Expected behaviour

Same performance between client / server on version 5.1 vs 5.2

Actual behaviour

We have upgraded some days ago to phpredis 5.2.1. After the upgrade we have seen that the CPU usage on our Redis servers have increased a lot (between 30-60%). Last night we got into trouble and narrowed the issue down to the phpredis upgrade.

We have downgraded all servers to 5.1.1. and we directly see the positive impact on our Redis servers. It looks like something has changed on persistent connection handling.

In version 5.2.1 we see a lot of echo requests with phpredis_pool, which doesn't exist on 5.1.1. The issue seems to come from here, at least its one visible aspect. In 5.1.1 we also do not see any "Ping" comments to keep alive the persistent connection
a5f9592

This graphs shows the CPU usage of one Redis server. Until 10:00 it was 5.2.1 and then downgraded to 5.1.1. Nothing else has changed application wise.

Connection is done like this:
$this->db_norepli2->pconnect('elasticache-hostname', 6379, 10, NULL, 20);

I'm seeing this behaviour on

• OS:
• Redis:
• PHP:PHP 7.3.11 (cli) (built: Oct 31 2019 18:30:52) ( NTS )
  Copyright (c) 1997-2018 The PHP Group
  Zend Engine v3.3.11, Copyright (c) 1998-2018 Zend Technologies
  with Zend OPcache v7.3.11, Copyright (c) 1999-2018, by Zend Technologies

Server

redis_version:5.0.4

Clients

connected_clients:16833
client_recent_max_input_buffer:2
client_recent_max_output_buffer:0
blocked_clients:0

instantaneous_ops_per_sec:33981

• phpredis: 5.2.1 vs 5.1.1

Steps to reproduce, backtrace or example script

I've checked

• There is no similar issue from other users
• Issue isn't fixed in develop branch

I have found this issue which might be related #1668
We never had any connection issues on persistent connections with version 5.1.1.