onebox-li opened a new pull request, #2078:
URL: https://github.com/apache/incubator-celeborn/pull/2078

   ### What changes were proposed in this pull request?
   Currently, `WorkerInfo` is used in many places, and allocationBuckets is 
only used when its own workers want to collect metrics `SLOTS_ALLOCATED`. If 
there are lots of workers in the RSS cluster, there may be a certain amount of 
memory waste, each `WorkerInfo` maintain a Array\[Int](61), so remove it from 
`WorkerInfo`. 
   And refactor the metrics `SLOTS_ALLOCATED` from gauge to counter. 
Originally, this metrics is approximately one hour's total only if there are 
continuous tasks. Now refactoring it into a counter can reduce the cost of 
maintaining time windows, including storage and timely expiration data, etc. It 
can also be more flexibly transformed according to user needs on the prometheus 
side. 
   
   
   ### Why are the changes needed?
   Ditto.
   
   
   ### Does this PR introduce _any_ user-facing change?
   Yes. metrics_SlotsAllocated_Count metrics change from gauge for 1 hour to a 
increasing counter. 
   
   
   ### How was this patch tested?
   Cluster test.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to