[ 
https://issues.apache.org/jira/browse/HBASE-9655?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13777736#comment-13777736
 ] 

Nick Dimiduk commented on HBASE-9655:
-------------------------------------

Aborting on all DoNotRetryIOEs is too generic I think. For instance, my last 
run aborted do to an UnknownScannerException. For an application rolling over 
cluster woes, we'd expect the scanner to not be retried but a new scanner to be 
acquired, making this a non-terminal exception for the sake of this test.

Of list of Exceptions derived from NoNotRetryIOE, I think only the following 
should be treated as fatal for the purposes of this test: 
AccessDeniedException, CoprocessorException, FatalConnectionException, 
InvalidFamilyOperationException, NamespaceExistException, 
NamespaceNotFoundException, NoSuchColumnFamilyException, TableExistsException, 
TableNotFoundException.

That leaves the following exceptions which will be retried by the test harness 
up to the specified number of times: HBaseSnapshotException, LeaseException, 
NotAllMetaRegionsOnlineException, ScannerTimeoutException, 
TableNotDisabledException, TableNotEnabledException, UnknownScannerException.

What say you?
                
> IntegrationTestMTTR can loop forever on improperly configured clusters
> ----------------------------------------------------------------------
>
>                 Key: HBASE-9655
>                 URL: https://issues.apache.org/jira/browse/HBASE-9655
>             Project: HBase
>          Issue Type: Test
>          Components: test
>    Affects Versions: 0.95.2
>            Reporter: Nick Dimiduk
>            Assignee: Nick Dimiduk
>         Attachments: HBASE-9655.00.patch
>
>
> IntegrationTestMTTR has a retry loop that can run infinitely. For instance, 
> running the test on a secure cluster as a user who lacks permissions to 
> perform table actions can cause the this scenario. Add another loop counter 
> and bail when a TimingCalable instance throws too many unexpected Exceptions.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to