hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andrew Purtell <apurt...@apache.org>
Subject sorry I won't be able to attend the meetup
Date Fri, 07 Aug 2009 16:44:55 GMT
Sorry I won't be able to attend the meetup. I've tried to make my
contributions to the discussion via email. Let me just take this
opportunity to finish that up:

First, I'll take care of the items on the Stargate roadmap for 0.21

Of the items on the roadmap list up at the wiki, my view is the
following are of primary importance:

- Insure that proper append() support in HDFS actually closes the
  WAL last block write hole.

- HBase-FSCK (HBASE-7) -- Suggest making this a blocker for 0.21. 

- Replication (HBASE-1295).

- Host HTD and HCD in ZK with copies in META only for cold boot
  scenarios (HBASE-1730) -- HBASE-1295 may require this depending on
  chosen implementation strategy.

- Distributed log splitting -- Writes fail with REE waiting for
  regions from failed server to be reassigned... while the master is
  busy crunching the logs, maybe > 10 minutes on a loaded cluster.

- Intra-row scanning (HBASE-1537) -- Without this, schemas with 
  millions of columns are not practical even if there is good store
  level support for it.

Thanks for your consideration,

   - Andy

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message