[
https://issues.apache.org/jira/browse/HIVE-11891?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14876773#comment-14876773
]
Brock Noland edited comment on HIVE-11891 at 9/19/15 1:03 AM:
--------------------------------------------------------------
I was looking at it again and I think it can be done by moving most
{{getPerfLogger}} calls to {{SessionState}}. Updated patch attached.
Note this patch logs at {{INFO}} since that is what PerfLogger does by default.
I think that is OK since the HMS logs a few messages per call to the HMS anyway.
Example output:
{noformat}
2015-09-18T18:01:04,389 INFO [pool-6-thread-1]: log.PerfLogger
(PerfLogger.java:PerfLogBegin(117)) - <PERFLOG method=set_ugi
from=org.apache.hadoop.hive.metastore.RetryingHMSHandler>
2015-09-18T18:01:04,389 INFO [pool-6-thread-1]: log.PerfLogger
(PerfLogger.java:PerfLogEnd(157)) - </PERFLOG method=set_ugi
start=1442624464388 end=1442624464389 duration=1
from=org.apache.hadoop.hive.metastore.RetryingHMSHandler threadId=1
retryCount=0 error=false>
{noformat}
was (Author: brocknoland):
I was looking at it again and I think it can be done by moving most
{{getPerfLogger}} calls to {{SessionState}}. Updated patch attached.
Note this patch logs at {{LOG}} since that is what PerfLogger does by default.
I think that is OK since the HMS logs a few messages per call to the HMS anyway.
Example output:
{noformat}
2015-09-18T18:01:04,389 INFO [pool-6-thread-1]: log.PerfLogger
(PerfLogger.java:PerfLogBegin(117)) - <PERFLOG method=set_ugi
from=org.apache.hadoop.hive.metastore.RetryingHMSHandler>
2015-09-18T18:01:04,389 INFO [pool-6-thread-1]: log.PerfLogger
(PerfLogger.java:PerfLogEnd(157)) - </PERFLOG method=set_ugi
start=1442624464388 end=1442624464389 duration=1
from=org.apache.hadoop.hive.metastore.RetryingHMSHandler threadId=1
retryCount=0 error=false>
{noformat}
> Add basic performance logging at trace level to metastore calls
> ---------------------------------------------------------------
>
> Key: HIVE-11891
> URL: https://issues.apache.org/jira/browse/HIVE-11891
> Project: Hive
> Issue Type: Improvement
> Components: Metastore
> Affects Versions: 1.0.0, 1.2.0, 1.1.0
> Reporter: Brock Noland
> Assignee: Brock Noland
> Priority: Minor
> Fix For: 2.0.0
>
> Attachments: HIVE-11891.patch, HIVE-11891.patch, HIVE-11891.patch
>
>
> At present it's extremely difficult to debug slow calls to the metastore.
> Ideally there would be some basic means of doing so.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
