jackrabbit-oak-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Julian Reschke (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (OAK-7709) Persistent cache files locked on restart of DocumentNodeStore
Date Wed, 29 Aug 2018 13:21:00 GMT

    [ https://issues.apache.org/jira/browse/OAK-7709?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16596317#comment-16596317
] 

Julian Reschke commented on OAK-7709:
-------------------------------------

AFAIU, the root cause here is DocumentNodeStore.dispose()'s attempt to write a tombstone revision,
which will fail with an exception upon lease failure, causing the remaining cleanup code to
be skipped.

I'll try to address this, but there may be more issues related to restart after lease failure
down the road.

> Persistent cache files locked on restart of DocumentNodeStore
> -------------------------------------------------------------
>
>                 Key: OAK-7709
>                 URL: https://issues.apache.org/jira/browse/OAK-7709
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: documentmk
>    Affects Versions: 1.8.2
>         Environment: * Linux
> * IBM JDK 8
>            Reporter: Jörg Hoh
>            Priority: Major
>
> During a maintenance of our database system the DocumentNodeStore bundle shut down because
of a lease timeout. When we restart the bundle, we ran into these exceptions:
> {code}
> 16.08.2018 11:32:44.139 *INFO* [Default Executor-thread-20544] org.apache.jackrabbit.oak.plugins.document.DocumentNodeStoreService
Configuring persistent cache to only cache nodes under paths [/libs, /apps, /jcr:system/jcr:nodeTypes,
/jcr:system/rep:namespaces]
> 16.08.2018 11:32:44.155 *INFO* [Default Executor-thread-20544] org.apache.jackrabbit.oak.plugins.document.persistentCache.PersistentCache
start, url=/opt/aem/repository/sling/_cqa/cache,size=2048,binary=0
> 16.08.2018 11:32:44.156 *WARN* [Default Executor-thread-20544] org.apache.jackrabbit.oak.plugins.document.persistentCache.MapFactory
Error in the background thread of the persistent cache: java.lang.IllegalStateException: The
file is locked: nio:/opt/aem/repository/sling/_cqa/cache/cache-0.data [1.4.194/7]
> 16.08.2018 11:32:44.157 *WARN* [Default Executor-thread-20544] org.apache.jackrabbit.oak.plugins.document.persistentCache.MapFactory
Could not open the store /opt/aem/repository/sling/_cqa/cache/cache-0.data
> java.lang.IllegalStateException: The file is locked: nio:/opta/aem/repository/sling/_cqa/cache/cache-0.data
[1.4.194/7]
> at org.h2.mvstore.DataUtils.newIllegalStateException(DataUtils.java:765)
> at org.h2.mvstore.FileStore.open(FileStore.java:168)
> at org.h2.mvstore.MVStore.<init>(MVStore.java:347)
> at org.h2.mvstore.MVStore$Builder.open(MVStore.java:2930)
> at org.apache.jackrabbit.oak.plugins.document.persistentCache.PersistentCache$1.openStore(PersistentCache.java:288)
> at org.apache.jackrabbit.oak.plugins.document.persistentCache.PersistentCache.createMapFactory(PersistentCache.java:361)
> at org.apache.jackrabbit.oak.plugins.document.persistentCache.PersistentCache.<init>(PersistentCache.java:210)
> at org.apache.jackrabbit.oak.plugins.document.DocumentNodeStoreBuilder.getPersistentCache(DocumentNodeStoreBuilder.java:648)
> at org.apache.jackrabbit.oak.plugins.document.DocumentNodeStoreBuilder.buildCache(DocumentNodeStoreBuilder.java:627)
> at org.apache.jackrabbit.oak.plugins.document.DocumentNodeStoreBuilder.buildDocumentCache(DocumentNodeStoreBuilder.java:586)
> at org.apache.jackrabbit.oak.plugins.document.DocumentNodeStoreBuilder.buildNodeDocumentCache(DocumentNodeStoreBuilder.java:594)
> at org.apache.jackrabbit.oak.plugins.document.rdb.RDBDocumentStore.initialize(RDBDocumentStore.java:967)
> {code}
> It seems that the mvstore wasn't shut down properly during the stop of the DocumentNodeStore.
For me it seems that the DocumentNodeStore doesn't use the persistent cache anymore until
there was a "hard" restart (on the JVM).
> It would be good if even in this situation the shutdown of the DocumentNodeStore could
drop the lock on the mvstore, so a clean restart of the bundle is possible.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message