Hi, On Tue, Apr 19, 2022 at 4:39 PM 胡玮文 <sehuww@xxxxxxxxxxxxxxxx> wrote: > > Hi Yuri, > > > Can we increase the priority of https://github.com/ceph/ceph/pull/44335 ? This issue prevents us (and many other users I believe) from upgrading to any newer version (MDS would crash). We had a painful journey when upgrading from 16.2.6 to .7, and we will experience this again when we decide to upgrade from current version. We hope we don't need to experence this the third time. We were running into a failed test with the PR which blocked merging it. We would need someone to look into the failed test so that the PR can be picked up again for validation. That said, we will surely prioritize it but I think it is unlikely to make it to the release. > > > I think maybe it worth a warning in the release note. The MDS may crash when reducing max_mds, and cephadm will reduce max_mds automatically when upgrading. Fair enough. You do mention (in the PR) that it is really easy to reproduce. I wonder why we haven't seen this in our testing. We do have (and run) MDS thrasher tests (with workloads) which would closely mimic your preproducer. > > > Thanks, > > Weiwen Hu > > > We merged all PRs known to be included into the next pacific and are ready to merge the repo later today and start QE validation. > > Thx > YuriW > > On Thu, Mar 31, 2022 at 3:51 AM Ilya Dryomov <idryomov@xxxxxxxxx> wrote: >> >> On Thu, Mar 31, 2022 at 12:15 AM Yuri Weinstein <yweinste@xxxxxxxxxx> wrote: >> > >> > We tested and merged many PRs in the last week and again want to start testing the next pacific ASAP. >> > >> > I see no new issues in the queue (is:open is:pr milestone:pacific label:needs-qa) >> > >> > Please let me know what is still planned, otherwise, the repo will be locked and we will start QE validation. >> >> For RBD: >> >> - https://github.com/ceph/ceph/pull/45638 >> labeled with wip-yuri2-testing, should be OK to merge once that >> branch goes through >> >> - https://github.com/ceph/ceph/pull/45184 >> already tested but can't be merged yet due to a unit test issue, >> Chris is looking into it >> >> Thanks, >> >> Ilya >> > _______________________________________________ > Dev mailing list -- dev@xxxxxxx > To unsubscribe send an email to dev-leave@xxxxxxx -- Cheers, Venky _______________________________________________ Dev mailing list -- dev@xxxxxxx To unsubscribe send an email to dev-leave@xxxxxxx
