hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ramkrishna.s.vasudevan (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-16651) LRUBlockCache#returnBlock should try return block to Victim Handler L2 cache.
Date Mon, 19 Sep 2016 11:52:21 GMT

    [ https://issues.apache.org/jira/browse/HBASE-16651?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15503193#comment-15503193

ramkrishna.s.vasudevan commented on HBASE-16651:

bq. There we have the SHARED mode check before deciding whether to decr a counter.
Fine then.
bq.Where is that comment/discussion?
This is long back. I don know whether it is in the JIRA or not. But am pretty sure we had
a discussion if to allow combinedMode=false.
bq.we will need L2 as victim for L1? When L1 is out of space cache them in L2. The blocks
there are more frequently used ones
Is there any minimum size that we say has to be configured for L1? Ya for bigger setups there
is a chance we may run out of space if the active set is big enough than the L1. So explicitly
we don't have to use the mode with false but inherently L1 may have a victim handler?

> LRUBlockCache#returnBlock should try return block to Victim Handler L2 cache.
> -----------------------------------------------------------------------------
>                 Key: HBASE-16651
>                 URL: https://issues.apache.org/jira/browse/HBASE-16651
>             Project: HBase
>          Issue Type: Sub-task
>          Components: regionserver, Scanners
>    Affects Versions: 2.0.0
>            Reporter: Anoop Sam John
>            Assignee: Anoop Sam John
>             Fix For: 2.0.0
>         Attachments: HBASE-16651.patch
> In case of L1 and L2 cache usage with combinedMode = false, L2 is used as a victim handler
cache for L1 cache.  When a getBlock() request comes, L1 will see if block is in it and if
not it will try to provide the block from L2 cache. In such a case, the return block must
return the block to L2 cache and count down the ref count for the block.  But right now we
just ignore the returnBlock call in LRUCache

This message was sent by Atlassian JIRA

View raw message