[
https://issues.apache.org/jira/browse/HADOOP-3422?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12598652#action_12598652
]
Jason commented on HADOOP-3422:
-------------------------------
The names being produced for ganglia are long and tend to overflow the bounding
box in the graph.
mapred.job.FQCN.COUNTER_NAME
where FQCN is the fully qualified class that contains the counter ie:
org.apache.hadoop.mapred.XXX.YYY
And it would be nice to get this from the individual machines as well as the
cluster agreggate so you could tell at a glance how the individual machines
were doing. I suspect that will be annother patch for that new feature.
> Ganglia counter metrics are all reported with the metric name "value", so the
> counter values can not be seen
> ------------------------------------------------------------------------------------------------------------
>
> Key: HADOOP-3422
> URL: https://issues.apache.org/jira/browse/HADOOP-3422
> Project: Hadoop Core
> Issue Type: Bug
> Components: metrics
> Affects Versions: 0.16.4
> Reporter: Jason
> Fix For: 0.18.0
>
> Attachments: diff-20080520-1025.txt
>
> Original Estimate: 1h
> Remaining Estimate: 1h
>
> The JobInProgress class reports all metrics with the name "value". The
> FileMetrics class puts all of the tags into the name when reporting the
> individual values, but the Ganglia Context does not put the tags into the
> name..
> This patch modifies the context to build names for the counter metrics out of
> the tag values. This enables the user to see the indivdual counter values
> with the ganglia web tool, on a per job basis
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
