[
https://issues.apache.org/jira/browse/AURORA-137?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15685633#comment-15685633
]
Mehrdad Nurolahzade commented on AURORA-137:
--------------------------------------------
In our clusters, I have noticed that the top three methods in terms of total
wait time to acquire the storage lock are currently the following:
- {{org.apache.aurora.scheduler.BatchWorker.processBatch()}}
-
{{org.apache.aurora.scheduler.reconciliation.TaskTimeout.TimedOutTaskHandler.run()}}
- {{org.apache.aurora.scheduler.TaskStatusHandlerImpl.run()}}
This is primarily due to very high frequency of invocations to these methods.
Wait times experienced by
{{org.apache.aurora.scheduler.mesos.MesosSchedulerImpl.resourceOffers()}} is an
order of magnitude smaller compared to the above three processes (perhaps
because the frequency of {{storage.write()}} invocations here similarly shrinks
an orders of magnitude). However, it might still be worthwhile to reduce this
frequency even further if all it takes is just a simple change.
> Save host attributes only when a task is being scheduled
> --------------------------------------------------------
>
> Key: AURORA-137
> URL: https://issues.apache.org/jira/browse/AURORA-137
> Project: Aurora
> Issue Type: Story
> Components: Scheduler
> Reporter: Bill Farner
> Priority: Minor
>
> The scheduler currently aggressively saves host attributes when handling
> {{resourceOffers}}, however it seems tractable for this to only happen when a
> task is actually scheduled. Context: the scheduler stores host attributes to
> satisfy scheduling constraints (like host/rack diversity). Doing this would
> allow us to avoid waiting for the storage write lock, and handle
> {{resourceOffers}} in a more deterministic time frame.
> One caveat with this approach is that the Offer would need to be plumbed into
> {{SchedulingFilterImpl}} in a way so as to ensure that the attributes are
> available for the offer being inspected. In other words, we need to avoid
> the chicken and egg of trying to read the attributes for a host when this is
> the first offer ever received for it.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
