[
https://issues.apache.org/jira/browse/HBASE-9037?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13721299#comment-13721299
]
Jean-Daniel Cryans commented on HBASE-9037:
-------------------------------------------
>From a real world, read-heavy, in-production region server:
{noformat}
$ grep "Block cache LRU eviction completed"
/path/to/logs/hbase-hadoop-regionserver-host.log.2013-07-25 | wc -l
49
{noformat}
Here's one example:
{noformat}
2013-07-25 00:13:36,934 DEBUG org.apache.hadoop.hbase.io.hfile.LruBlockCache:
Block cache LRU eviction completed; freed=1.69 GB, total=12.69 GB, single=5 GB,
multi=9.24 GB, memory=0 KB
{noformat}
bq. I see these things when I run perf tests all the time
Like I said previously, this is the only time it's going to annoy you.
bq. it is unreasonable to expect for all data to be in cache in real-world
scenarios imho
Real-world scenarios usually retain most of the working data set in cache.
> LruBlockCache is too verbose at Debug level
> -------------------------------------------
>
> Key: HBASE-9037
> URL: https://issues.apache.org/jira/browse/HBASE-9037
> Project: HBase
> Issue Type: Bug
> Reporter: Lars Hofhansl
>
> From Stack on the mailing list:
> {quote}
> I see this every 4ms or so when reading.
> 21 2013-07-24 10:55:24,594 DEBUG
> org.apache.hadoop.hbase.io.hfile.LruBlockCache: Block cache LRU eviction
> completed; freed=24.78 MB, total=185.13 MB, single=0 KB, multi=207.88 MB,
> memory=0.64 KB
> 20 2013-07-24 10:55:28,975 DEBUG
> org.apache.hadoop.hbase.io.hfile.LruBlockCache: Block cache LRU eviction
> started; Attempting to free 24.75 MB of total=209.91 MB
> Loads of it.
> {quote}
> Should either limit the log frequency of this, or change to trace.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
