2014-10-20 8:01 GMT+02:00 Rainer Gerhards <[email protected]>:
> I have an idea... some time ago we had problems that the child process > (the one backgrounded ) had problems signaling the parent (startup process) > that it finished initializing. This currently (and historically has been) > done via kill(). I think I'll give it a try to move it to some other > method. I'll try to work on this today. It would be great if Whissi could > try out any patch that will result from that work... > > While implementing this, I think I have found a potential cause. It's different from what I thought, but it can lead to non-signaling the startup process, so that it hangs. In any case, I'll now also change the way the communication is handled. I am half through in any case ;) Rainer > Rainer > > 2014-10-20 6:48 GMT+02:00 David Lang <[email protected]>: > >> The only time I have seen this happen is when there are a LOT of messages >> being generated at boot time, and rsyslog fills it's queue and can't >> deliver the messages (frequently because they're trying to go to a database >> or remote machine that's not up yet) >> >> If you configure impstats to write directly to a file, you can see what >> the queue sizes are like at startup, both normally and when you have a >> problem. >> >> If the queue fills up you can't login because ssh and login will try to >> log the fact that you are logging in (especially if you are trying to get >> to root) and block until the log is accepted by the syslog daemon. >> >> the default queue size is pretty large (10k or 100k messages IIRC), but I >> can see this being exceeded on systems that have either very verbose device >> drivers or some ongoing problem. >> >> David Lang >> >> On Sun, 19 Oct 2014, Thomas D. wrote: >> >> Date: Sun, 19 Oct 2014 19:29:45 +0200 >>> From: Thomas D. <[email protected]> >>> Reply-To: rsyslog-users <[email protected]> >>> To: [email protected] >>> Subject: Re: [rsyslog] 1/30 Reboots will fail due to rsyslog >>> >>> >>> Hi, >>> >>> I created a VM to trigger this and it works very well. >>> >>> But only if I don't start with "-dn". In debug mode, when rsyslogd won't >>> fork, I am yet unable to trigger this (tried about 80 restarts). >>> >>> When rsyslog forks on start triggering the hang works well, but because >>> I don't have a shell I cannot do anything :/ >>> But the system reacts on CTRL+ALT+DEL... >>> >>> Any idea what I can do to track this down? >>> >>> >>> -Thomas >>> _______________________________________________ >>> rsyslog mailing list >>> http://lists.adiscon.net/mailman/listinfo/rsyslog >>> http://www.rsyslog.com/professional-services/ >>> What's up with rsyslog? Follow https://twitter.com/rgerhards >>> NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad >>> of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you >>> DON'T LIKE THAT. >>> >>> _______________________________________________ >> rsyslog mailing list >> http://lists.adiscon.net/mailman/listinfo/rsyslog >> http://www.rsyslog.com/professional-services/ >> What's up with rsyslog? Follow https://twitter.com/rgerhards >> NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad >> of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you >> DON'T LIKE THAT. >> > > _______________________________________________ rsyslog mailing list http://lists.adiscon.net/mailman/listinfo/rsyslog http://www.rsyslog.com/professional-services/ What's up with rsyslog? Follow https://twitter.com/rgerhards NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE THAT.
