You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
…cheduling (palantir#102)
* partial work to use node affinity instead of labels
* get dependencies right
* Remove occurences of pod.Spec.NodeSelector
* Expose FindInstanceGroup
* move find instance group in internal package
* updated tests
* fixed vendor dependencies
* upgrade k8s.spark-scheduler-lib
* fix unit tests
* running verify
* publish docker snapshot from any branch temporarily
* some refactor
* use local filter instead of all for publishing
* Instead of label match use node affinity predicate
* test the release from my branch
* revert circle config changes after testing
* Fix flaky test
* Address some of Onur's comments
* remove named return type
Co-authored-by: Onur Satici <[email protected]>
Today, if the
metric-emit-frequency
is set to0
we use the default of 60s. This is different thantrace-sample-rate
, which disables logging of <= 0.It would be nice to be able to disable metrics via config alone.
The text was updated successfully, but these errors were encountered: