[
https://issues.apache.org/jira/browse/HBASE-1458?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack resolved HBASE-1458.
--------------------------
Resolution: Fixed
Committed the below:
{code}
Index: src/java/org/apache/hadoop/hbase/regionserver/HLog.java
===================================================================
--- src/java/org/apache/hadoop/hbase/regionserver/HLog.java (revision
780040)
+++ src/java/org/apache/hadoop/hbase/regionserver/HLog.java (working copy)
@@ -202,7 +202,7 @@
throw new IOException("Target HLog directory already exists: " + dir);
}
fs.mkdirs(dir);
- this.maxLogs = conf.getInt("hbase.regionserver.maxlogs", 64);
+ this.maxLogs = conf.getInt("hbase.regionserver.maxlogs", 32);
this.enabled = conf.getBoolean("hbase.regionserver.hlog.enabled", true);
LOG.info("HLog configuration: blocksize=" + this.blocksize +
", rollsize=" + this.logrollsize +
{code}
> 64 commit logs as upper bound is too many -- make it half
> ---------------------------------------------------------
>
> Key: HBASE-1458
> URL: https://issues.apache.org/jira/browse/HBASE-1458
> Project: Hadoop HBase
> Issue Type: Task
> Reporter: stack
> Fix For: 0.20.0
>
>
> Running an upload, 64 commit logs as upper bound before we force flushes to
> clear the oldest edit is too much. I can see an upload running in my little
> cluster and even after running for an hour we still have not hit the 64 logs
> max. The more logs we have, the longer recovery on crash. We should halve
> the number I'd say.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
