[
https://issues.apache.org/jira/browse/KYLIN-3953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16861723#comment-16861723
]
XiaoXiang Yu commented on KYLIN-3953:
-------------------------------------
It looks like that it is caused by zookeeper lock not released as expected. You
may try to restart Job Server or use ZK CLI to delete that lock path.
> cube run many building instances，but dict file locked，failed to acquire lock
> ----------------------------------------------------------------------------
>
> Key: KYLIN-3953
> URL: https://issues.apache.org/jira/browse/KYLIN-3953
> Project: Kylin
> Issue Type: Bug
> Components: Storage - HBase
> Affects Versions: v2.6.1
> Environment: cdh57-hadoop, kylin 2.6.1
> Reporter: leh
> Priority: Major
> Labels: build
> Original Estimate: 96h
> Remaining Estimate: 96h
>
> when i build the cube history data for many days,and run many instances in
> monitor, but hang out the step" #4 Step Name: Build Dimension Dictionary"
> kylin.log as follow:
> 2019-04-12 11:20:28,519 DEBUG [Scheduler 635400189 Job
> b66ccafa-a2ce-621d-b76d-bac2b16bcca3-240] util.ZookeeperDistributedLock:147 :
> [email protected] trying to lock
> /kylin/kylin_metadata/dict/DEFAULT.APP_STRATEGY_ADS_VIEW_BASE_INFO_DI_DURATION_CUID/lock
> 2019-04-12 11:20:28,575 DEBUG [Scheduler 635400189 Job
> b66ccafa-a2ce-621d-b76d-bac2b16bcca3-240] util.ZookeeperDistributedLock:152 :
> [email protected] see
> /kylin/kylin_metadata/dict/DEFAULT.APP_STRATEGY_ADS_VIEW_BASE_INFO_DI_DURATION_CUID/lock
> is already locked
> 2019-04-12 11:20:28,609 DEBUG [Scheduler 635400189 Job
> b66ccafa-a2ce-621d-b76d-bac2b16bcca3-240] util.ZookeeperDistributedLock:162 :
> [email protected] failed to acquire lock at
> /kylin/kylin_metadata/dict/DEFAULT.APP_STRATEGY_ADS_VIEW_BASE_INFO_DI_DURATION_CUID/lock,
> which is held by [email protected]
> -------
> kylin.properties:
> ## Max count of concurrent jobs running
> kylin.job.max-concurrent-jobs=50
> kylin.cube.max-building-segments=62
> -------
> and the same cube ,can I build many instance parallismly? and how to avoid
> this question?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
