> On June 19, 2014, 10:47 p.m., Bill Farner wrote:
> > Can you elaborate (preferably in a comment) why an operator would want to 
> > adjust this, rather than exposing the data as soon as it's available?
> 
> Kevin Sweeney wrote:
>     What do you have in mind here?

Presumably this would be similar to the motivation for this change.  However, i 
just realized there's a gotcha here - you're probably interested in tightening 
the time to export this once a leader is available, but this will shorten the 
first export attempt after application startup, which won't work well for a 
non-leader.


- Bill


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/22794/#review46235
-----------------------------------------------------------


On June 19, 2014, 10:32 p.m., Kevin Sweeney wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/22794/
> -----------------------------------------------------------
> 
> (Updated June 19, 2014, 10:32 p.m.)
> 
> 
> Review request for Aurora and Bill Farner.
> 
> 
> Bugs: AURORA-511
>     https://issues.apache.org/jira/browse/AURORA-511
> 
> 
> Repository: aurora
> 
> 
> Description
> -------
> 
> Expose command-line flags to control the initial delay for stats
> collection independent of the collection interval.
> 
> This is related to AURORA-511 because it allows an operator to control the 
> interval that the machine-parseable /vars.json endpoint is updated with 
> resource utilization information, avoiding the need to scrape /utilization.
> 
> 
> Diffs
> -----
> 
>   src/main/java/org/apache/aurora/scheduler/stats/AsyncStatsModule.java 
> 3b5c61713f8b4a9f921e49ce12abbfd0bb28089f 
> 
> Diff: https://reviews.apache.org/r/22794/diff/
> 
> 
> Testing
> -------
> 
> ./gradlew -Pq build
> 
> 
> Thanks,
> 
> Kevin Sweeney
> 
>

Reply via email to