Hi Nick, Apologies for the delayed reply. If you see the current patch we are just ensuring that there is no behaviour change by moving the archival to an async thread. So I don't expect any behaviour change there. The existing behaviour was to abort the server on an archival failure and now we are still doing it except the fact that if needed user can configure a retry there. By default there is no retry. Yes the test related failures and in turn the changes required in the postLogArchival has been taken care by Duo. If you still feel it is too invasive then I will refrain from committing the patch there. Thanks Nick.
Regards Ram On Wed, Oct 21, 2020 at 11:14 PM Nick Dimiduk <[email protected]> wrote: > Hi Ram, > > Per my response on jira, I think this looks too invasive for a patch > release, but I'm willing to be convinced otherwise. Looks like you and Duo > were able to sort out the associated issue in HBASE-25186. Do you have a > plan to test that extra WALs don't accumulate inadvertently? > > Thanks, > Nick > > On Wed, Oct 21, 2020 at 9:57 AM ramkrishna vasudevan < > [email protected]> wrote: > > > Hi Nick > > > > Can I port https://issues.apache.org/jira/browse/HBASE-25065 to > branch-2.2 > > - Note that it is an improvement? > > > > Regards > > Ram > > > > On Wed, Oct 21, 2020 at 10:05 PM Nick Dimiduk <[email protected]> > wrote: > > > > > Hello team, > > > > > > It's come around to that time again. We have about 30 improvements on > the > > > branch, which is my litmus test for another patch release. Please try > to > > > land any patches you have in flight, and plan for an RC on or around > > > Tuesday, 27 Oct. > > > > > > Thanks, > > > Nick > > > > > >
