You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@hbase.apache.org by "Andrew Purtell (JIRA)" <ji...@apache.org> on 2015/04/11 02:38:13 UTC

[jira] [Resolved] (HBASE-3302) To prevent possible overrun of block cache with CacheOnWrite, add safeguard so we reject blocks if completely full

     [ https://issues.apache.org/jira/browse/HBASE-3302?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Andrew Purtell resolved HBASE-3302.
-----------------------------------
    Resolution: Incomplete
      Assignee:     (was: Jonathan Gray)

> To prevent possible overrun of block cache with CacheOnWrite, add safeguard so we reject blocks if completely full
> ------------------------------------------------------------------------------------------------------------------
>
>                 Key: HBASE-3302
>                 URL: https://issues.apache.org/jira/browse/HBASE-3302
>             Project: HBase
>          Issue Type: Improvement
>          Components: io, regionserver
>            Reporter: Jonathan Gray
>
> With the aggressive caching when CacheOnWrite is turned on, and given the current LRU architecture, there's potential (though low probability) we could overrun the block cache capacity by caching faster than we can evict.
> Currently the block cache triggers eviction at 85% capacity.  If somehow we attempt to cache a block but the cache is at 100% capacity, we should reject caching of that block.



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