Ted Yu created HBASE-17841:
------------------------------
Summary: ServerCrashProcedure is not triggered when meta server
with unflushed edits is aborted
Key: HBASE-17841
URL: https://issues.apache.org/jira/browse/HBASE-17841
Project: HBase
Issue Type: Bug
Affects Versions: 2.0
Reporter: Ted Yu
When writing unit test for HBASE-17287, I noticed that the wait for master to
come down after hdfs enters safe mode times out (where meta server still has
unflushed edits).
The same test in branch-1 passes fine.
Looking at
org.apache.hadoop.hbase.master.procedure.TestSafemodeBringsDownMaster-output.txt
, I don't see occurrence of ServerCrashProcedure.
While in branch-1, there is something similar to the following:
{code}
at org.apache.hadoop.hdfs.DFSClient.rename(DFSClient.java:1661)
at
org.apache.hadoop.hdfs.DistributedFileSystem.rename(DistributedFileSystem.java:525)
at
org.apache.hadoop.hbase.master.MasterFileSystem.getLogDirs(MasterFileSystem.java:364)
at
org.apache.hadoop.hbase.master.MasterFileSystem.splitLog(MasterFileSystem.java:429)
at
org.apache.hadoop.hbase.master.MasterFileSystem.splitMetaLog(MasterFileSystem.java:343)
at
org.apache.hadoop.hbase.master.MasterFileSystem.splitMetaLog(MasterFileSystem.java:334)
at
org.apache.hadoop.hbase.master.procedure.ServerCrashProcedure.processMeta(ServerCrashProcedure.java:351)
at
org.apache.hadoop.hbase.master.procedure.ServerCrashProcedure.executeFromState(ServerCrashProcedure.java:239)
at
org.apache.hadoop.hbase.master.procedure.ServerCrashProcedure.executeFromState(ServerCrashProcedure.java:73)
at
org.apache.hadoop.hbase.procedure2.StateMachineProcedure.execute(StateMachineProcedure.java:139)
at org.apache.hadoop.hbase.procedure2.Procedure.doExecute(Procedure.java:506)
at
org.apache.hadoop.hbase.procedure2.ProcedureExecutor.execProcedure(ProcedureExecutor.java:1152)
{code}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
