Github user vanzin commented on a diff in the pull request:
https://github.com/apache/spark/pull/6394#discussion_r31572115
--- Diff:
yarn/src/main/scala/org/apache/spark/deploy/yarn/YarnAllocator.scala ---
@@ -427,11 +474,25 @@ private[yarn] class YarnAllocator(
containerSet.remove(containerId)
if (containerSet.isEmpty) {
allocatedHostToContainersMap.remove(host)
+ // If all the containers of this node are removed, remove this
preferred location,
+ // since it means currently we don't need to this node locality.
+ preferredNodeLocationToUses.remove(host)
} else {
allocatedHostToContainersMap.update(host, containerSet)
}
allocatedContainerToHostMap.remove(containerId)
+
+ // If current there's no container allocated on this rack, remove
it,
+ // since we don't need this rack currently.
+ val remainingRacks = allocatedHostToContainersMap.keys.map { h =>
+ RackResolver.resolve(conf, h).getNetworkLocation
--- End diff --
Probably not a huge deal, but isn't `RackResolver.resolve` a little
expensive? IIRC it launches a subprocess that can do arbitrary things. Perhaps
at some point it would be nice to cache this information.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
