You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Michael Stack (Jira)" <ji...@apache.org> on 2020/06/19 16:56:00 UTC
[jira] [Updated] (HBASE-24585) Failed start recovering crash in
standalone mode if procedure-based distributed WAL split &
hbase.wal.split.to.hfile=true
[ https://issues.apache.org/jira/browse/HBASE-24585?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Michael Stack updated HBASE-24585:
----------------------------------
Summary: Failed start recovering crash in standalone mode if procedure-based distributed WAL split & hbase.wal.split.to.hfile=true (was: If RSProcedureHandler throws exception, it aborts the hosting RS)
> Failed start recovering crash in standalone mode if procedure-based distributed WAL split & hbase.wal.split.to.hfile=true
> -------------------------------------------------------------------------------------------------------------------------
>
> Key: HBASE-24585
> URL: https://issues.apache.org/jira/browse/HBASE-24585
> Project: HBase
> Issue Type: Bug
> Reporter: Michael Stack
> Priority: Major
>
> HBASE-24574 proc v2 distributed log splitting is enabled. A remote split fails because it was interrupted. The InterruptedException became an IOE and then bubbled up and out of the RSPH below causing a RS abort.
> {code}
> 2020-06-17 21:20:37,472 ERROR [RS_LOG_REPLAY_OPS-regionserver/localhost:16020-0] handler.RSProcedureHandler: Error when call RSProcedureCallable:
> java.io.IOException: Failed WAL split, status=RESIGNED, wal=file:/Users/stack/checkouts/hbase.apache.git/tmp/hbase/WALs/localhost,16020,1592440848604-splitting/localhost%2C16020%2C1592440848604.meta.1592440852959.meta
> at org.apache.hadoop.hbase.regionserver.SplitWALCallable.splitWal(SplitWALCallable.java:106)
> at org.apache.hadoop.hbase.regionserver.SplitWALCallable.call(SplitWALCallable.java:86)
> at org.apache.hadoop.hbase.regionserver.SplitWALCallable.call(SplitWALCallable.java:49)
> at org.apache.hadoop.hbase.regionserver.handler.RSProcedureHandler.process(RSProcedureHandler.java:49)
> at org.apache.hadoop.hbase.executor.EventHandler.run(EventHandler.java:104)
> at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:748)
> {code}
> The remote-procedure framework needs to be more resilient? Log the exception unless an ERROR and keep going? Otherwise, makes features like procedurev2 distributed log splitting brittle.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)