[
https://issues.apache.org/jira/browse/HBASE-16450?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15431909#comment-15431909
]
Matteo Bertozzi commented on HBASE-16450:
-----------------------------------------
looks good to me. below a list of "cleanups". but nothing important.
(up to you if put those as improvement for another jira)
I don't see the new ZKUtil.getReplicationZnodesDump(), is this a left over from
the initial patch or is there a missing file calling this method?
why the --hdfs and --distributed parseOpts() uses startsWith() and not equals()?
that parseOpts() with the queue is interesting. in general we do a simple for
loop, but the while looks nice. it is unfortunate the double conversion
list->linkedList in the run() but not important here. (not sure if we want to
keep the standard for loop, but I like the form here with the poll()). There is
also AbstractHBaseTool that Appy just moved to common that have a different way
to handle args, but at the moment any tool is different. so.. keep it as it is
for now. we may bulk change every tool when we will decide something.
why dumpReplicationQueues() uses LOG.info() instead of System.out? I think in
other tools we use System.out
In getTotalWALSize() you can try to use new WALLink(getConf(), serverName,
wal).getFileStatus() which resolve the logdir and oldLogDir for you
> Shell tool to dump replication queues
> --------------------------------------
>
> Key: HBASE-16450
> URL: https://issues.apache.org/jira/browse/HBASE-16450
> Project: HBase
> Issue Type: Improvement
> Components: Operability, Replication
> Affects Versions: 2.0.0, 1.3.0, 1.1.5, 1.2.2
> Reporter: Esteban Gutierrez
> Assignee: Esteban Gutierrez
> Attachments: HBASE-16450.branch-1.001.patch,
> HBASE-16450.master.001.patch, HBASE-16450.master.002.patch
>
>
> Currently there is no way to dump list of the configured queues and the
> replication queues when replication is enabled. Unfortunately the HBase
> master only offers an option to dump the whole content of the znodes but not
> details on the queues being processed on each RS.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
