2015-06-04 10:34 GMT+02:00 Rainer Gerhards <[email protected]>: > 2015-06-03 21:36 GMT+02:00 Thomas D. <[email protected]>: >> Hi, >> >> Tim Eifler wrote: >>> under some versions we're experiencing a strange problem if rsyslog should >>> use the write to database action. The strange thing is that this error >>> first occurs under the version v8.10.0 but we've not changed anything >>> related to MySQL. >>> >>> [...] >>> >>> I was able to produce this problem under RHEL 7 with the following >>> versions: >>> v8-stable - (Official Repository) >>> v8.11.0.master - (Tarball of the master - Official Git) - I create a >>> tarball via make dist and compiled it as the normal git version >>> v8.10.0 - (Tarball) >>> >>> It works under the following versions: >>> v8.11.0.master - (Master Branch - Official Git) >>> V8.9.0 - (Tarball) >>> >>> I can't explain why it not works with the tarball of the git master >>> version.. Please let us know if you have an idea why this error occurs. >> >> Please post URLs you are using to get the version, because it is unclear >> which version you are really using >> >> - Github can create a tarball for you >> >> - There's a tarball uploaded to GitHub >> >> - There's a release tarball on rsyslog.com ... >> >> - Something else? > > > do: > > git clone rsyslog > git checkout master > ... build from here --> works > > make distcheck > extract tarball created by make distcheck > ... build from here --> does not work > > Now that I write this I remember some patches from you to check wether > this is git or not. Probably we should try to reverse them to see if > that changes the picture. In the light of above, this seems to make > quite some sense (I didn't realize that before I just wrote it up). > > Else I'll probably ask Tim to git bisect until hopefully something comes up.
We finished the bisect. This is the commit in questions: https://github.com/rsyslog/rsyslog/commit/23c5b3e1e62ee2daf5cce61c7e80f5e20898c03b I replaced deprecated GnuTLS functions with that commit. I do not see any direct relation to why it makes ommysql fail if compiled from tarball, but there obviously is some. The problem goes away if during configure we --disable-gnutls and reapears reproducible when doing --enable-gnutls. If I remove the two lines added to configure.ac with that commit, ommysql works well. The strange thing is that GnuTLS does not even need to be used in an config to make the issue appear. It's sufficient if the option is enabled. It looks like for now I'll revert to using the deprecated GnuTLS functions. But that's for sure not a solution for eternity... Does anone have a clue what this could be? Do I overlook some autotools automagic? Thanks, Rainer _______________________________________________ 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.
