Now I also see some pre-releases pass w/o failures. Thanks Tully! Let me ask one follow-up question regarding what Jack said; On Fri, Sep 27, 2013 at 10:04 AM, Jack O'Quin wrote: > I like to run the "devel" pre-release on the source repo before using bloom-release. Is there a ways to run the pre-release test on ROS' jenkins against source repo (I know local pre-release is possible as in the wiki#1)? #1 http://wiki.ros.org/bloom/Tutorials/PrereleaseTest#Locally Thanks, Isaac On Sat, Sep 28, 2013 at 6:03 AM, Tully Foote wrote: > > > > On Fri, Sep 27, 2013 at 12:56 PM, Tully Foote > wrote: >> >> With the change in servers recently I've just identified an issue with >> different memory usages between servers that is causing complete failures. > > > We updated the scripts and the prerelease jobs no longer run out of memory. > I've tested a few of the recent failures now pass: > http://jenkins.ros.org/job/prerelease_groovy_erratic_robot_precise_amd64/ > > Tully > > >> >> >> >> >> On Fri, Sep 27, 2013 at 10:04 AM, Jack O'Quin >> wrote: >>> >>> >>> On Fri, Sep 27, 2013 at 11:56 AM, Isaac Isao Saito <130s@lateeye.net> >>> wrote: >>>> >>>> Happy Friday ROS-Release, >>>> >>>> I'm afraid I've not yet gotten the idea of ROS pre-release. One thing >>>> is, what does the failure of the majority of previous prerelease tests >>>> mean? >>>> http://jenkins.ros.org/view/Prerelease/ >>>> Do maintainers fix on their upstrem repositories after getting >>>> pre-release error, and release w/o pre-release? >> >> >> Unfortunately at the moment we don't have completely clean prereleases so >> most things will cause failures. Part of that is we don't have a clear line >> for how to scope prereleases so they build everything downstream. The >> important thing to do is look through the errors and confirm that your >> release does not create new ones. >> >>> >>> >>> Not sure what others do. I like to run the "devel" pre-release on the >>> source repo before using bloom-release. I usually skip the "latest" option, >>> which uses the release repo, instead. >>> >>> I would be interested to know what different tests those options actually >>> perform. >> >> >> They do the same thing, except that they have a different source of the >> code. >> >> Tully >> >>> >>> -- >>> joq >>> >>> _______________________________________________ >>> ros-release mailing list >>> ros-release@lists.ros.org >>> http://lists.ros.org/mailman/listinfo/ros-release >>> >> >