[
https://issues.apache.org/jira/browse/OAK-3330?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14727206#comment-14727206
]
Michael Dürig commented on OAK-3330:
------------------------------------
Caching all segments of the current tar writer improves the situation quite a
bit. With below configuration of {{SegmentCompactionIT}} I was able to write
2.5 GB of content vs. 5.8 GB of content with the cache. In the former case the
{{FileStore}} monitor was fully contented while in the latter no lock was
contended more then 4%.
{code}
private volatile int lockWaitTime = 60;
private volatile int maxReaders = 10;
private volatile int maxWriters = 16;
private volatile long maxStoreSize = 120000000000L;
private volatile int maxBlobSize = 10000;
private volatile int maxStringSize = 10000;
private volatile int maxReferences = 10;
private volatile int maxWriteOps = 10000;
private volatile int maxNodeCount = 1000;
private volatile int maxPropertyCount = 1000;
private volatile int nodeRemoveRatio = 10;
private volatile int propertyRemoveRatio = 10;
private volatile int nodeAddRatio = 40;
private volatile int addStringRatio = 20;
private volatile int addBinaryRatio = 0;
private volatile int compactionInterval = 120;
{code}
> FileStore lock contention with concurrent writers
> -------------------------------------------------
>
> Key: OAK-3330
> URL: https://issues.apache.org/jira/browse/OAK-3330
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: segmentmk
> Reporter: Michael Dürig
> Assignee: Michael Dürig
> Labels: compaction
>
> Concurrently writing to the file store can lead to a sever lock contention in
> {{FileStore#readSegment}}. That method searches the current {{TarWriter}}
> instance for the segment once it could not be found in any of the
> {{TarReader}} instances. This is the point where synchronizes on the
> {{FileStore}} instance, which leads to the contention.
> The effect is only observable once the segment cache becomes full and reads
> actually need to go to the file store. Thus a possible improvement could be
> to pin segments from the current tar writer to the cache. Alternatively we
> could try to ease locking by employing read/write locks where possible.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
