> On Oct. 18, 2017, 2 p.m., Stephan Erb wrote:
> > src/main/java/org/apache/aurora/scheduler/offers/OfferManager.java
> > Lines 203-205 (original), 204-206 (patched)
> > <https://reviews.apache.org/r/63121/diff/1/?file=1862746#file1862746line204>
> >
> >     By default, `min_offer_hold_time` is set to 5 minutes so unused offers 
> > are declined roughly every 5 minutes. This will also clean up related 
> > static bans.
> >     
> >     Are you seeing the memory increase within these 5 minutes? Or have you 
> > increased the offer hold time significantly?

> Or have you increased the offer hold time significantly?

We are effectively disabling the proactive offer decline mechanism by raising 
offer hold time.  Proactively declining is unnecessary in clusters with 
singleton frameworks, which is how Twitter uses Aurora.  Holding offers 
indefinitely reduces a major source of churn in the scheduler, and makes the 
static offer bans more effective.


- Bill


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/63121/#review188578
-----------------------------------------------------------


On Oct. 18, 2017, 12:40 p.m., Bill Farner wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/63121/
> -----------------------------------------------------------
> 
> (Updated Oct. 18, 2017, 12:40 p.m.)
> 
> 
> Review request for Aurora and Jordan Ly.
> 
> 
> Repository: aurora
> 
> 
> Description
> -------
> 
> This alleviates a (slow) memory leak in static offer bans, as entries are only
> removed when an offer is removed.  If a pending task group is depleted
> (either by fully scheduling the group, or terminating the job), the entry
> remains.  This issue is exacerbated when offers are held for a longer 
> duration,
> as is proposed in https://reviews.apache.org/r/62956/.
> 
> 
> Diffs
> -----
> 
>   src/main/java/org/apache/aurora/scheduler/events/PubsubEvent.java 
> 0637eb7f85125cf70b588d56fa7dc88130947837 
>   src/main/java/org/apache/aurora/scheduler/offers/OfferManager.java 
> e8334310a2a46a0ccb09ee6e4122c515892d3996 
>   src/main/java/org/apache/aurora/scheduler/scheduling/TaskGroups.java 
> 2d3492d05986ef65519fd7a8c71396d055b6881f 
>   src/test/java/org/apache/aurora/scheduler/scheduling/TaskGroupsTest.java 
> b88d5f13889b81ba4b0171efaf6c759d23976a39 
> 
> 
> Diff: https://reviews.apache.org/r/63121/diff/1/
> 
> 
> Testing
> -------
> 
> This patch needs testing, and is not yet ready for submission.
> 
> 
> Thanks,
> 
> Bill Farner
> 
>

Reply via email to