Github user foxish commented on a diff in the pull request:
https://github.com/apache/spark/pull/20032#discussion_r158148895
--- Diff:
resource-managers/kubernetes/core/src/main/scala/org/apache/spark/scheduler/cluster/k8s/KubernetesClusterSchedulerBackend.scala
---
@@ -86,7 +86,7 @@ private[spark] class KubernetesClusterSchedulerBackend(
private val initialExecutors =
SchedulerBackendUtils.getInitialTargetExecutorNumber(conf)
- private val podAllocationInterval =
conf.get(KUBERNETES_ALLOCATION_BATCH_DELAY)
+ private val podAllocationInterval =
conf.getTimeAsMs(KUBERNETES_ALLOCATION_BATCH_DELAY.key)
--- End diff --
That was the previous state actually. Not sure what the best practice is
here. @mridulm, thoughts? `getTimeAsMs` seemed better as it might protect us
from changes in the config if at all that happens. It enforces the contract
that it must be milliseconds, which is essential for the allocator to function
correctly.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
