GitHub user agresch opened a pull request:
https://github.com/apache/storm/pull/2659
STORM-3042 restore topology.acker.cpu.pcore.percent
TOPOLOGY_ACKER_CPU_PCORE_PERCENT and other similar settings in
adjustResourcesForExec() were orphaned accidentally by a previous pull request.
This adds the functionality back.
When we deployed the change to clusters that were near full with some
topologies that had this setting, because we were not previously honoring it,
the nodes now looked over 100% full and were throwing exceptions during
scheduling.
The fix for this was to instead zero out the resources and track that the
issue is occurring. We had also logged that the error was occurring, and found
that the logging became excessive, which also slowed scheduling. A further
change was added by @revans2 to limit the logging.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/agresch/storm agresch_storm3042
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/storm/pull/2659.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #2659
----
commit 75dc1825252ecbadf3317a141458bee6323404f6
Author: Aaron Gresch <agresch@...>
Date: 2018-05-03T18:09:56Z
STORM-3042 restore topology.acker.cpu.pcore.percent
----
---
