On Wed, Jan 29, 2014 at 1:02 PM, Igor Sverkos <[email protected]>wrote:
> Hi, > > on slide 6 there's "ter-minal" (not sure about the dash). > > will see, I think I tried to break it so that it fits... > I am not yet familiar with rsyslog plugins but after reading the > slides and what you have written about threading it sounds like > FASTCGI which will raise some questions: > > I don't know exactly about FASTCGI. > 1) In your example you open a file for writing, how does that play > nice with your threading (e.g. spawning multiple instances)? > > good point. I tried to use a very simple sample that everyone understands. Obviously, writing to a file, without appending, always causes trouble. Maybe I should add a comment about this. I am hesitant to put connecting to a more realistic output destination into the skeleton, as I am not sure everyone understands it. Any suggestions? > 2) I expect that any output plugin wants to ensure that the data will > be written in the right order (you don't want a log message from 12:30 > in your data after you wrote a log message from 12:31, do you?). Next > to the locking problem from #1, could this be a problem when rsyslog > will use multiple instances? E.g. is that more like a middleware and > the final data store has to ensure proper ordering? > > "proper ordering" you never have when you process on multiple threads. Here are the details: http://www.gerhards.net/download/LinuxKongress2010rsyslog.pdf See chapter 7. 3) If my FASTCGI comparison is right: First there was mod_fastcgi... > then there was a need for mod_fcgid because spawning a new process for > each request proved to be a problem. Does the output plugin system > work the same, so will it spawn a new instance of the output plugin > for each message? Is it a problem in rsyslog? If not, why? :) > > no. It will spawn the plugin once, and then continuesly shuffle data to it via stdin. So it continues to run. If we ignore multiple worker threads for a bit, the plugin will only be terminated when rsyslog itself is terminated. > I hope this was the feedback you were looking for. If not, I am sorry > but this were the first things which came to my mind after reading the > slides. > > Yeah, everything that helps. I'll wait for some more feedback and will see that I get all of this into an update! Thanks, Rainer > > -- > Regards, > Igor > _______________________________________________ > 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.
