nexus/sreweekly/markdown/130/07-relp-100-cpu-rsyslog-stop-after-start-issue-13-rsyslog-librelp-github.md

relp 100% cpu – rsyslog stop after start · Issue #13 · rsyslog/librelp · GitHub

• 期号: SRE Weekly Issue #130（2018-07-15）
• 作者: Full disclosure: Fastly, my employer, is mentioned.
• 链接: https://github.com/rsyslog/librelp/issues/13

简介

I had some fun this week unearthing the cause for the chronic lockups in Rsyslog that we’ve experienced at work. I found the cause (overeager retries of socket writes) and put together a bug report and a pull request.

正文

Hi,

The issue i have faced:

• 100% on rsyslog server with imrelp, and just imrelp ad addiction to the standard debian jessie or debian wheezy configuration. But the problem seems to be the client with 1.0.0 librelp installed.

The problem is quite described in the two links below:

http://lists.adiscon.net/pipermail/rsyslog/2013-July/033303.html

http://bugzilla.adiscon.com/show_bug.cgi?id=208

I have tested the patch below, and this won't hit the cpu of the collector.

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=771218

2 problems:

1. i have not found the patch for librelp elsewhere, can you attach this one to the bug 208 if you think is relevant?
2. is possible to develop a patch for the server, in order to avoid this kind of behavior or at least put warnings on logs?

Thanks,

Daniele