hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jean-Daniel Cryans (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HBASE-3049) HLog size explosion
Date Fri, 22 Oct 2010 17:18:04 GMT

     [ https://issues.apache.org/jira/browse/HBASE-3049?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Jean-Daniel Cryans updated HBASE-3049:
--------------------------------------

    Fix Version/s:     (was: 0.89.20100924)
                   0.90.0

> HLog size explosion
> -------------------
>
>                 Key: HBASE-3049
>                 URL: https://issues.apache.org/jira/browse/HBASE-3049
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Jean-Daniel Cryans
>             Fix For: 0.90.0
>
>
> Recently I've been seeing some strange behavior around HLogs, HBASE-3038 is an example.
Looking in my oldlogs folder, I see:
> {noformat}
> -rw-r--r--   3 hadoop supergroup  716716459 2010-09-29 01:08 /hbase/.oldlogs/10.20.20.176%3A60020.1285722528252
> -rw-r--r--   3 hadoop supergroup   64841781 2010-09-29 01:10 /hbase/.oldlogs/10.20.20.176%3A60020.1285722620286
> bunch of normally sized HLogs... then
> -rw-r--r--   3 hadoop supergroup  769729956 2010-09-29 01:13 /hbase/.oldlogs/10.20.20.176%3A60020.1285722785347
> {noformat}
> 680MB is way off the 64MB limit. My feeling is that the optimizations I did in the scope
of HBASE-2922 that remove some blocking behavior have the side effect of getting a lot more
data in a lot faster in some edge cases. Need to investigate more.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message