You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@openwhisk.apache.org by gi...@git.apache.org on 2017/09/21 22:47:58 UTC

[GitHub] tysonnorris opened a new pull request #2795: enable concurrent activation processing

tysonnorris opened a new pull request #2795: enable concurrent activation processing
URL: https://github.com/apache/incubator-openwhisk/pull/2795
 
 
   - requires rebuilding nodejs6Action
   - considers all actions support concurrency
   - log collection is disabled for now due to interleaved logging when running concurrent activations in a single container
   
   To validate operation, I've tested with the throughput.sh script using an action that has an async response after 175ms. In this type of scenario, I see throughput 2x over previous sequential processing (e.g. 10-15 rps before, 20-40 rps after).
   
   Note that as discussed offline, in current form, with kafka between controller and invoker, there is no way to account for non-concurrent actions blocking resource usage by those that support concurrency, so currently the code assumes that all actions support concurrency, and the ContainerPoolScheduler accounts for concurrency settings per action, while MessageFeed only accounts for concurrency across the invoker in general (via maximumHandlerCapacity)
 
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services