On Mon, Dec 5, 2016 at 6:53 AM, Abhishek L <abhishek@xxxxxxxx> wrote: > > Abhishek L writes: > >> Sage Weil writes: >> >>> On Wed, 30 Nov 2016, Abhishek L wrote: >>>> Abhishek L writes: >>>> >>>> > Sage Weil writes: >>>> > >>>> >> On Wed, 30 Nov 2016, Abhishek L wrote: >>>> >>> >>>> >>> Abhishek L writes: >>>> >>> >>>> >>> > Sage Weil writes: >>>> >>> > >>>> >>> >> On Wed, 23 Nov 2016, Abhishek L wrote: >>>> >>> >>> >>>> >>> >>> Hi Sage, Greg, >>>> >>> >>> >>>> >>> >>> >>>> >>> >>> Yuri Weinstein writes: >>>> >>> >>> >>>> >>> >>> > See updated status - http://tracker.ceph.com/issues/17487#note-32 >>>> >>> >>> > >>>> >>> >>> > Outstanding issues: >>>> >>> >>> > >>>> >>> >>> > knfs - http://tracker.ceph.com/issues/16397 (same as in v10.2.3, Greg >>>> >>> >>> > pls review/approve, assumed Approved ?) >>>> >>> >>> > >>>> >>> >>> > upgrade/hammer-x (jewel) - http://tracker.ceph.com/issues/17847 ((Sage >>>> >>> >>> > pls review/approve, seems persistent, but maybe not a showstopper) >>>> >>> >>> > >>>> >>> >>> > upgrade/infernalis-x (jewel) - deprecated (Nathan is still working >>>> >>> >>> > to make it pass, see issues in the tacker summary above) >>>> >>> >>> > >>>> >>> >>> > Sage, jewel 10.2.4 can be released as soon as you agree with the >>>> >>> >>> > findings/summary. >>>> >>> >>> >>>> >>> >>> Do you think we're ready to release 10.2.4 yet? >>>> >>> >> >>>> >>> >> I'm reproducing http://tracker.ceph.com/issues/17847 with logs to make >>>> >>> >> sure this isn't a regression. >>>> >>> >> >>>> >>> >> We can ignore the infernalis runs. >>>> >>> >> >>>> >>> >> I think we can ignore the knfs selinux issue too.. Greg, can you confirm? >>>> >>> >> >>>> >>> > Added the prs 12001 & 12167 on top of the jewel branch and scheduled rados runs >>>> >>> > at >>>> >>> > http://pulpito.ceph.com/abhi-2016-11-29_10:22:25-rados-wip-jewel-10-2-4-distro-basic-smithi/ & >>>> >> >>>> >> https://github.com/ceph/ceph-qa-suite/pull/1292 >>>> >> >>>> >> avoids xenial for the rados upgrade tests (in jewel branch). >>>> >>>> Ah alright, maybe this was the cause after all, let's get this in (or >>>> push a branch to ceph-qa-suite so that I can schedule against that) >>> >>> I added a patch for the client-upgrade tests to the same branch. You >>> should be able to just schedule with --suite-branch jewel-avoid-xenial >>> instead of --suite-branch jewel (for both rados and >>> upgrade/client-upgrade). >>> >>> sage >>> >>> >>>> >> >>>> >> The upgrade tests already explicitly call out trusty, so they should be >>>> >> fine. >>>> >> >>>> >>> > updated the tracker at http://tracker.ceph.com/issues/17851#note-17, >>>> >>> > with details of the test. I'll update the progress once the suite goes >>>> >>> > through >>>> >>> >>>> >>> Had around 11 tests fail from 296 scheduled, there were a couple of >>>> >>> valgrind issues on ceph-mon (which were seen at earlier runs on jewel as >>>> >>> well) and an s3test failure, rest of the issues were looking related to >>>> >>> infrastructure as they were failing to get specific version numbers from >>>> >>> gitbuilders. >>>> >>> >>>> >>> Reported this issue as: >>>> >>> >>>> >>> http://tracker.ceph.com/issues/18089 >>>> >>> >>>> >>> subsequent re runs are still failing with similar errors. The details >>>> >>> are updated at >>>> >>> >>>> >>> http://tracker.ceph.com/issues/17487#note-37 >>>> >>> >>>> >>> The upgrade suite has also failed with similar errors. >>>> >> >>>> >> Which one? >>>> > The upgrade/client-upgrade suite (I hope this is the right upgrade >>>> > suite), the errors are the same failed to fetch package version errors >>>> > seen for the rados suite, so not actual test run errors yet. >>>> > >>>> > http://pulpito.ceph.com/abhi-2016-11-30_09:58:50-upgrade:client-upgrade-wip-jewel-10-2-4-distro-basic-smithi/ >>>> > is the run >>>> > >> Update on the current status, >> For prs 12001 & 12167: (sage/Sam) >> >> The upgrade passed with the exception of infernalis/jewel which failed >> with RBD (test_librbd_api.sh) (jdillaman please approve) >> >> http://pulpito.ceph.com/abhi-2016-12-02_09:45:37-upgrade:client-upgrade-wip-jewel-10-2-4-distro-basic-smithi/ >> >> the rados run (with the exception of the valgrind issue) are still >> waiting for completion for 4 of the 296 jobs, this one suite always >> seems to die when scheduled, this is the current url of the rados jobs >> (still waiting) >> http://pulpito.ceph.com/abhi-2016-12-02_10:06:39-rados-wip-jewel-10-2-4-rc-distro-basic-smithi/ > > Loic was able to spot the error in this run (which was a regression > introduced by f95ed3e, this was later fixed in > https://github.com/ceph/ceph-qa-suite/pull/1297, so running this suite > on the latest branch came out green, > > http://pulpito.front.sepia.ceph.com/abhi-2016-12-05_10:25:26-rados-wip-jewel-10-2-4-rc-distro-basic-smithi/ > > Which means 12167 is also good to go. >> >> >> for pr 12207: (josh) >> the rgw suite also saw the same valgrind issue >> (http://pulpito.ceph.com/abhi-2016-12-01_22:19:07-rgw-wip-jewel-10-2-4-rc-distro-basic-smithi/) >> >> https://github.com/ceph/ceph/pull/12267, (mon cleanup) was merged on Sam's approval. >> >> prs 11884 & 12067 (Sam, Alfredo): >> the ceph create-keys pr #11884 hasn't been merged yet into jewel, is >> there any suite that must be run to validate this. Same applies for >> https://github.com/ceph/ceph/pull/12067 (OSDMon: MOSDBoot ) > > 12067 was merged on Sam's approval as well. Which only leaves us with > 11884. Bummed that 11884 didn't make it. It fixes issues where it hangs for ever for any tool that tries to bootstrap and encounters problems during deployment. There isn't a specific suite that must be run to validate this. > > Best, > Abhishek Lekshmanan > SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg) -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
