[GitHub] athanatos opened a new pull request #875: Issue 865: LongHierarchicalLedgerManager: refactor to fix races

[GitBox <gi...@apache.org>]

athanatos opened a new pull request #875: Issue 865: LongHierarchicalLedgerManager: refactor to fix races
URL: https://github.com/apache/bookkeeper/pull/875
 
 
   LongHierarchicalLedgerRangeIterator.initialize() could erroneously exit
   iteratorDone if the lexicographically first path in zk had fewer
   than 4 levels.  This can happen for a few reasons including a case where
   a client creating a ledger on that path crashed during the zk updates or
   the iterator.hasNext() call simply raced with an in-progress node
   creation or removal.  ScanAndCompareGarbageCollector is hasNext()
   returns false will delete all ledgers on the bookie, so this is a fairly
   serious bug.
   
   Ruling out other such bugs was fairly dificult with the structure of the
   code as written, so instead use a simpler recursive iterator design with
   simpler pre/post conditions.
   
   Also, surface KeeperException.NoNodeException from
   ZkUtils.getChildrenInSingleNode so that we can actually handle it in
   LHLM.
   
   Master Issue: 865
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services