[poudriere] poudriere non-responsive, zombie sh

From: O. Hartmann <ohartmann_at_walstatt.org>
Date: Mon, 14 Aug 2017 08:19:10 +0200
Running FreeBSD 12.0-CURRENT #87 r322472: Sun Aug 13 21:59:36 CEST 2017 amd64
with jail of the same revision, lately the poudriere build system started to
get inresponsive when hitting Ctrl-C or, very often, starts to stop when
showing up which package is deleted or has to be rebuild due to changed
dependencies. usually, the list of deleted/to-be-rebuild packages show up and
then the output flows as packages are build. This stops somehow in the middle
of the output.

Checking the box then via ps/top, I see the a "sh" eating up a tremendous
portion of the CPU time. I have a 4 core/8 threads XEON (IvyBridge based) with
16 GB of RAM using ZFS on a RAIDZ for the poudriere stuff (which induced never
problems in the past). 

When havin hit the Ctrl-C key, there are only two jails left not dying, I have
to use "poudriere jail -k" to kill the jail. But then, the zombie-shell (sh)
remains eating CPU time - no idea wht the shell is doing so far.

This strange behaviour occured within the last two weeks on several poudriere
hosts the same time with unchanged configurations working prior to this
observation.

Waiting long enough - in some cases hours! - the shell will finally die (after
Ctrl-C). I haven't checked whether the poudriere jobs work to the end in the
back when not showing progress on the terminal, I got impatient after a couple
of hours and stopped them.

Seems therer is an issue lately introduced.

Can someone shed some light on it? The problem is erratic - I can not easily
reproduce it, and I also can not say whether it is a ZFS or shell or kernel
issue.


kind regards,

Oliver
Received on Mon Aug 14 2017 - 04:19:14 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:41:12 UTC