + + +-h, --help +show this help message and exit +-a NAME, --name=NAME +cluster name
+
-
+
- +do_run(instance, arg)[source]¶ +
None +Usage: run [options] arg
+-
+
- Options: +
+
++ + + +-h, --help +show this help message and exit +-a NAME, --name=NAME +cluster name +-n NUMBER, --number=NUMBER +number of computation nodes +-t TYPE, --type=TYPE +instance type +-i IMAGE, --image=IMAGE +image id
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/search.html b/doc/build/html/search.html
new file mode 100644
index 0000000..9e74c19
--- /dev/null
+++ b/doc/build/html/search.html
@@ -0,0 +1,148 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ Python Module Index
+ +
+ f
+
+
+ | + f | ||
| + | + futuregrid | + |
| + | + futuregrid.virtual.cluster.CloudInstances | + |
| + | + futuregrid.virtual.cluster.FGCluster | + |
| + | + futuregrid.virtual.cluster.Shell | + |
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/doc/build/html/searchindex.js b/doc/build/html/searchindex.js
new file mode 100644
index 0000000..e04a4fc
--- /dev/null
+++ b/doc/build/html/searchindex.js
@@ -0,0 +1 @@
+Search.setIndex({objects:{"futuregrid.virtual.cluster":{Shell:[2,0,1,""],CloudInstances:[2,0,1,""],FGCluster:[2,0,1,""]},"futuregrid.virtual.cluster.CloudInstances":{CloudInstances:[2,4,1,""]},"futuregrid.virtual.cluster.FGCluster.Cluster":{restore_cluster:[2,2,1,""],upload_bundle:[2,2,1,""],create_cluster:[2,2,1,""],deploy_services:[2,2,1,""],set_flag:[2,2,1,""],euca_get_ip:[2,2,1,""],define_repo:[2,2,1,""],del_known_host:[2,1,1,""],get_instance_from_reservation:[2,2,1,""],clean_repo:[2,2,1,""],euca_start_new_instance:[2,2,1,""],add_keypair:[2,1,1,""],save_node:[2,2,1,""],config_slurm:[2,2,1,""],boto_change_ip:[2,2,1,""],shut_down:[2,2,1,""],get_list:[2,2,1,""],boto_describe_addresses:[2,2,1,""],msg:[2,1,1,""],check_avaliable:[2,2,1,""],checkpoint_cluster:[2,2,1,""],set_interface:[2,2,1,""],boto_run_instances:[2,2,1,""],get_kernel_id:[2,2,1,""],euca_associate_address:[2,2,1,""],describe_images:[2,2,1,""],euca_describe_addresses:[2,2,1,""],update:[2,2,1,""],disassociate_address:[2,2,1,""],parse_conf:[2,2,1,""],copyto:[2,2,1,""],euca_register:[2,2,1,""],start_slurm:[2,2,1,""],print_section:[2,2,1,""],euca_change_ip:[2,2,1,""],terminate_instance:[2,2,1,""],boto_associate_address:[2,2,1,""],boto_start_new_instance:[2,2,1,""],execute:[2,2,1,""],set_cloud:[2,2,1,""],installation:[2,2,1,""],install:[2,2,1,""],if_status:[2,2,1,""],if_keypair_exits:[2,2,1,""],euca_run_instance:[2,2,1,""],ec2_connect:[2,2,1,""],get_command_result:[2,2,1,""],save_instance:[2,2,1,""],debug:[2,2,1,""],get_ramdisk_id:[2,2,1,""],show_status:[2,2,1,""]},"futuregrid.virtual.cluster.Shell":{Shell:[2,4,1,""]},"futuregrid.virtual.cluster.CloudInstances.CloudInstances":{if_exist:[2,2,1,""],set_instance:[2,2,1,""],checkpoint_cloud_instances:[2,2,1,""],set_backup_file:[2,2,1,""],if_status:[2,2,1,""],get_cluster_size:[2,2,1,""],clear:[2,2,1,""],save_instances:[2,2,1,""],del_by_name:[2,2,1,""],set_cloud_instances_by_name:[2,2,1,""],del_instance:[2,2,1,""],get_status:[2,2,1,""],set_ip_by_id:[2,2,1,""],set_status:[2,2,1,""],get_by_id:[2,2,1,""],get_index:[2,2,1,""],get_list:[2,2,1,""],get_all_cloud_instances:[2,2,1,""],get_cloud_instances_by_name:[2,2,1,""]},"futuregrid.virtual.cluster.FGCluster":{Cluster:[2,4,1,""],commandline_parser:[2,3,1,""]},"futuregrid.virtual.cluster.Shell.Shell":{do_restore:[2,2,1,""],do_run:[2,2,1,""],do_config:[2,2,1,""],do_checkpoint:[2,2,1,""],do_terminate:[2,2,1,""],do_status:[2,2,1,""]}},terms:{all:[1,2],code:1,dist:1,forget:1,confifur:1,mnt:2,mpirun:1,define_repo:2,edu:2,follow:[1,2],scp:1,concaten:1,content:[0,2],privat:1,middl:2,nodem:2,pem:[1,2],former:2,program:[0,1,2],present:2,under:[1,2],exit:2,set:2,ramdisk:2,sourc:[1,2],everi:[1,2],fals:2,subcommand:1,none:2,mpi:[0,1],util:1,untouch:1,look:1,list:[0,1,2],upload:[1,2],helloworld:1,"try":2,small:1,pleas:1,work:1,cfg:[1,2],second:2,pass:1,download:1,get_cluster_s:2,append:2,"0000001d":1,index:[0,1,2],unfinish:1,section:[1,2],cloud_inst:2,current:[1,2],delet:2,version:[1,2],compute_node_id:2,cloud_list:2,"new":2,method:2,bucket_nam:2,if_exist:2,gener:[0,1,2],here:[1,2],ubuntu:[1,2],euca_change_ip:2,modifi:1,sinc:1,valu:2,search:0,ahead:1,larger:1,do_config:2,queue:2,chang:[1,2],via:1,get_cloud_instances_by_nam:2,packagenam:2,appli:1,modul:[0,1,2],wil:1,do_checkpoint:2,set_flag:2,put:[1,2],leav:1,instal:[0,1,2],cloud:2,node_typ:2,from:[1,2],describ:[1,2],zip:1,"super":1,univers:2,regist:[1,2],upgrad:1,next:1,typi:2,readi:2,openmpi:2,call:1,usr:1,recommend:1,msg:2,type:[1,2],more:1,eucalyptussoftwar:1,usign:2,boto_run_inst:2,exce:2,known:2,account:1,restor:[0,1,2],outlin:1,uniqu:1,can:[1,2],"9th":2,those:[1,2],deb:2,purpos:1,get_ramdisk_id:2,address:2,control:[1,2],tar:1,give:1,process:[1,2],sudo:1,india:1,indic:[0,2],unzip:1,unavail:2,want:1,multipl:1,hom:1,write:2,how:[0,1],env:1,parse_conf:2,simpl:[0,1],updat:2,resourc:1,max:2,clone:1,after:[1,2],prerequisit:[0,1],befor:[1,2],known_host:2,mai:[1,2],end:2,disassociate_address:2,associ:[1,2],parallel:2,man:2,relinquish:1,alloc:1,github:1,classmethod:2,credenti:1,element:2,inform:[1,2],print_sect:2,combin:2,order:[1,2],ip_addr:2,help:[1,2],check_avali:2,becaus:2,paramet:[1,2],img:1,mpicc:1,reomot:2,requir:[0,1,2],main:2,boto_start_new_inst:2,them:1,set_interfac:2,"return":2,vaild:1,python:[1,2],"break":2,now:1,introduct:[0,1],sbatch:1,name:[1,2],euca_start_new_inst:2,config:[1,2],separ:2,easili:1,achiev:1,mumg:2,each:[1,2],cluster2:2,found:1,cluster1:2,"000023c9":1,just:1,blank:2,continu:1,keypair:[1,2],set_statu:2,connect:2,our:1,acut:1,cloudinst:2,out:1,variabl:1,ftp:2,shown:1,shut_down:2,max_retri:2,get_all_cloud_inst:2,print:2,occurr:1,correct:[1,2],file_nam:2,shut:2,obatin:1,given:[1,2],free:2,infomraion:2,git:1,dictionari:2,releas:1,org:1,thread:2,could:[1,2],omit:1,installt:[1,2],filter:2,length:2,place:1,first:[1,2],oper:[1,2],softwar:[1,2],instance_nam:2,do_restor:2,onc:1,set_backup_fil:2,number:[1,2],rank:1,sometim:2,alreadi:[1,2],done:2,overwrit:2,if_statu:2,gettingstart:1,open:2,size:[1,2],differ:[1,2],ip_list:2,script:1,system:[0,1],messag:2,get_list:2,checkpoint:[1,2],attack:2,master:1,termin:[1,2],store:2,editor:1,shell:2,option:[1,2],welcom:0,tool:[1,2],copi:[1,2],specifi:[1,2],copyto:2,part:1,pars:2,than:1,show_statu:2,provid:1,remov:[1,2],get_index:2,structur:2,current_ip:2,project:1,returen:2,info:[1,2],comput:[1,2],argument:[1,2],fgclusterrunprogram:[0,1],raw:1,pip:1,computaion:1,have:1,tabl:0,need:[1,2],euca:[1,2],self:2,currenli:2,del_by_nam:2,note:1,also:[1,2],exampl:1,which:[1,2],environ:[1,2],sure:1,distribut:[0,1],usernam:1,get_kernel_id:2,previou:1,statu:[0,1,2],deploi:2,"class":2,clear:2,later:[1,2],doe:[1,2],do_termin:2,show:[0,1,2],text:1,random:2,del_known_host:2,newest:1,fine:2,find:[1,2],access:1,onli:[0,1,2],submit:1,locat:[1,2],boto_change_ip:2,verison:1,control_node_id:2,explain:1,configur:[0,1,2],activ:1,should:1,controln:2,checkpoint_clust:2,local:1,controlb:2,reset:2,pypa:1,get:[1,2],euca_associate_address:2,mung:2,upload_bundl:2,compli:1,repo:2,cannot:2,ssh:[1,2],create_clust:2,truel:2,instance_ip:2,start_slurm:2,"public":[1,2],instance_id:2,remot:2,ec2_connect:2,priviledg:1,contain:1,mynam:1,where:[1,2],kernel:2,computeb:2,dump:2,computen:2,see:1,result:[1,2],arg:2,fail:2,reserv:2,best:1,config_slurm:2,correctli:[1,2],state:1,limit:2,attribut:2,accord:[1,2],kei:[1,2],image_id:2,job:1,succeed:2,cluster_inst:2,boto:[1,2],admin:1,etc:1,tutori:1,logic:2,login:1,com:1,load:[1,2],simpli:1,describe_imag:2,boto_describe_address:2,point:1,if_keypair_exit:2,chmod:1,header:2,shutdown:[0,1],linux:2,assum:1,quit:2,commandline_pars:2,due:2,empti:2,if_debug:2,tini:1,mycluster2:1,mycluster1:1,salloc:1,sources_list:2,imag:[1,2],ani:[1,2],restore_clust:2,boto_associate_address:2,"case":2,address_public_ip:2,packag:[1,2],servic:2,"while":1,get_command_result:2,abov:[1,2],error:2,loop:2,bin:1,munge_key_fil:2,vol:2,ramdisk_id:2,site:1,cluster:[0,1,2],kwarg:2,conf:[1,2],controlmachin:[1,2],add_keypair:2,kernal:2,sever:1,deploy_servic:2,commandlin:2,develop:[0,1],do_statu:2,grant:[1,2],make:1,format:[1,2],same:2,proce:1,instanc:[1,2],enrc:[1,2],get_by_id:2,document:[0,1],novarc:[1,2],finish:1,http:[1,2],resotr:1,kernel_id:2,euca_describe_address:2,get_instance_from_reserv:2,save_inst:2,portal:1,user:[1,2],without:1,command:[1,2],thi:[1,2],disassoci:2,del_inst:2,fgcluster:2,latest:1,identifi:1,execut:[1,2],set_cloud_instances_by_nam:2,obtain:1,euca_get_ip:2,"000023c8":1,openstack:1,natti:2,had:1,add:[1,2],valid:1,blob:1,input:[1,2],save:[0,1,2],match:2,ami:1,source_list:2,get_statu:2,read:2,euca_regist:2,test:1,youruserkei:1,step:1,world:1,like:1,specif:[0,1,2],filenam:2,html:1,manifest:2,necessari:[1,2],output:2,manag:1,instaruct:1,eucalyptu:2,www:1,right:1,old:2,deal:2,creation:2,back:2,hellowrld:1,unspecifi:2,smoothli:1,home:1,successfulli:1,if_default:2,virtualenv:1,cloud_id:2,"_arg":2,substitut:2,larg:1,save_nod:2,set_cloud:2,refer:[1,2],machin:[1,2],plu:1,run:[0,1,2],usag:2,if_boto:2,debug:2,host:2,set_inst:2,repositori:[1,2],wget:1,subtract:2,src:2,about:[1,2],actual:[1,2],euca_run_inst:2,terminate_inst:2,page:0,backup:[1,2],"1gb":2,instance_typ:2,euca2ool:[1,2],ussg:2,free_ip:2,within:1,slurm:[1,2],down:2,set_ip_by_id:2,been:[1,2],checkpoint_cloud_inst:2,your:[1,2],accordingli:2,processor:1,log:1,if_create_kei:2,nova:[1,2],custom:1,avail:2,start:[1,2],interfac:[1,2],includ:[1,2],instance_index:2,cluster_nam:2,infom:2,bundl:2,link:1,line:2,futuregrid:[0,1,2],"true":2,count:2,pull:1,succe:2,temp:2,possibl:2,whether:2,userkei:[1,2],bucker:1,bucket:[1,2],displai:2,directori:[1,2],below:1,slurm_conf_fil:2,image_nam:2,out_queu:2,creat:[0,1,2],certain:2,dure:2,cluster_s:2,exist:[1,2],file:[0,1,2],clean_repo:2,check:[1,2],again:2,when:[1,2],detail:1,virtual:[0,1,2],"default":[1,2],other:[1,2],futur:2,hpc:1,you:1,node:[1,2],create_kei:2,do_run:2,inforamt:2,flag:2,time:2,hello:1,"8th":2},objtypes:{"0":"py:module","1":"py:classmethod","2":"py:method","3":"py:function","4":"py:class"},titles:["Welcome to Virtual Cluster documentation!","INTRODUCTION","Modules"],objnames:{"0":["py","module","Python module"],"1":["py","classmethod","Python class method"],"2":["py","method","Python method"],"3":["py","function","Python function"],"4":["py","class","Python class"]},filenames:["index","intro","modules"]})
\ No newline at end of file
diff --git a/doc/build/source/futuregrid.rst b/doc/build/source/futuregrid.rst
new file mode 100644
index 0000000..8234495
--- /dev/null
+++ b/doc/build/source/futuregrid.rst
@@ -0,0 +1,10 @@
+futuregrid Package
+==================
+
+Subpackages
+-----------
+
+.. toctree::
+
+ futuregrid.virtual
+
diff --git a/doc/build/source/futuregrid.virtual.cluster.rst b/doc/build/source/futuregrid.virtual.cluster.rst
new file mode 100644
index 0000000..89385e3
--- /dev/null
+++ b/doc/build/source/futuregrid.virtual.cluster.rst
@@ -0,0 +1,35 @@
+cluster Package
+===============
+
+:mod:`cluster` Package
+----------------------
+
+.. automodule:: futuregrid.virtual.cluster
+ :members:
+ :undoc-members:
+ :show-inheritance:
+
+:mod:`CloudInstances` Module
+----------------------------
+
+.. automodule:: futuregrid.virtual.cluster.CloudInstances
+ :members:
+ :undoc-members:
+ :show-inheritance:
+
+:mod:`FGCluster` Module
+-----------------------
+
+.. automodule:: futuregrid.virtual.cluster.FGCluster
+ :members:
+ :undoc-members:
+ :show-inheritance:
+
+:mod:`Shell` Module
+-------------------
+
+.. automodule:: futuregrid.virtual.cluster.Shell
+ :members:
+ :undoc-members:
+ :show-inheritance:
+
diff --git a/doc/build/source/futuregrid.virtual.rst b/doc/build/source/futuregrid.virtual.rst
new file mode 100644
index 0000000..6e5b07b
--- /dev/null
+++ b/doc/build/source/futuregrid.virtual.rst
@@ -0,0 +1,10 @@
+virtual Package
+===============
+
+Subpackages
+-----------
+
+.. toctree::
+
+ futuregrid.virtual.cluster
+
diff --git a/doc/build/source/modules.rst b/doc/build/source/modules.rst
new file mode 100644
index 0000000..932e905
--- /dev/null
+++ b/doc/build/source/modules.rst
@@ -0,0 +1,7 @@
+futuregrid
+==========
+
+.. toctree::
+ :maxdepth: 4
+
+ futuregrid
diff --git a/doc/source/_themes/bootstrap/__init__.py b/doc/source/_themes/bootstrap/__init__.py
deleted file mode 100644
index 743c635..0000000
--- a/doc/source/_themes/bootstrap/__init__.py
+++ /dev/null
@@ -1,223 +0,0 @@
-import os
-
-from docutils import nodes
-
-from sphinx.locale import admonitionlabels
-from sphinx.writers.html import HTMLTranslator as SphinxHTMLTranslator
-
-class HTMLTranslator(SphinxHTMLTranslator):
- """
- Handle translating to bootstrap structure.
- """
-
- def visit_document(self, node):
- SphinxHTMLTranslator.visit_document(self, node)
-
- # used to detect that bullet_lists are for the global and page tocs
- self.is_partial = node.get('source') == '
+
+
+
+
+
+
+
+
+ Search
+
+
+
+ + Please activate JavaScript to enable the search + functionality. +
++ From here you can search these documents. Enter your search + words into the box below and click "search". Note that the search + function will automatically search for all of the words. Pages + containing fewer words won't appear in the result list. +
+ + +
+
+
+
+
-
-
-
-
-{{ super() }}
-{% endblock %}
-
-
-{# put the sidebar before the body #}
-{% block sidebar1 %}{{ sidebar() }}{% endblock %}
-{% block sidebar2 %}{% endblock %}
-
-{%- block document%}
-
-
-{{ super() }}
-{% endblock %}
-
-{%- block extrahead %}
-{{ super() }}
-
-
-
-{% endblock %}
diff --git a/download.html b/download.html
new file mode 100644
index 0000000..e840d05
--- /dev/null
+++ b/download.html
@@ -0,0 +1,130 @@
+
+
+
+
+
+
+
+
+ |
- |
-
- FutureGrid - Virtual Cluster - |
-
+
+
+
+
\ No newline at end of file
diff --git a/etc/futuregrid.cfg b/etc/futuregrid.cfg
deleted file mode 100644
index 9621cdf..0000000
--- a/etc/futuregrid.cfg
+++ /dev/null
@@ -1,18 +0,0 @@
-[virtual-cluster]
-# Backup file for saving and loading virtual cluster(s)
-backup = ~/.futuregrid/virtual-cluster
-
-# Slurm configuration input file
-slurm = ~/.futuregrid/slurm.conf.in
-
-# userkey pem file
-userkey = ~/PUT-YOUR-USER-NAME-HERE.pem
-
-# environment file
-enrc = ~/novarc
-
-# interface to use
-interface = euca2ools
-
-# cloud
-cloud = nova
diff --git a/etc/helloworld.c b/etc/helloworld.c
deleted file mode 100644
index 2bb3bc3..0000000
--- a/etc/helloworld.c
+++ /dev/null
@@ -1,27 +0,0 @@
-#include
+
+
+
+
+
+
+
+
+
+
+
+ Download¶
+The code can be downloaded from
+ +The release version can be installed via:
+pip install futuregird-virual-cluser
+
+
+
+
+
\ No newline at end of file
diff --git a/index.html b/index.html
new file mode 100644
index 0000000..ef60392
--- /dev/null
+++ b/index.html
@@ -0,0 +1,177 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ Index
+ +
+
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/intro.html b/intro.html
new file mode 100644
index 0000000..b007793
--- /dev/null
+++ b/intro.html
@@ -0,0 +1,832 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+Welcome to Virtual Cluster documentation!¶
+Contents:
+
+
+
+
+
+
+
+ Indices and tables¶
+-
+
- Index +
- Module Index +
- Search Page +
+
+
+
+
\ No newline at end of file
diff --git a/modules.html b/modules.html
new file mode 100644
index 0000000..e5930f4
--- /dev/null
+++ b/modules.html
@@ -0,0 +1,126 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+INTRODUCTION¶
+
+
+System requirements¶
+-
+
- euca2ools: verison 2.0.2 +
- python: version 2.7 +
- virtualenv (optional, if you do not have sudo rights) +
+
+Code Repository¶
+Our code and web page is maintained in github
+-
+
- Source: https://github.com/futuregrid/virtual-cluster +
- Documentation: http://futuregrid.github.com/virtual-cluster +
+
+Introduction¶
+This project contains a virtual cluster management +software utilizing OpenStack and Eucalyptus resources on FutureGrid. +The software contains several parts that could help users easily +manage their operations to virtual clusters on FutureGrid.
+Managment operations include:
+-
+
- Create virtual cluster(s) +
- Save currently running virtual cluster(s) +
- Restore saved virtual cluster(s) +
- Show status of virtual cluster(s) +
- List brief information of all created clusters +
- Terminate virtual cluster(s) +
- Run a simple MPI program +
+
+HOW TO INSTALL¶
+
+
+First Setup¶
+We assume that yo do not have super user priviledges on the computer +where you like to install our tool.
+-
+
Step 0: Prerequisites (not using india.futuregrid.org machines):
+In order to make the installation process work smoothly, please make +sure that the computer you like to install our tool has already +installed euca2ools (version 2.0.2) and Python (version 2.7). If you do +not have those tools or correct version installed. you may install +euca2ools (version 2.0.2) from
+ +and install python (version 2.7) from
+ +After you check all the tools with version are correctly installed, +you may proceed with following steps to start installation.
+
+Step 1: Prerequisites (using india.futuregrid.org machines):
++
+india.futuregrid.org has installed every tools you may need to finish this +installation, so to activate euca2ools (version 2.0.2) after you login into +india futuregrid machines, just simply do:
++$ module load euca2ools
+To activate python 2.7, simply do:
++$ module load python
+Those commands will help you load tools with correct version you need +to finish installation. So now you may proceed with following +installation steps.
+
+Step 2: Download virtualenv
++
+Since you do not have super user priviledges, you need virtualenv in +order to finish the installtion. You may download virtualenv.py by +following command:
++$ wget https://raw.github.com/pypa/virtualenv/master/virtualenv.py
+
+Step 3: Install virtualenv
++
+After you downloaded virtualenv, you can install it by following +command:
++$ python virtualenv.py --system-site-packages ENV
+
+Step 4: Activate virtualenv
++
+After installation of virtualenv, you can activate virtualenv by +following command:
++$ source ENV/bin/activate
+
+Step 5: install the virtual cluster with pip
++
+Once virtualenv is activated, you can use pip to install our tool by +following command:
++$ ENV/bin/pip install fgvirtualcluster
+If you already had our tool installed, and you want to upgrade to +newest version, you can do it by following command:
++$ ENV/bin/pip install --upgrade fgvirtualcluster
+
+
+
+Repeated Use¶
+We assume that you are using the same machine as the first time you ran our tool.
+-
+
Step 1: Environment Setup (using india.futuregrid.org machines):
++
+india.futuregrid.org has installed every tools you may need to finish this +installation, so to activate euca2ools (version 2.0.2) after you login into +india futuregrid machines, just simply do:
++$ module load euca2ools
+To activate python 2.7, simply do:
++$ module load python
+Those commands will help you load tools with correct version you need +to finish installation. So now you may proceed with following +installation steps.
+
+Step 2: Activate virtualenv
++
+You can activate virtualenv again by following command:
++$ source ENV/bin/activate
+
+
Now you can use our tool to manage your virtual clusters.
+
+
+FutureGrid Specific Installation¶
+
+
+Install without admin rights on india futuregrid¶
+Go to the futuregrid portal
+ +If you do not have a futuregrid account, please apply for one at link:
+ +In order for you to get access to FutureGrid resources, you need to +apply for a portal account and create a Futuregrid project. This is in +detail explained at
+ +Do not forget to upload your public key. (see also +https://portal.futuregrid.org/generating-ssh-keys-futuregrid-access)
+Once you have a vaild portal account and a valid project, you can go +ahead and use FutureGrid.
+Our virtual cluster is best executed on our machine called +india.futuregrid.org
+Please log into this machine and follow the steps that we have outlined +in the previous section to install the software and then run it while +following the instaructions from the next section
+
+
+HOW TO RUN¶
+
+
+
+
+
+
+Prerequisites¶
+In order to use our tool, you need to obatin nova, eucalyptus +credentials and configuration files for FutureGrid system. These +should have been placed in your home directory on the INDIA +system. If you do not have a portal and HPC account please create one. +The credential zip file contains the rc file. Unzip this file in your +home directory. The novarc (OpenStack) and eucarc (Eucalyptus) file contains +the necessary environment variables.
+Create your private key by (Recommended: Use your user name as your +private key name):
+$ euca-add-keypair youruserkey > youruserkey.pem
+$ chmod 0600 youruserkey.pem
+You can also use our tool to create a userkey for you, but you need +to specify the userkey name in configuration file which is listed +below. The key can be created if you use --create-key argument before +any subcommands when you first run our tool. For more help see also:
+$ fg-cluster --help
+However, this documentation is much more comprehensive than the help message.
+NOTE: For more information about credentials, you can refer +to tutorial at:
+* https://portal.futuregrid.org/tutorials/openstack
+and:
+* https://portal.futuregrid.org/using-eucalyptus-futuregrid
+
+
+
+
+
+
+Create configuration file¶
+You need to create a configuration file which needs to be passed to +this tool for locating necessary files in order to run. You can create +configuration file using text editor, or using an example we provide +to you
+ +It has the following format:
+[virtual-cluster]
+# Backup file for saving and loading virtual cluster(s)
+backup = ~/.futuregrid/virtual-cluster
+# Slurm configuration input file
+slurm = ~/.futuregrid/slurm.conf.in
+# userkey pem file
+userkey = ~/PUT-YOUR-USER-NAME-HERE.pem
+# environment file
+enrc = ~/novarc
+# program interface
+interface = euca2ools
+# cloud to use
+cloud = nova
+You will have to modify the <PUT-YOUR-USER-NAME-HERE> occurrence +within the file with the name that you use to associate your key. The +file is to be placed at ~/.futuregrid/futuregrid.cfg or concatenated +to an already existing futuregrid.cfg file. In order to use different +interface (euca2ools/boto) or cloud to run this tool, you can change +interface or cloud parameter in the configuration file to achieve that.
+If you want to use a different configuration file instead of changing +one configuration file back and forth, you can use argument --file +before you specify each subcommand you would like to run.:
+$ fg-cluster --file <configuration-file> <subcommands>
+NOTE: Please make sure all the files are placed under the location as +you specified in configuration file. You can also find an example of +slurm.conf.in file which is used by configuring SLURM system from
+ +You can modify SLURM configuration parameters for your customized +SLURM. But please leave "controlMachine", "ControlAddr" and "COMPUTE NODES" +untouched.
+In the configuration file, you can choose different interface for this tool to use +(euca2ool or boto), just need to change interface parameter in the configuration file. +You also need to change enrc and cloud parameters in order to switch between different +clouds (nova/eucalyptus).
+After you finished all steps above, you can use our tool to manage +your virtual clusters
+-
+
- OpenStack configuration example: +
To correctly configure the tool to use OpenStack resources, you can refer to example:
+[virtual-cluster]
+# Backup file for saving and loading virtual cluster(s)
+backup = ~/.futuregrid/virtual-cluster
+# Slurm configuration input file
+slurm = ~/.futuregrid/slurm.conf.in
+# userkey pem file
+userkey = ~/PUT-YOUR-USER-NAME-HERE.pem
+# environment file
+enrc = ~/novarc
+# program interface
+interface = euca2ools
+# cloud to use
+cloud = nova
+-
+
- Eucalyptus configuration example: +
To correctly configure the tool to use Eucalyptus resources, you can refer to example:
+[virtual-cluster]
+# Backup file for saving and loading virtual cluster(s)
+backup = ~/.futuregrid/virtual-cluster
+# Slurm configuration input file
+slurm = ~/.futuregrid/slurm.conf.in
+# userkey pem file
+userkey = ~/PUT-YOUR-USER-NAME-HERE.pem
+# environment file
+enrc = ~/eucarc
+# program interface
+interface = euca2ools
+# cloud to use
+cloud = eucalyptus
+
+
+
+
+Create a virtual cluster¶
+Run following command will create a virtual cluster of given +parameters (command example is given below):
+$ fg-cluster run -n <number-of-computation-nodes> -t <instance-type> -i <image-id> -a <cluster-name>
+Parameters:
++++
++ + + + +-n +Number of computation nodes. + +-t +Instance type. +Instance type includes: m1.tiny, m1.small and m1.large. + +-i +Image id. You can obtain image with a commandline tool such as euca-describe-images.+ + +-a +Cluster name. +The virtual cluster name which uniquely identifies your cluster.
Please note that the number of computation nodes does not include the control node, so the actual number for virtual cluster. +nodes is the number of computations node plus one control node.
+For example:
+$ fg-cluster run -n 2 -t m1.small -i ami-0000001d -a mycluster1
+Virtual cluster info will be saved in backup file specified in +futuregrid configuration file. Note: Cluster name should be different +as other virtual clusters which is running if you want to run multiple +virtual clusters.
+
+
+
+
+
+Save a virtual cluster¶
+Run following command will save a currently running virtual cluster into one +control image and compute image for later resotre. (Installed softwares and +unfinished jobs will also be saved):
+$ fg-cluster checkpoint -c <control-node-bucket> -t <control-node-name> -m <compute-bucket> -e <compute-name> -a <cluster-name> -s <image-size>
+Parameters:
++++
++ + + + +-c +Control node bucket name. Bucket name which you can identify control image + +-t +Control node image name. Image name which you can use to identify your control image + +-m +Compute node bucket name. Bucket name which you can identify your compute image + +-e +Compute node image name. Image name which you can use to identify your compute image + +-a +Virtual cluster name + + +-s +Image size, in MB
For example:
+$ fg-cluster checkpoint -c myname -t c1.img -m myname -e c2.img -a mycluster1 -s 1024
+If you successfully upload your control image and compute image, you +can find them by command:
+$ euca-describe-images
+Note: Cluster name should be a name of cluster which is +currently running. Generated image ids (including one control +node image id and one compute image id) will be registered which +are used for later restore. Please CHECK avaliable space on instances +before you save cluster, image size should be smaller than the avaliable +space on each instance of the virtual cluster you want to save.
+
+
+
+
+
+Restore a virtual cluster¶
+Run following command will restore a virtual cluster state including +installed softwares, unfinished jobs which was saved before, so that +you can continue your work from that saved point:
+$ fg-cluster restore -a <cluster-name>
+Parameters:
++++
++ + + + + +-a +Cluster name. The virtual cluster name which uniquely identifies your cluster.
For example:
+$ fg-cluster restore -a mycluster2
+Note: Cluster name should be the name of cluster which had been saved +before. You can check the images you saved, the images you saved will +have the bucket name and image name you specified from checkpoint command, +and which can be shown by following command:
+$ euca-describe-images
+
+
+
+
+Shutdown a virtual cluster¶
+Run following command will terminate a virtual cluster:
+$ fg-cluster terminate -a <cluster-name>
+Parameters:
++++
++ + + + + +-a +Virtual cluster name
For example:
+$ fg-cluster terminate -a mycluster2
+Note: Cluster name should be a name of cluster which is currently +running.
+
+
+
+
+
+Show status of virtual cluster(s)¶
+Run following command will show status of currently running +virtual cluster(s) including cluster size, image id, instance id, IP addresses:
+$ fg-cluster status -a <cluster-name>
+Parameters:
++++
++ + + + + +-a +Virtual cluster name
For example:
+Show status of one specific cluster given cluster name:
+fg-cluster status -a mycluster1
+Show status of all currently running clusters:
+fg-cluster status
+Note: If argument -a is specified, then the cluster you want to show status should be +a cluster that is currently running. If argument -a is not specified, the tool will +show the status of all virtual clusters.
+
+
+
+
+List the virtual clusters¶
+Run following command will give you a list of virtual clusters and their status:
+$ fg-cluster list
+For example:
+$ fg-cluster list
+
+
+RUN SIMPLE MPI PROGRAMS UNDER SLURM¶
+A simple MPI version of helloworld can be found at:
+ +You may use this for test purpose.
+We assume that you are using helloworld.c from above link. So in order to run this MPI program +on the cluster you created using SLURM system, you can conduct the following steps.
+-
+
Step 1: Copy helloworld.c to HOME directory on each node in virtual cluster
++
++$ scp -i <your-userkey-pem-file> helloworld.c ubuntu@<instance-ip>:~/
+
+Step 2: Login to instances, complie helloworld.c on each node, run
++
++$ ssh -i <your-userkey-pem-file> ubuntu@<instance-ip> +$ mpicc hellowrld.c -o helloworld
+
+Step 3: run MPI program, you need to login into control node
++
+Option 1: Using salloc command:
++$ salloc -N 2 mpirun helloworld
+where -N is the number of computation nodes you want to run with. And +should not be larger than the actual number of computation nodes
+Option 2: Using sbatch command by submitting a job script:
++$ sbatch helloworld.sh
+You can find example helloworld.sh at
+ +Execution result:
++Running program helloworld +salloc: Granted job allocation 2 +Hello world from processor i-000023c8, rank 0 out of 2 processors +Hello world from processor i-000023c9, rank 1 out of 2 processors +salloc: Relinquishing job allocation 2
+
+
+
+
+
+
+
+
+
+
+
+Using fg-cluster tool¶
+A much simpler way to run a MPI program is to use our tool
+You can choose different ways to run your MPI program, one way is to +use salloc command in SLURM and another way is to use sbatch command. +And you can also use our tool to achieve this.
+If you want to directly run MPI program using salloc, you can simply run command:
+# fg-cluster mpirun -p <program-source-file> -n <compute-nodes-to-use> -a <cluster-name>
+For example:
+# fg-cluster mpirun -p helloworld.c -n 2 -a mycluster1
+If you want to submit a job script to the SLURM, you can simply run command:
+# fg-cluster mpirun -p <program-source-file> -n <compute-nodes-to-use> -a <cluster-name> -s <script-name>
+For example:
+# fg-cluster mpirun -p helloworld.c -n 2 -a mycluster1 -s helloworld.sh
+Parameters
++++
++ + + + +-p +Program source code file + +-n +Number of computaion nodes you want to run with. + +-a +Name of virtual cluster you want to run program on + + +-s +Job script you would like to submit to SLURM
Make sure that the number you input is no larger than the acutal number of computaion node you created. +The virtual cluster name should be a name of cluster which is currently running.
+
+
+PERFORMANCE TEST¶
+You can use our tool to run performance test on OpenStack and Eucalyptus.
+
+
+Prerequisites¶
+In order to use our tool to run performance test on Openstack and Eucalyptus. You must have +our tool correctly installed. You can refer to the tutorial about how to run in the pervious +sections. Once you can successfully run our tool, you can proceed with the following steps.
+
+
+How to run¶
+Followings are steps you need to follow in order to successfully run performance test using our tool
+-
+
Step 1: Download our peroformance test tool
++
+You can download the performan test tool from
+ +Performance_Testall.py is the test script that you can use to run the performance test.
+ +Performance_Statistic is the data process program which could prodeuce excel sheets on data you collected
+ +Performance_Plot is the graph produce program which could prodeuce graphs for each cloud you tested
+ +Performance_Comp is the comparison graph produce program which could produce comparison graphs +between clouds you tested on
+Also, you can download our source code from github, and then you can find performance tool under +virtual-cluster/performance folder.
+NOTE: When you switch performance test between OpenStack and Eucalyptus, please make sure that you have +futuregrid.cfg file correctly configuared. See below section about switching between clouds.
+
+Step 2: Run performance test script
++
+Beaucase each test involves running a MPI program, so please download our sample MPI helloworld from
+ +and put it where you would like you run the test script.
+If you have done all the steps above, then you can run the test scripte by:
++$ python Performance_Testall.py -o <output-file> -i <image-id>
+This will run tests which involve creating different virtual clusters with various parameters, +running MPI program and terminating virtual clusters, then produces output file which contains +all the performance test result data you collected.
+When you finish performance test, you will get result like following:
++Test Name Total Time Installation Configuration Execution Termination IP association IP association fail IP change Restart +euca-m1.small-1 115.702837944 96.9913449287 6.05437302589 0.58861207962 0.159124135971 N/A N/A N/A N/A +euca-m1.small-1 111.77609396 92.9926450253 6.03100919724 0.55158996582 0.157529830933 N/A N/A N/A N/A +euca-m1.small-1 110.741933107 92.9937160015 5.04305911064 0.598108053207 0.16206908226 N/A N/A N/A N/A +nova-m1.small-1 151.426457167 134.004024982 2.22711896896 0.196369886398 1.20041799545 4.12035417557 0 0 0 +nova-m1.small-1 163.470904827 146.006072998 2.24714803696 0.179543972015 1.0476629734 4.10231184959 0 0 0 +nova-m1.small-1 153.810782194 136.004303932 2.69106817245 0.219621181488 1.00952887535 4.1146697998 0 0 0
+NOTE: The script will create clusters with size 1, 2, 4, 8, 16, 24, 32; with instance type small, +medium, large
+
+Step 3: Process performance test data
++
+Once you have done the performance test and outpus the raw data file. You can create excel sheets +or graphs using our tool.
+However, before you can proceed, you need to install numpy and matplotlib which are required by the tool:
++$ pip install numpy +$ pip install matplotlib
+Then, you can creats excel sheets by the following command:
++$ python Performance_Statistic.py -f <your-performance-test-result-file>
+This will create two excels for you which you can view via excel. One is for OpenStack data, +and the other one is for Eucalyptus data.
+The file has the following format (in plain text):
+++name,t_total_avg,t_total_min,t_total_max,t_total_stdev,t_setup_install_avg,t_setup_install_min,t_setup_install_max,t_setup_install_stdev,t_setup_configure_avg,t_setup_configure_min,t_setup_configure_max,t_setup_configure_stdev,t_execute_avg,t_execute_min,t_execute_max,t_execute_stdev,t_shutdown_avg,t_shutdown_min,t_shutdown_max,t_shutdown_stdev +euca-m1.small-2,139.072920442,113.236577034,164.909263849,25.8363434075,114.992971659,85.9957351685,143.990208149,28.9972364903,6.78620207309,6.69066214561,6.88174200058,0.095539927485,0.642117619514,0.613047122955,0.671188116074,0.0290704965595,0.256532430649,0.254338026047,0.258726835251,0.002194404602 +euca-m1.small-1,112.740288337,110.741933107,115.702837944,2.13696003674,94.3259019852,92.9926450253,96.9913449287,1.88475283095,5.70948044459,5.04305911064,6.05437302589,0.471327566829,0.579436699549,0.55158996582,0.598108053207,0.0200686125267,0.159574349721,0.157529830933,0.16206908226,0.00188028720646 +
You can also create graphs for each cloud by the following command:
++$ python Performance_Plot.py -f <your-performance-test-result-file>
+and create comparison graphs by:
++$ python Performance_Comp.py -ie <your-performance-test-result-about-eucalyptus-on-india> -se <your-performance-test-result-about-eucalyptus-on-sierra> -in <your-performance-test-result-about-openstack-on-india>
+
+
+
+Switch between OpenStack and Eucalyptus¶
+If you want to use our tool to do performance test on OpenStack and Eucalyptus, you can switch between +clouds just by changing futuregrid.cfg configuration file. You can refer to the configuration file section +to change your configuration file accodringly.
+Before you run a test script, you need to choose to a proper image (ubuntu) before each run (You can use euca2ools +to test the image before you run our script). To avoid output files overwritten, please use different file names, +and later you can use these performance output files to produce nice graphs.
+
+
+
+
+ FOR DEVELOPERS ONLY¶
+
+
+Generating the Distribution¶
+Assume that you have git correctly installed, configured on your +computer. And you also added your ssh public key on github. So you +can proceed with step 1.
+If you use machines on indiana futuregrid, you can load git by
++++module load git+
And added ssh public key on github.
+-
+
Step 1: You can pull source code from github by
++
++git clone git@github.com:futuregrid/virtual-cluster.git
+
+Step 2: Create tar file for installation
++
++make pip
+This creates the tar file that you can install via pip in ./dist
+
+Step 3: Install
++
++sudo pip install --upgrade dist/*.tar.gz
+This wil install the files by default into /usr/local/bin/fg-cluster
+
+
+
+
+
+
\ No newline at end of file
diff --git a/objects.inv b/objects.inv
new file mode 100644
index 0000000..6099798
Binary files /dev/null and b/objects.inv differ
diff --git a/performance/Performance_Comp.py b/performance/Performance_Comp.py
deleted file mode 100644
index b5ac044..0000000
--- a/performance/Performance_Comp.py
+++ /dev/null
@@ -1,210 +0,0 @@
-#!/usr/bin/python
-
-import argparse
-import os
-
-from pylab import *
-from numpy import *
-
-'''
-Performance_Comp.py (python)
--------------------------
-
-Creates charts for comparison of collected data
-
-If data collected from Eucalyptus on india and
-sierra are provided, a comparison graph
-india-euca-sierra-euca will be produced.
-
-If data collected from Eucaluptus and OpenStack
-on india are provided, a comparsion graph
-india-euca-india-nova will be prouced
-
-Description
-
-usage:
-
- python Performance_Plot.py
+
+
+
+
+
+
+
+
+
+
+ Modules¶
+
+
+
+
+
\ No newline at end of file
diff --git a/searchindex.js b/searchindex.js
new file mode 100644
index 0000000..625c6f4
--- /dev/null
+++ b/searchindex.js
@@ -0,0 +1 @@
+Search.setIndex({objects:{},terms:{all:3,code:[0,1,3],dist:3,forget:3,mpirun:3,follow:3,scp:3,concaten:3,privat:3,graph:3,program:[1,3],under:[1,3],leav:3,sourc:3,everi:3,subcommand:3,mpi:[1,3],util:3,untouch:3,t_execute_avg:3,cluser:0,eucarc:3,t_shutdown_avg:3,occurr:3,list:[1,3],upload:3,helloworld:3,small:3,pleas:[2,3],smaller:3,cfg:3,pass:3,download:[0,3],"0000001d":3,index:1,section:3,brief:3,access:3,version:[0,3],conduct:3,"public":3,gener:[1,3],t_execute_min:3,here:3,address:3,t_setup_configure_stdev:3,modifi:3,sinc:3,search:1,ahead:3,larger:3,chang:3,via:[0,3],prerequisit:[1,3],appli:3,modul:[1,3,4],wil:3,releas:0,prodeuc:3,instal:[0,1,3],total:3,cloud:[2,3],from:[0,3],describ:3,zip:3,regist:3,upgrad:3,next:3,call:3,usr:3,recommend:3,indiana:3,type:3,more:3,fgvirtualclust:3,t_shutdown_min:3,must:3,account:3,restor:[1,3],setup:[1,3],outlin:3,uniqu:3,can:[0,3],purpos:3,control:3,tar:3,give:3,process:3,sudo:3,india:3,indic:1,unzip:3,want:3,multipl:3,anoth:3,how:[1,3],env:3,instead:3,simpl:[1,3],t_total_avg:3,resourc:3,clone:3,after:3,befor:3,mai:3,data:3,relinquish:3,alloc:3,github:[0,3],credenti:3,processor:3,perviou:3,"switch":[1,3],maintain:3,environ:3,order:3,help:[2,3],paramet:3,img:3,mpicc:3,restart:3,comprehens:3,them:3,vaild:3,python:3,framework:2,now:3,introduct:[1,3],sbatch:3,name:3,easili:3,achiev:3,each:3,found:3,continu:3,keypair:3,our:[2,3],out:3,variabl:3,shown:3,space:3,content:1,t_execute_max:3,forth:3,correct:3,obatin:3,given:3,put:3,org:[2,3],could:3,installt:3,t_shutdown_stdev:3,place:3,first:[1,3],oper:3,softwar:3,directli:3,onc:3,number:3,rank:3,alreadi:3,done:3,submit:3,gettingstart:3,size:3,differ:3,sheet:3,script:3,associ:3,system:[1,3],messag:3,checkpoint:3,enrc:3,termin:3,editor:3,option:3,tool:[1,3],copi:3,specifi:3,part:3,performance_plot:3,than:3,performan:3,provid:3,project:3,comput:3,ran:3,ani:3,raw:3,computaion:3,have:3,tabl:1,need:3,euca:3,inform:3,note:3,also:3,without:3,which:3,sure:3,distribut:[1,3],previou:3,later:3,doe:3,link:3,show:[1,3],text:3,newest:3,unfinish:3,find:3,involv:3,current:3,onli:[1,3],locat:3,just:3,verison:3,explain:3,configur:[1,3],activ:3,should:3,beaucas:3,sierra:3,folder:3,local:3,futuregird:0,overwritten:3,pypa:3,variou:3,get:3,t_execute_stdev:3,compli:3,ssh:3,requir:[1,3],acut:3,priviledg:3,contain:3,where:3,view:3,commandlin:3,see:3,result:3,fail:3,outpu:3,best:3,t_setup_configure_max:3,statu:[1,3],correctli:3,state:3,between:[1,3],mynam:3,kei:3,job:3,matplotlib:3,performance_testal:3,boto:3,admin:3,howev:3,etc:3,tutori:3,login:3,com:[0,3],load:3,simpli:3,point:3,ubuntu:3,chmod:3,shutdown:[1,3],assum:3,numpi:3,been:3,much:3,tini:3,mycluster2:3,mycluster1:3,salloc:3,t_setup_install_avg:3,imag:3,argument:3,those:3,packag:3,plain:3,"while":3,abov:3,ami:3,configuar:3,metric:2,site:3,cluster:[0,1,3],conf:3,controlmachin:3,sever:3,develop:[1,3],welcom:1,grant:3,perform:[1,3],make:3,same:3,proce:3,t_setup_install_max:3,instanc:3,document:[1,3],novarc:3,finish:3,http:[0,2,3],again:3,portal:[2,3],user:3,t_setup_install_stdev:3,controladdr:3,exampl:3,command:3,thi:3,choos:3,t_total_min:3,accodringli:3,identifi:3,execut:3,when:[2,3],obtain:3,"000023c8":3,openstack:[1,3],web:3,had:3,add:3,valid:3,blob:3,input:3,save:[1,3],bin:3,format:3,t_total_stdev:3,avali:3,hpc:3,youruserkei:3,world:3,like:3,specif:[1,3],collect:3,necessari:3,output:3,manag:3,instaruct:3,eucalyptu:[1,3],www:3,right:3,performance_statist:3,back:3,hellowrld:3,virual:0,sampl:3,smoothli:3,proper:3,home:3,successfulli:3,t_setup_configure_min:3,virtualenv:3,avoid:3,pem:3,larg:3,refer:3,machin:3,plu:3,run:[1,2,3],performance_comp:3,step:3,repositori:[1,3],wget:3,"super":3,simpler:3,comparison:3,about:3,actual:3,would:3,t_setup_configure_avg:3,page:[1,3],backup:3,produc:3,euca2ool:3,within:3,slurm:[1,3],two:3,master:3,"000023c9":3,your:3,git:3,log:3,wai:3,support:2,nova:3,custom:3,start:3,interfac:3,includ:3,medium:3,form:2,t_shutdown_max:3,futuregrid:[0,1,2,3],info:3,pull:3,"default":3,userkei:3,bucket:3,below:3,problem:2,creat:[1,3],t_setup_install_min:3,exist:3,file:[1,3],pip:[0,3],work:3,check:3,resotr:3,excel:3,detail:3,virtual:[0,1,3],other:3,test:[1,3],you:[2,3],nice:3,node:3,repeat:[1,3],t_total_max:3,peroform:3,directori:3,time:3,hello:3},objtypes:{},titles:["Download","Welcome to Virtual Cluster documentation!","Support","INTRODUCTION","Modules"],objnames:{},filenames:["download","index","support","intro","modules"]})
\ No newline at end of file
diff --git a/setup.cfg b/setup.cfg
deleted file mode 100644
index c00a298..0000000
--- a/setup.cfg
+++ /dev/null
@@ -1,3 +0,0 @@
-[egg_info]
-tag_build =
-tag_svn_revision = true
diff --git a/setup.py b/setup.py
deleted file mode 100644
index 280fdb3..0000000
--- a/setup.py
+++ /dev/null
@@ -1,80 +0,0 @@
-"""FutureGrid: Creating a SLURM virtual Cluster via euca2tools
-
-This project creates a virtual cluster with the help of euca
-tools. The cluster uses SLURM as resource manager. A simple MPI
-program is included to test the functionality. of teh virtual cluster.
-"""
-
-from setuptools import setup, find_packages
-import sys, os
-
-version = '0.3.0'
-
-# due to a bug we are not including VERION.py yet
-# execfile('VERSION.py)
-
-classifiers = """\
-Intended Audience :: Developers
-Intended Audience :: Education
-Intended Audience :: Science/Research
-Development Status :: 4 - Beta
-Intended Audience :: Developers
-License :: OSI Approved :: Apache Software License
-Programming Language :: Python
-Topic :: Database
-Topic :: Software Development :: Libraries :: Python Modules
-Operating System :: POSIX :: Linux
-Programming Language :: Python :: 2.7
-Operating System :: MacOS :: MacOS X
-Topic :: Scientific/Engineering
-Topic :: System :: Clustering
-Topic :: System :: Distributed Computing
-"""
-
-if sys.version_info < (2, 7):
- _setup = setup
- def setup(**kwargs):
- if kwargs.has_key("classifiers"):
- del kwargs["classifiers"]
- _setup(**kwargs)
-
-doclines = __doc__.split("\n")
-
-
-setup(
- name='fgvirtualcluster',
- version=version,
- description = doclines[0],
- classifiers = filter(None, classifiers.split("\n")),
- long_description = "\n".join(doclines[2:]),
- keywords='FutureGrid OpenStack virtual cluster',
- maintainer='Gregor von Laszewski, Xiuwen Yang',
- maintainer_email="laszewski@gmail.com",
- author='Gregor von Laszewski, Xiuwen Yang',
- author_email='laszewski@gmail.com',
- url='https://github.com/futuregrid/virtual-cluster',
- license='Apache 2.0',
- package_dir = {'': '.'},
- packages = find_packages(exclude=['ez_setup', 'examples', 'tests']),
-
- #include_package_data=True,
- #zip_safe=True,
- #install_requires=[
- # # -*- Extra requirements: -*-
- #],
-
-
- entry_points={
- 'console_scripts': [
- 'fg-cluster = fgvirtualcluster.FGCluster:commandline_parser',
- 'fg-csh = fgvirtualcluster.FGShell:main',
- ]},
-
- install_requires = [
- 'setuptools',
- 'pip',
- 'fabric',
- 'boto',
- ],
- )
-
diff --git a/support.html b/support.html
new file mode 100644
index 0000000..ead351c
--- /dev/null
+++ b/support.html
@@ -0,0 +1,125 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ Search
+
+
+
+ + Please activate JavaScript to enable the search + functionality. +
++ From here you can search these documents. Enter your search + words into the box below and click "search". Note that the search + function will automatically search for all of the words. Pages + containing fewer words won't appear in the result list. +
+ + +
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/test/index.html b/test/index.html
deleted file mode 100644
index bbf8214..0000000
--- a/test/index.html
+++ /dev/null
@@ -1,133 +0,0 @@
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+ Support¶
+If you run into problems when using the cloud metric framework, please use our +help form at https://portal.futuregrid.org/help.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ 
+ 
+ 
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ 
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ 1 +2 +3 +4 +5 +6 +7 +8 +9 +10 +11 +12 +13 +14 +15 +16 +17 +18 +19 +20 +21 +22 +23 +24 +25 +26 +27 +28 +29 +30 +31 +32 +33 +34 +35 +36 +37 +38 +39 +40 +41 +42 +43 +44 +45 +46 +47 +48 +49 +50 +51 +52 +53 +54 +55 +56 +57 +58 +59 +60 +61 +62 +63 +64 +65 +66 +67 +68 +69 +70 +71 +72 +73 +74 +75 +76 +77 +78 +79 +80 +81 +82 +83 +84 +85 +86 +87 +88 +89 +90 +91 +92 +93 +94 +95 +96 +97 +98 +99 +100 +101 +102 +103 +104 +105 +106 +107 +108 +109 +110 +111 +112 +113 +114 +115 +116 +117 +118 +119 +120 +121 +122 +123 +124 +125 +126 +127 +128 +129 +130 +131 +132 +133 +134 +135 +136 +137 +138 +139 +140 +141 +142 +143 +144 +145 +146 +147 +148 +149 +150 +151 +152 +153 +154 +155 +156 +157 +158 +159 +160 +161 +162 +163 +164 +165 +166 +167 +168 +169 +170 +171 +172 +173 +174 +175 +176 +177 +178 +179 +180 +181 +182 +183 +184 +185 +186 +187 +188 +189 +190 +191 +192 +193 +194 +195 +196 +197 +198 +199 +200 +201 ++ |
+
+ |
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ -
+
- Terms of Service +
- Privacy +
- Security +
© 2012 GitHub Inc. All rights reserved.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/tips.rst b/tips.rst
deleted file mode 100644
index 2a1b741..0000000
--- a/tips.rst
+++ /dev/null
@@ -1,97 +0,0 @@
-TIPS
-====
-
-This file contains a number of tips to consider when doing the
-programming
-
-PICKLE
-------
-
-To save data ina persitnat file one can use either
-
-* pickle
-* ZODB
-
-Here a simple example of pickle::
-
- >>> import cPickle
- >>> strList = ['instane1','instance2']
- >>> pickleFile = open("insance.dat", "w")
- >>> cPickle.dump(strList, pickleFile)
- >>> pickleFile.close()
-
- >>> pickleFile = open("instance.dat")
- >>> unpickledList = cPickle.load(pickleFile)
- >>> print unpickledList
- ['instance1', 'instance2']
- >>> pickleFile.close()
- >>>
-
-Naturally we would use a dict to store that
-
-If you like to use a calss like save method, you could use
-
-ZODB
-
-a good totorial is here:
-http://www.zodb.org/documentation/tutorial.html
-
-SSH-ADD
-=======
-
-to avoid typing in all the time passwords, i recommend you use ssh add
-and keyagent
-
-a good tutorial on this is posted at http://mah.everybody.org/docs/ssh
-
-CREATING A KEY
---------------
-
-::
-
- >>>> ssh-keygen -t dsa -f ~/.ssh/id_dsa -C "label f the key"
-
-REMOTE HOST PREPARATION
------------------------
-
-Make sure the public key is in the ~/.ssh/authorized_keys file on the
-hosts you wish to connect to::
-
- >>>> cat ~/.ssh/id_dsa.pub | ssh you@remote-host 'cat - >> ~/.ssh/authorized_keys'
-
-ADDING TO YOUR BASH
--------------------
-
-add the following to your .bash_profile or .bashrc file
-
-It detects if the agent is running stats it and asks you for your
-passphrase. any other shell will than use the same agent and you do
-not have to retype you phrase till you have closed the shell wher you
-started it::
-
- SSH_ENV="$HOME/.ssh/environment"
-
- function start_agent {
- echo "Initialising new SSH agent..."
- /usr/bin/ssh-agent | sed 's/^echo/#echo/' > "${SSH_ENV}"
- echo succeeded
- chmod 600 "${SSH_ENV}"
- . "${SSH_ENV}" > /dev/null
- /usr/bin/ssh-add;
- }
-
-# Source SSH settings, if applicable
-
-::
-
- if [ -f "${SSH_ENV}" ]; then
- . "${SSH_ENV}" > /dev/null
- #ps ${SSH_AGENT_PID} doesn't work under cywgin
- ps -ef | grep ${SSH_AGENT_PID} | grep ssh-agent$ > /dev/null || {
- start_agent;
- }
- else
- start_agent;
- fi
-
-
Markdown Cheat Sheet
+ +
+
+
+
+
+
+
+
+
+
+ Format Text
+Headers
++# This is an <h1> tag +## This is an <h2> tag +###### This is an <h6> tag+
Text styles
++*This text will be italic* +_This will also be italic_ +**This text will be bold** +__This will also be bold__ + +*You **can** combine them* ++
+
+ Lists
+Unordered
++* Item 1 +* Item 2 + * Item 2a + * Item 2b+
Ordered
++1. Item 1 +2. Item 2 +3. Item 3 + * Item 3a + * Item 3b+
+
+ Miscellaneous
+Images
++ +Format:  ++
Links
++http://github.com - automatic! +[GitHub](http://github.com)+
Blockquotes
++As Kanye West said: + +> We're living the future so +> the present is our past. ++
Code Examples in Markdown
+
+
+ Syntax highlighting with GFM
+
+```javascript
+function fancyAlert(arg) {
+ if(arg) {
+ $.facebox({div:'#foo'})
+ }
+}
+```
+
+
+ Or, indent your code 4 spaces
++Here is a Python code example +without syntax highlighting: + + def foo: + if not bar: + return true+
+
+ Inline code for comments
++I think you should use an +`<addr>` element here instead.+
\ + Sort by:\ + best rated\ + newest\ + oldest\ +
\ +\ +
Add a comment\ + (markup):
\ +