hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ted Yu (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-17131) Avoid livelock caused by HRegion#processRowsWithLocks
Date Mon, 21 Nov 2016 02:06:58 GMT

    [ https://issues.apache.org/jira/browse/HBASE-17131?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15682241#comment-15682241
] 

Ted Yu commented on HBASE-17131:
--------------------------------

Reproduced the following failure with branch-1 patch:
{code}
testClientScannerIsResetWhenScanThrowsIOException(org.apache.hadoop.hbase.client.TestFromClientSide)
 Time elapsed: 1.667 sec  <<< FAILURE!
java.lang.AssertionError: expected:<17576> but was:<0>
	at org.apache.hadoop.hbase.client.TestFromClientSide.testClientScannerIsResetWhenScanThrowsIOException(TestFromClientSide.java:702)
{code}
Please investigate.

> Avoid livelock caused by HRegion#processRowsWithLocks
> -----------------------------------------------------
>
>                 Key: HBASE-17131
>                 URL: https://issues.apache.org/jira/browse/HBASE-17131
>             Project: HBase
>          Issue Type: Bug
>    Affects Versions: 2.0.0, 1.4.0, 1.3.1
>            Reporter: ChiaPing Tsai
>            Assignee: ChiaPing Tsai
>             Fix For: 2.0.0, 1.4.0, 1.3.1
>
>         Attachments: HBASE-17131.branch-1.v0.patch, HBASE-17131.v0.patch
>
>
> {code:title=HRegion.java|borderStyle=solid}
> try {
>       // STEP 2. Acquire the row lock(s)
>       acquiredRowLocks = new ArrayList<RowLock>(rowsToLock.size());
>       for (byte[] row : rowsToLock) {
>         // Attempt to lock all involved rows, throw if any lock times out
>         // use a writer lock for mixed reads and writes
>         acquiredRowLocks.add(getRowLockInternal(row, false));
>       }
>       // STEP 3. Region lock
>       lock(this.updatesLock.readLock(), acquiredRowLocks.size() == 0 ? 1 : acquiredRowLocks.size());
>       locked = true;
>       boolean success = false;
>       long now = EnvironmentEdgeManager.currentTime();
>       try {
> {code}
> We should lock all involved rows in the second try-finally. Otherwise, we won’t release
the previous locks if any subsequent lock times out.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message