[
https://issues.apache.org/jira/browse/HBASE-12422?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-12422:
--------------------------
Attachment: 12422v3.txt
Try this patch. Should fix the TestSecureLoadIncrementalHFilesSplitRecovery.
More work to do on the TestHBaseFsck failure (should have fixes one test
failure).
Here is fix for the TestSecureLoadIncremental...
45 - Table table = new HTable(conn.getConfiguration(),
getTableName());
46 - secureClient = new SecureBulkLoadClient(table);
47 - success = secureClient.bulkLoadHFiles(famPaths,
fsDelegationToken.getUserToken(),
48 - bulkToken, getLocation().getRegionInfo().getStartKey());
49 + try (Table table = conn.getTable(getTableName())) {
50 + secureClient = new SecureBulkLoadClient(table);
51 + success = secureClient.bulkLoadHFiles(famPaths,
fsDelegationToken.getUserToken(),
52 + bulkToken, getLocation().getRegionInfo().getStartKey());
53 + }
Use existing Connection rather than create new one w/ 'wrong' zk config -- so
fail to connect to zk.
Patch is big because some of the tests have been refactored to use new idiom
just so we do clean up of resources, action that was not there previous.
Also includes addition of timeout on tests so they fail rather than go zombie.
Lets see how this does one hadoopqa.
> Use ConnectionFactory in HTable constructors
> --------------------------------------------
>
> Key: HBASE-12422
> URL: https://issues.apache.org/jira/browse/HBASE-12422
> Project: HBase
> Issue Type: Bug
> Components: Client
> Reporter: Solomon Duskis
> Assignee: stack
> Priority: Minor
> Fix For: 1.0.0, 2.0.0
>
> Attachments: 12422v2.txt, 12422v3.txt, HBASE-12422.patch,
> HBASE-12422.patch, HBASE-12422.patch, HBASE-12422.patch, HBASE-12422.patch
>
>
> Use ConnectionFactory in HTable instead of ConnectionManager.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
