2015-03-31 12:27 GMT+02:00 Rainer Gerhards <rgerha...@hq.adiscon.com>: > 2015-03-31 1:25 GMT+02:00 David Lang <da...@lang.hm>: >> On Wed, 25 Mar 2015, Rainer Gerhards wrote: >> >>> 2015-03-25 12:06 GMT+01:00 Rainer Gerhards <rgerha...@hq.adiscon.com>: >>>> >>>> 2015-03-25 11:49 GMT+01:00 Rainer Gerhards <rgerha...@hq.adiscon.com>: >>>>> >>>>> 2015-03-24 23:22 GMT+01:00 David Lang <da...@lang.hm>: >>>>>> >>>>>> when rsyslog gets hupped (closing it's outputs), omprog currently does >>>>>> nothing, so programs that it's running keep writing to anything they >>>>>> have >>>>>> open. >>>>>> >>>>>> there should be an option so that when rsyslog is sent a hup, it sends >>>>>> a hup >>>>>> to the binaries it's managing via omprog. >>>>> >>>>> >>>>> sounds like a great idea. I begun to implement it, but notice that the >>>>> module interface is missing some capabilities. Right now, I don't can >>>>> enumerate the worker instance data pointers. So it's a larger change. >>>>> Not sure if I get it done today. If you don't hear back, please ping >>>>> me mid next week. >>>> >>>> >>>> I added a tracker just in case. It also has some more info: >>>> >>>> https://github.com/rsyslog/rsyslog/issues/281 >>>> >>> >>> David, >>> >>> things progressed better than I thought. Could you please try this branch: >>> >>> https://github.com/rgerhards/rsyslog/tree/rger-omprog-hup >>> >>> This should activate HUP forwarding (default is off): >>> >>> action(type="omprog" binary="..." hup.forward="on") >>> >>> I've only given it very brief testing. The patch set also needs some >>> cleanup and doc. I'll do that as soon as time is left. In the >>> meantime, some feedback would be great. >> >> >> I haven't had a chance to test this, but I discovered that the signal to be >> passed along is not always HUP. >> >> In the case of sec, HUP is a full restart that breaks correlations, USR2 is >> log rotation (closing all outputs) >> >> so instead of hup.forward="on" it will need to be something like >> signal.rotate="USR2" > > OK, I'll see I change it that way. Makes sense.
David, I think it makes sense to limit the signals that can be sent. I'd say to the set HUP, INT, USR1, USR2, TERM (maybe). Are others required? If so, why? Or do we need the full set? Feedback appreciated, Rainer > Rainer > >> >> more complex (and I'm not sure this is the best way). But we do want to be >> sure of killing children when we exit, and we need to have a way of >> specifying what to do for log rotation. > >> >> >> David Lang >> _______________________________________________ >> 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.