[
https://issues.apache.org/jira/browse/OAK-4543?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15375201#comment-15375201
]
Vikas Saurabh edited comment on OAK-4543 at 7/18/16 11:05 AM:
--------------------------------------------------------------
bq. we should not only log slow cases but also provide metrics that propagate
to the system/console and allow identifying slow listeners over time better.
Yes, I was thinking about that and it seemed that we should probably attach
such feature in PerfLogger itself (rename it too??). So, the advantage gets
propagated to anyone interested in logging for performance.
Btw, an open question still is "can we get a more consolidated event for
diff/processing-of-event/diff+processing for a given revision pair instead of
timings for revision-pair,path tuple?". I think the jmx data would be useful if
we can get consolidated timings. i.e. while post-processing logs can give
insights. I "think" it's hard to post-processing on data exposed on jmx
(rather, I'd probably like to have a consolidated info on jmx for easier
consumtion)
was (Author: catholicon):
bq. we should not only log slow cases but also provide metrics that propagate
to the system/console and allow identifying slow listeners over time better.
Yes, I was thinking about that and it seemed that we should probably attach
such feature in PerfLogger itself (rename it too??). So, the advantage gets
propagated to anyone interested in logging for performance.
> Add PerfLogger around observation event generation and event consumption by
> observer
> ------------------------------------------------------------------------------------
>
> Key: OAK-4543
> URL: https://issues.apache.org/jira/browse/OAK-4543
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Reporter: Vikas Saurabh
> Assignee: Vikas Saurabh
> Priority: Minor
> Labels: candidate_oak_1_0, candidate_oak_1_2, candidate_oak_1_4,
> monitoring, observation, performance
>
> I'm not sure if it's possible in the current scheme of things
> (implementation), but it'd useful to be able to easily differentiate between
> slow diff calculation or slow observer as a reason to see why observation
> queue might fill up.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
