Hi guys, Sorry to flood -devel with many Patchy failures, but we're still at an early stage of experimenting concurrent runs of Patchy-staging, ones on-demand made on fast machines, and ones made on Grenouille that take hours but (hopefully) runs every 8 hours [1] without human intervention. Previous failures have come a lot from my experiments on Patchy code, and a little from bugs in existing code, I'm commenting a bit on this failure below.
[1] I decreased frequency a bit because a build takes between three and four hours, and I'm setting up a build of translation branch (without any merging operation) to be run every 8 hours too, interleaved with Patchy-staging. Le dimanche 05 août 2012 à 21:02 +0000, [email protected] a écrit : > 08:55:13 (UTC) Begin LilyPond compile, previous commit at > ac0e95c8112250bcaddf0bce57ba2980767a2f43 > 08:55:18 Merged staging, now at: c28baa15608b96e15ae3971ac2a70cc65282c110 > 08:55:23 Success: ./autogen.sh --noconfigure > 08:56:29 Success: ../configure --disable-optimising > 08:58:45 Success: nice make clean > 09:41:03 Success: nice make -j2 CPU_COUNT=2 > ANTI_ALIAS_FACTOR=1 > 11:42:29 Success: nice make test -j2 CPU_COUNT=2 > ANTI_ALIAS_FACTOR=1 > 21:02:39 Success: nice make doc -j2 CPU_COUNT=2 > ANTI_ALIAS_FACTOR=1 Wow, I can't figure out why it took so long to build this time, whereas it went back to normal this morning (3h12min). It doesn't seem that there have been other processes with high resource consumption at the same time, and the hard disk is sane. > 21:02:41 *** FAILED STEP *** > merge from staging > Command '['git', 'push', 'origin', 'test-master-lock:master']' returned > non-zero exit status 1 > error: src refspec test-master-lock does not match any. Mmh, I thought I fixed locking functionality in lilypond-extra:24e9bc986062605022eba75ff168310de2d614c, but I didn't; a Patchy should not remove test-master-lock branch if it exists already and another Patchy is running at the same time; I'll try to fix this with writing PID to $LILYPOND_BUILD_DIR/patchy.pid, so that Patchy should remove test-master-lock in case some previous Patchy run died unexpectedly. > error: failed to push some refs to '/var/src/lilypond-git' This is normal and wpuld have happened even without the deletion of test-master-lock by the Patchy run at 16:55 UTC: Phil's Patchy picked up a revision in staging newer than when my Patchy started, and his run completed a lot earlier; in this case I think a message should say "Info: origin/staging has been fastforwarded to a newer revision than last fetch to the repository of this computer." and not CC lilypond-devel. Best, John _______________________________________________ lilypond-devel mailing list [email protected] https://lists.gnu.org/mailman/listinfo/lilypond-devel
