> On 31 Jan 2019, at 14:12, Marco Varlese <mvarl...@suse.de> wrote: > > Damjan, > > On 1/31/19 12:41 PM, Damjan Marion wrote: >> >> >>> On 31 Jan 2019, at 08:57, Marco Varlese <mvarl...@suse.de >>> <mailto:mvarl...@suse.de>> wrote: >>> >>> Damjan, >>> >>> On 1/30/19 3:29 PM, Damjan Marion via Lists.Fd.Io wrote: >>>> >>>> Folks, >>>> >>>> Anybody can help with understanding why opensuse jobs are failing with: >>>> >>>> *13:54:57* >>>> /w/workspace/vpp-verify-master-osleap15/build-root/rpmbuild/BUILD/vpp-19.04/src/plugins/nsim/nsim.c:620:1: >>>> fatal error: error writing to /tmp/ccJaw6Cw.s: Cannot allocate memory >>>> *13:54:57* }; >>>> *13:54:57* ^ >>>> *13:54:57* compilation terminated. >>>> >>>> >>>> *13:54:57* gcc-7: internal compiler error: Killed (program cc1) >>>> *13:54:57* Please submit a full bug report, >>>> *13:54:57* with preprocessed source if appropriate. >>>> *13:54:57* See <http://bugs.opensuse.org/> for instructions. >>> This to me looks like an infrastructure issue. >>> If it was a problem with the distro / setup it would have happened all >>> the time whilst things were pretty stable over the past months... >> >> Well, that line clearly asks for filling bug with opensuse.org >> <http://opensuse.org>... > Sure... that's pretty standard I'd say. > > Aaaaanywaaaayyyy... I did ask around to some compiler guys and that > looks like an out-of-memory issue (e.g. no more memory left on the node > where the compilation was running)... > >> >>> >>>> >>>> >>>> >>>> Also, how we can avoid in future that people need to deal with this kind >>>> of issues? >>> I am not sure how to answer this question. >>> >>>> >>>> Personally I don't see lot of value in compiling VPP agains every single >>>> distro on the planet in each verify job, >>>> but i'm fine with that as long as things are stable and not PITA for >>>> people trying to get patch verified.... >>> Are you suggesting to remove openSUSE builds from Jenkins for VPP? >> >> That is not what I said. I just don't see lot of value of running each >> verify job against 3 distros, >> I believe one is enough, and I'm not suggesting which one is that. >> >>> Beside, I see verify jobs are passing now. Has anything been done in the >>> meantime or all started to work auto-magically? >> >> No, centos job was failing after that. but then everything passed. >> it is lottery, whole infra is fragile and more jobs we run makes it >> harder to have winning lottery ticket. > I think it's rather a matter of sizing the running jobs queue to the > actual capacity we have in the back-end. I remember something similar > happened during summer / end-of summer time where jobs were randomly > failing due to infra issues. > Perhaps, another approach (instead of reducing the number of distros) > could be to have only couple of patches building at a time. I would > imagine that could be driven by Jenkins jobs/queues/etc. > In that way, we could try addressing the capacity issue without reducing > distros support.
Fine with me if we have volunteers to drive that. At the moment I see many people expecting that their jobs are part of infra, but I don't see any initiative to make infra less fragile (with Paul who offered to look at this 2 days ago being an exception). -- Damjan
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#12095): https://lists.fd.io/g/vpp-dev/message/12095 Mute This Topic: https://lists.fd.io/mt/29594604/21656 Group Owner: vpp-dev+ow...@lists.fd.io Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
