Issue #10418 has been updated by Jo Rhett.
So apparently the listen bit is a factor, according to discussion in that bug. Apparently when ruby has only a single file open, it just opens the file in /proc which works fine. But if more than one file will be open (ie, the listen socket plus the /proc file) then ruby uses select() on the file instead which is what trips this bug. So yeah, disabling listen might relieve the symptoms a bit. I'll add you to the bug if I can. Erm, no. Yeah, they locked the bug down to just redhat internal plus reporter and CC list, and I don't seem to be able to edit the bug. Good news is that they tested a new kernel this morning that appears to fix the problem: (remainder is copy/paste from there) David Howells 2011-11-16 05:14:07 EST I've put a test kernel with the patch applied for download at: http://people.redhat.com/~dhowells/.067ac120438d738257e2a305a3ddac64/kernel-2.6.18-298.el5.bz751214.1.x86_64.rpm [reply] [-] Comment 31 Dmitry Zamaruev 2011-11-16 05:53:06 EST I could confirm that with given kernel test passes: [root@app ~]# uname -r 2.6.18-298.el5.bz751214.1 [root@app ~]# ./test /proc/uptime 153.99 137.31 And Ruby applications (Chef/Shef in my case) which hung on 274.7.1 - works as expected on this kernel. ---------------------------------------- Bug #10418: "Caught TERM; calling stop" with state/puppetdlock left in place https://projects.puppetlabs.com/issues/10418 Author: Jo Rhett Status: Investigating Priority: Normal Assignee: Category: agent Target version: Affected Puppet version: 2.6.12 Keywords: enabledisable Branch: Mon Oct 31 23:03:31 +0000 2011 Puppet (notice): Caught TERM; calling stop Ever since the 2.6.12 upgrade I've been seeing these reports reach us. As in, about a hundred of a half thou machines. Most of the time we find that $vardir/state/puppetdlock is in place and blocking further puppet runs, which requires a manual resolution. I wrote a quick cron script to look for puppetdlock files older than one hour, remove them and mail me a report and I've received several dozen in the last few hours. Something is clearly broken in 2.6.12, we are backgrading our systems to 2.6.11. No-- I have no other information than that it crosses all of our machine types, and we have had no significant changes in our modules in this time period. Many of the machines which have failed have had zero module or manifest changes which would apply to them. I cannot get this to replicate on the command line. -- You have received this notification because you have either subscribed to it, or are involved in it. To change your notification preferences, please click here: http://projects.puppetlabs.com/my/account -- You received this message because you are subscribed to the Google Groups "Puppet Bugs" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/puppet-bugs?hl=en.
