[
https://issues.apache.org/jira/browse/HBASE-10714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13926099#comment-13926099
]
stack commented on HBASE-10714:
-------------------------------
Good on you [~jxiang]
Could this code be put into a method, hidden? There is some duplication of
code from replaceWriter:
{code}
+ SafePointZigZagLatch zigzagLatch = (ringBufferEventHandler == null)
+ ? null: ringBufferEventHandler.zigzagLatch;
+ if (zigzagLatch != null && !zigzagLatch.isCocked()) {
+ // Someone is replacing the writer now. To avoid racing with
+ // writer#close, let's wait till the writer is closed.
+ zigzagLatch.safePointReleasedLatch.await();
+ }
{code}
Also, since no locking going on, we could come out of the above and the sync
could still fail because its being swapped under us -- but that should be ok..
right. We'll catch the IOE and try a new log roll?
Thanks Jimmy for fixing holes in this stuff.
> SyncFuture hangs when sequence is 0
> -----------------------------------
>
> Key: HBASE-10714
> URL: https://issues.apache.org/jira/browse/HBASE-10714
> Project: HBase
> Issue Type: Bug
> Components: wal
> Reporter: Jimmy Xiang
> Assignee: Jimmy Xiang
> Fix For: 0.99.0
>
> Attachments: hbase-10714.patch
>
>
> In SyncFuture, NOT_DONE = 0. The initial value of the ringBuffer is -1. So
> ringBuffer.next() gives 0 for the first call. If we create a SyncFuture with
> sequence = 0, even when we set it's done (ie. doneSequence = 0), it is still
> not done since doneSequence == NOT_DONE == 0. Can we set NOT_DONE to -1, and
> the initial doneSequence to -2?
--
This message was sent by Atlassian JIRA
(v6.2#6252)
