[
https://issues.apache.org/jira/browse/HBASE-12384?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14190378#comment-14190378
]
Andrew Purtell edited comment on HBASE-12384 at 10/30/14 5:03 PM:
------------------------------------------------------------------
This test makes some assumptions about compaction policy and in one case relies
on winning a race between flush and subsequent count of store files and a
compaction. Elsewhere for checking compaction progress the test uses an API
available for that purpose. We are lacking an API for confirming flush request
completion. Rather than count files, sleep for a short time. We lose the
confirmation of flush completion but the test won't hang now and if in the
future we change the default compaction policy.
was (Author: apurtell):
This patch makes some assumptions about compaction policy and in one case
relies on winning a race between flush and subsequent count of store files and
a compaction. Elsewhere for checking compaction progress the test uses an API
available for that purpose. We are lacking an API for confirming flush request
completion. Rather than count files, sleep for a short time. We lose the
confirmation of flush completion but the test won't hang now and if in the
future we change the default compaction policy.
> TestTags can hang on fast test hosts
> ------------------------------------
>
> Key: HBASE-12384
> URL: https://issues.apache.org/jira/browse/HBASE-12384
> Project: HBase
> Issue Type: Bug
> Reporter: Andrew Purtell
> Assignee: Andrew Purtell
> Fix For: 2.0.0, 0.98.8, 0.99.2
>
> Attachments: HBASE-12384-0.98.patch, HBASE-12384-master.patch
>
>
> Waiting indefinitely expecting flushed files to reach a certain count after
> triggering a flush but compaction has happened between the flush and check
> for number of store files.
> {code}
> admin.flush(tableName);
> regions = TEST_UTIL.getHBaseCluster().getRegions(tableName);
> for (HRegion region : regions) {
> Store store = region.getStore(fam);
> - Flush and compaction has happened before here --->
> while (!(store.getStorefilesCount() > 2)) {
> - Hung forever in here --->
> Thread.sleep(10);
> }
> }
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
