bq. Apache projects are supposed to encourage collaboration I totally agree with you.
Cheers On Sat, Feb 10, 2018 at 10:32 AM, anil gupta <[email protected]> wrote: > Thanks Ted. Will try to do the clean-up. Unfortunately, we ran out of > support for this cluster since its nearing End-of-life. For our new > clusters we are in process of getting support. > > PS: IMO, I agree that i should use vendor forum/list for any vendor > specific stuff but i think its appropriate to use this mailing Apache HBase > questions/issues related to HBase. As per my understanding, Apache projects > are supposed to encourage collaboration rather building boundaries around > vendors.("collaboration and openness" is one of the reason i like Apache > Projects) > > On Sat, Feb 10, 2018 at 10:11 AM, Ted Yu <[email protected]> wrote: > > > You can cleanup oldwal directory beginning with oldest data. > > > > Please open support case with the vendor. > > > > On Sat, Feb 10, 2018 at 10:02 AM, anil gupta <[email protected]> > > wrote: > > > > > Hi Ted, > > > > > > We cleaned up all the snaphsots around Feb 7-8th. You were right that i > > > dont see the CorruptedSnapshotException since then. Nice observation! > > > So, i am again back to square one. Not really, sure why oldwals and > > > recovered.edits are not getting cleaned up. I have already removed all > > the > > > replication peer and deleted all the snapshots. > > > Is it ok if i just ahead and cleanup oldwal directory manually? Can i > > also > > > clean up recovered.edits? > > > > > > Thanks, > > > Anil > > > > > > On Sat, Feb 10, 2018 at 9:37 AM, Ted Yu <[email protected]> wrote: > > > > > > > Can you clarify whether /apps/hbase/data/.hbase-snapshot/.tmp/ > became > > > > empty > > > > after 2018-02-07 09:10:08 ? > > > > > > > > Do you see CorruptedSnapshotException for file outside of > > > > /apps/hbase/data/.hbase-snapshot/.tmp/ ? > > > > > > > > Cheers > > > > > > > > > > > > > > > > -- > > > Thanks & Regards, > > > Anil Gupta > > > > > > > > > -- > Thanks & Regards, > Anil Gupta >
