hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Xiao Chen (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-12116) BlockReportTestBase#blockReport_08 and #blockReport_08 intermittently fail
Date Sat, 15 Jul 2017 00:09:00 GMT

    [ https://issues.apache.org/jira/browse/HDFS-12116?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16088315#comment-16088315
] 

Xiao Chen commented on HDFS-12116:
----------------------------------

Thanks a lot for the review Eddy!

Updated patch 3 to address both comments.
bq. do you need to schedule block report?
Not for this test... This is because in {{BPServiceActor#offerService}}, {{isBlockReportDue}}
is checked iff {{isHeartbeatDue}} is true. But since the new method is updating block report
interval and named so, your proposal sounds to be the right thing to do in case it's used
by future tests.

> BlockReportTestBase#blockReport_08 and #blockReport_08 intermittently fail
> --------------------------------------------------------------------------
>
>                 Key: HDFS-12116
>                 URL: https://issues.apache.org/jira/browse/HDFS-12116
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: test
>    Affects Versions: 0.22.0
>            Reporter: Xiao Chen
>            Assignee: Xiao Chen
>         Attachments: HDFS-12116.01.patch, HDFS-12116.02.patch, TEST-org.apache.hadoop.hdfs.server.datanode.TestNNHandlesBlockReportPerStorage.xml
>
>
> This seems to be long-standing, but the failure rate (~10%) is slightly higher in dist-test
run in using cdh.
> In both _08 and _09 tests:
> # an attempt is made to make a replica in {{TEMPORARY}}
>  state, by {{waitForTempReplica}}.
> # Once that's returned, the test goes on to verify block reports shows correct pending
replication blocks.
> But there's a race condition. If the replica is replicated between steps #1 and #2, {{getPendingReplicationBlocks}}
could return 0 or 1, depending on how many replicas are replicated, hence failing the test.
> Failures are seen on both {{TestNNHandlesBlockReportPerStorage}} and {{TestNNHandlesCombinedBlockReport}}



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org


Mime
View raw message