About the issue you were raising, it's a bit different than what you described.
If you try disabling a table that still has split parents in .META., those parents will be put in the regionsInTransitions list and marked as closing... but they won't be closed since they are offline already. It's when you re-enable that table that it may mess things up a bit, nothing a restart can't fix (I didn't dig that far in my testing). J-D On Wed, May 5, 2010 at 1:10 PM, Andrew Purtell <apurt...@apache.org> wrote: > I missed that the vote closed early. Disregard the -1 as it came too late. > > - Andy > >> From: Andrew Purtell <apurt...@apache.org> >> Subject: Re: [VOTE] Shall we release hbase-0.20.4 Release Candidate 6 as >> hbase-0.20.4? >> To: hbase-dev@hadoop.apache.org >> Date: Wednesday, May 5, 2010, 12:55 PM >> Do I understand correctly that >> HBASE-2515 means we will lose regions (stuck in transition) >> if a table is disabled at an inopportune time after a split? >> If so, -1, let's pull HBASE-2515 into 0.20.4. This may have >> been there for a while but now it's a known problem. It's >> still necessary to disable tables to make schema changes. >> >> - Andy >> >> > From: Stack <st...@duboce.net> >> > Subject: [VOTE] Shall we release hbase-0.20.4 Release >> Candidate 6 as hbase-0.20.4? >> > To: "HBase Dev List" <hbase-dev@hadoop.apache.org> >> > Date: Tuesday, May 4, 2010, 4:52 PM >> > I've posted an hbase 0.20.4 release >> > candidate 6 here: >> > >> > http://people.apache.org/~stack/hbase-0.20.4-candidate-6/ >> > >> > Should we release this candidate as hbase 0.20.4? >> >> >> >> >> >> > > > > >
