Github user ehnalis commented on a diff in the pull request:
https://github.com/apache/spark/pull/6082#discussion_r30338968
--- Diff:
yarn/src/main/scala/org/apache/spark/deploy/yarn/ApplicationMaster.scala ---
@@ -300,11 +300,14 @@ private[spark] class ApplicationMaster(
val expiryInterval =
yarnConf.getInt(YarnConfiguration.RM_AM_EXPIRY_INTERVAL_MS, 120000)
// we want to be reasonably responsive without causing too many
requests to RM.
- val schedulerInterval =
- sparkConf.getTimeAsMs("spark.yarn.scheduler.heartbeat.interval-ms",
"5s")
+ val heartbeatInterval = math.max(0, math.min(expiryInterval / 2,
+ sparkConf.getTimeAsMs("spark.yarn.scheduler.heartbeat.interval-ms",
"5s")))
- // must be <= expiryInterval / 2.
- val interval = math.max(0, math.min(expiryInterval / 2,
schedulerInterval))
+ // we want to check more frequently for pending containers
+ val initialAllocationInterval = math.min(heartbeatInterval,
+
sparkConf.getTimeAsMs("spark.yarn.scheduler.initial-allocation.interval",
"200ms"))
--- End diff --
Thanks for finding that! I've tested it with 50ms and did not notice it
after I've refactored the debug log. About to fix it soon.
---
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]
