[
https://issues.apache.org/jira/browse/HBASE-10765?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13937365#comment-13937365
]
Ted Yu commented on HBASE-10765:
--------------------------------
>From TestHBaseFsck#deleteTable():
{code}
void deleteTable(TableName tablename) throws IOException {
HBaseAdmin admin = new HBaseAdmin(conf);
admin.getConnection().clearRegionCache();
{code}
The HBaseAdmin instance retrieves currently active connection for the mini
cluster.
The cache of connection used by HBaseAdmin is cleared.
For testNoHdfsTable, deleteTable() is not called - deleteTableDir() is called
instead.
So the step of clearing connection cache was skipped.
> TestHBaseFsck#testNoHdfsTable doesn't clear region cache
> --------------------------------------------------------
>
> Key: HBASE-10765
> URL: https://issues.apache.org/jira/browse/HBASE-10765
> Project: HBase
> Issue Type: Test
> Reporter: Ted Yu
> Assignee: Ted Yu
> Attachments: 10765-v1.txt
>
>
> TestHBaseFsck#testNoHdfsTable doesn't call deleteTable(), which clears region
> cache for the connection, at the end of the test due to the way the test is
> constructed.
> However, region cache for the underlying connection should be cleared.
> I found this issue when investigating test failure such as the following:
> https://builds.apache.org/job/HBase-TRUNK/5018/testReport/org.apache.hadoop.hbase.util/TestHBaseFsck/testHbckThreadpooling/
--
This message was sent by Atlassian JIRA
(v6.2#6252)
