[ 
https://issues.apache.org/jira/browse/HBASE-3016?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

stack resolved HBASE-3016.
--------------------------

      Assignee: stack
    Resolution: Fixed

Committed.  Here's what was done:

{code}
M conf/log4j.properties
  Make zkutil logging be INFO level so default doesn't include
  detail on size of znode, etc., whenever we interact w/ zk (but
  later if needed can reenable)
M src/main/java/org/apache/hadoop/hbase/zookeeper/ZooKeeperWatcher.java
  Add sessionid to what was called the zookeeperwatcher 'name'.
  Make toString return this identifier.  Add a prefix method
  that will prefix this zkw identifier to all log messages.
  Removed all of the logging methods.
M src/main/java/org/apache/hadoop/hbase/zookeeper/ZKUtil.java
M src/main/java/org/apache/hadoop/hbase/zookeeper/ZKAssign.java
  Use new zkw.prefix method and use new local LOG instance
  rather than have zkw log for us.
M src/main/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java
  Remove extaneous log.
M src/main/java/org/apache/hadoop/hbase/master/ActiveMasterManager.java
  Use zkw prefix and local logger rather than zkws.
M src/main/java/org/apache/hadoop/hbase/client/HConnectionManager.java
  Make a more sensible identifier
{code}

> Refactor zk logging in trunk; do less and and identifier to log messages to 
> help debugging
> ------------------------------------------------------------------------------------------
>
>                 Key: HBASE-3016
>                 URL: https://issues.apache.org/jira/browse/HBASE-3016
>             Project: HBase
>          Issue Type: Improvement
>          Components: zookeeper
>            Reporter: stack
>            Assignee: stack
>             Fix For: 0.90.0
>
>         Attachments: 3016.txt
>
>
> Currently all logging is done by zookeeperwatcher so can add a zkw identifier 
> on the front of the log message.  This happens for logging done by other 
> classes -- zkutil, zkconfig -- which makes it so can't turn off logging using 
> log4j config (its all or nothing).
> Also, zk gives connections a sessionid.  To tie zk logs to hbase logs, we 
> should include sessionid in the zkw identifier.  It helps debugging 
> especially when using minihbasecluster where clients, master, rs, and zk 
> server are all logging to the same file.

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

Reply via email to