Environment
- CP version: v0.9.4 (
f925e5f6); libpcap 1.10.4
Description
The libpcap capturer fails task creation when pcap_activate fails (libpcap.c: if (pcap_activate(p) < 0) { error_format(...); goto error; }). Interface flapping (network failures, container network rebuilds) therefore makes the task fail to start rather than survive and wait for the interface to come back; it is not retried after recovery either.
Steps to reproduce
ip link set <if> down.- Create a
libpcap task.
- Observe
pcap_activate error: That device is not up and no task; nothing retries after the interface returns.
Expected
Make it configurable: "fail task creation on error" vs "keep the task and wait for the interface".
Please confirm
Is failing hard intentional? Some deployments expect the task to survive a brief interface-down. For comparison, the Rust port keeps the task alive and retries with a 1ms->100ms backoff.
中文原文
测试环境
- CP 版本：v0.9.4（
f925e5f6）；libpcap 1.10.4
问题描述
libpcap capturer 在 pcap_activate 失败时直接让 task 创建失败（libpcap.c 中 if (pcap_activate(p) < 0) { error_format(...); goto error; }）。接口 flapping（网络故障、容器网络重建期间常见）会让该 task 直接不启动，而不是存活等待接口恢复；接口恢复后也不会自动重试。
重现方法
ip link set <if> down；- 创建
libpcap 任务；
- 观察
pcap_activate error: That device is not up 且任务未创建。
期望
可配置为"创建失败即报错"或"保持任务、等待接口恢复"。
请确认
当前"直接失败"是刻意设计吗？（有现场会希望接口短暂 down 时 task 存活。）对照：移植实现让 task 存活并按 1ms→100ms 退避重试（差异记于 PARITY.md §2.6）。
Environment
f925e5f6); libpcap 1.10.4Description
The libpcap capturer fails task creation when
pcap_activatefails (libpcap.c:if (pcap_activate(p) < 0) { error_format(...); goto error; }). Interface flapping (network failures, container network rebuilds) therefore makes the task fail to start rather than survive and wait for the interface to come back; it is not retried after recovery either.Steps to reproduce
ip link set <if> down.libpcaptask.pcap_activate error: That device is not upand no task; nothing retries after the interface returns.Expected
Make it configurable: "fail task creation on error" vs "keep the task and wait for the interface".
Please confirm
Is failing hard intentional? Some deployments expect the task to survive a brief interface-down. For comparison, the Rust port keeps the task alive and retries with a 1ms->100ms backoff.
中文原文
测试环境
f925e5f6）；libpcap 1.10.4问题描述
libpcap capturer 在
pcap_activate失败时直接让 task 创建失败（libpcap.c中if (pcap_activate(p) < 0) { error_format(...); goto error; }）。接口 flapping（网络故障、容器网络重建期间常见）会让该 task 直接不启动，而不是存活等待接口恢复；接口恢复后也不会自动重试。重现方法
ip link set <if> down；libpcap任务；pcap_activate error: That device is not up且任务未创建。期望
可配置为"创建失败即报错"或"保持任务、等待接口恢复"。
请确认
当前"直接失败"是刻意设计吗？（有现场会希望接口短暂 down 时 task 存活。）对照：移植实现让 task 存活并按 1ms→100ms 退避重试（差异记于
PARITY.md §2.6）。