hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Appy (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-14743) Add metrics around HeapMemoryManager
Date Sat, 18 Jun 2016 01:47:05 GMT

    [ https://issues.apache.org/jira/browse/HBASE-14743?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15337420#comment-15337420

Appy commented on HBASE-14743:

Testing the metrics:
Set these configs to turn on tuner and set memstore size to ~ 250bytes (4G * 0.00000005)

So i see the metrics being updated. But one weird thing is, one moment blockedflush histogram
has all these values, and they are all gone in next refresh. You can compare num_ops values
to get sense of relative timing of snapshots.

> Add metrics around HeapMemoryManager
> ------------------------------------
>                 Key: HBASE-14743
>                 URL: https://issues.apache.org/jira/browse/HBASE-14743
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: Elliott Clark
>            Assignee: Reid Chan
>            Priority: Minor
>         Attachments: HBASE-14743.001.patch, HBASE-14743.002.patch, HBASE-14743.003.patch,
HBASE-14743.004.patch, HBASE-14743.005.patch, HBASE-14743.006.patch, HBASE-14743.007.patch,
HBASE-14743.008.patch, HBASE-14743.009.patch, HBASE-14743.009.rw3.patch, HBASE-14743.009.v2.patch,
Screen Shot 2016-06-16 at 5.39.13 PM.png
> it would be good to know how many invocations there have been.
> How many decided to expand memstore.
> How many decided to expand block cache.
> How many decided to do nothing.
> etc.
> When that's done use those metrics to clean up the tests.

This message was sent by Atlassian JIRA

View raw message