[
https://issues.apache.org/jira/browse/OAK-9913?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17598459#comment-17598459
]
Stefan Egli commented on OAK-9913:
----------------------------------
I would see the following requirements for throttling metrics:
* it should be (easily) possible to determine the time window during which
throttling kicked in. As this is basically a binary thing: either throttling is
on and all (applicable) operations are throttled - or it is off (note that I
don't mean enabled/disabled here, but actually on or off). The throttling is
not per operation but on all operations of a particular Oak instance (actually
on all Oak instances in the cluster). I'm not sure if this aspect is covered by
the current plan. But it would seem like this would be a very simple, global,
singleton timer.
* it should also be possible to see how much each of the applicable operation
is throttled vs how much it would normally take. Depending on where exactly the
existing metrics are measured and where the new ones go, we'd have to ensure
that we can calculate the "{{normal time + throttling time = total time}}". One
aspect of that equation is already available from the existing metrics, so we'd
have to ensure we provide one of the other variable (but not more)
> Expose Metrics for Mongo DocumentStore throttling
> -------------------------------------------------
>
> Key: OAK-9913
> URL: https://issues.apache.org/jira/browse/OAK-9913
> Project: Jackrabbit Oak
> Issue Type: New Feature
> Reporter: Rishabh Kumar
> Priority: Major
>
> Expose metrics for the Mongo DocumentStore throttling feature. This allows us
> to better observe when the DocumentStore is throttled.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
