I'm having the same issues, and have put a couple of band-aids in place as workarounds. Would moving jabber from berkely db to mysql or postgres help this?
From: [email protected] [mailto:[email protected]] On Behalf Of Stephen Benjamin Sent: Wednesday, February 10, 2010 10:04 PM To: [email protected] Subject: Re: [Spacewalk-list] Endless osad issues Josh, Thanks, I think the cronjob will work for the dead osad problem, and my idle fix will keep them from going into the phantom connection state. So far our deployment has been only a few hundred servers with Satellite, I just worry about this solutions scalability. Even 2-3,000 servers constantly re-connecting every 10 minutes when the idle timeout hits or the cronjob restarts the service seems like a bad idea. - Steve On Wed, Feb 10, 2010 at 8:54 PM, <[email protected]<mailto:[email protected]>> wrote: Steve, I was able to resolve some similar issues the other day and will update the wiki sometime tomorrow hopefully. http://www.mail-archive.com/[email protected]/msg03702.html I also have a cronjob on all my clients that restarts osad and runs rhn_check roughly every 4 hours randomly. I also have rhnsd disabled on the clients. It's a lotta crap, but it has been working flawlessly for several weeks. Scheduled actions get picked up on 250+ servers within 5 mins. Let me know if this works for you. I'll try to update the wiki with everything hopefully by this weekend. Good luck -Josh ----- Original Message ----- From: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> To: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> Sent: Wed Feb 10 20:03:17 2010 Subject: Re: [Spacewalk-list] Endless osad issues Hi Steve Unfortunately, I can't offer any useful advise other than to log this with RedHat GSS. If lots of people do this then maybe the problem will be fixed sooner. FWIW, I'd like func to replace osad sooner rather than later also. There used to be an entry on the roadmap (https://fedorahosted.org/spacewalk/roadmap) with func on the timeline but it's gone now. CC On Thu, Feb 11, 2010 at 8:33 AM, Stephen Benjamin <[email protected]<mailto:[email protected]>> wrote: > I inherited a Satellite system with a couple of proxies, and have had > endless issues with osad. I finally fixed it and it was working well for a > whole week. The exact issue I was having was this: > https://fedorahosted.org/spacewalk/wiki/JabberAndOSAD. Changed the idle > timeout to 600 and except for an increase in traffic, it works ok, but not > great (since at disconnect, the osads wait up to 120 seconds to reconnect, > so effectively giving my osads 80-90% uptime -- better than the previous of > near-zero). > > Now I came in yesterday and osad on 300+ boxes were just dead - service osad > status showed pid file existed, but osad not running. No idea why, no logs > were generated on any of those boxes or on the Satellite server. Jabberd's > were fine. Any ideas? > > osad is really unreliable, and I'm not the only one. Is there any priority > on replacing it with Func someday? > > > > - Steve > > -- > Stephen Benjamin > [email protected]<mailto:[email protected]> > > _______________________________________________ > Spacewalk-list mailing list > [email protected]<mailto:[email protected]> > https://www.redhat.com/mailman/listinfo/spacewalk-list > -- RHCE#805007969328369 _______________________________________________ Spacewalk-list mailing list [email protected]<mailto:[email protected]> https://www.redhat.com/mailman/listinfo/spacewalk-list _______________________________________________ Spacewalk-list mailing list [email protected]<mailto:[email protected]> https://www.redhat.com/mailman/listinfo/spacewalk-list
_______________________________________________ Spacewalk-list mailing list [email protected] https://www.redhat.com/mailman/listinfo/spacewalk-list
