apex-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chandni Singh (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (APEXMALHAR-2321) Improve Buckets memory management
Date Tue, 01 Nov 2016 17:45:00 GMT

    [ https://issues.apache.org/jira/browse/APEXMALHAR-2321?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15626117#comment-15626117
] 

Chandni Singh commented on APEXMALHAR-2321:
-------------------------------------------


>>AbstractManagedStateImpl.maxMemorySize probably will be misunderstand as the max memory
size of total managed state, but in fact it was used as memory size of each bucket. Better
to rename it.

This is incorrect. This setting is not used per bucket. Please look at lines 94-98 in StateTracker.
Bytes are summed over all buckets and then compared to maxMemorySize.

> Improve Buckets memory management
> ---------------------------------
>
>                 Key: APEXMALHAR-2321
>                 URL: https://issues.apache.org/jira/browse/APEXMALHAR-2321
>             Project: Apache Apex Malhar
>          Issue Type: Improvement
>            Reporter: bright chen
>            Assignee: bright chen
>
> Currently buckets were managed as an array. Each bucket have memory limitation, and free
memory will be triggered if the bucket memory usage over the limitation.
> - For ManagedTimeUnifiedStateImpl, the default bucket number is 345600, which probably
too large.
> - AbstractManagedStateImpl.maxMemorySize probably will be misunderstand as the max memory
size of total managed state, but in fact it was used as memory size of each bucket. Better
to rename it.
> - The default maxMemorySize is zero. It's better to give a default reasonable value to
avoid too much garbage collection 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message