rmatharu commented on a change in pull request #1104: SAMZA-2266: Introduce a
backoff when there are repeated failures for host-affinity allocations
URL: https://github.com/apache/samza/pull/1104#discussion_r303562373
##########
File path:
samza-core/src/main/java/org/apache/samza/config/ClusterManagerConfig.java
##########
@@ -87,7 +88,10 @@
*/
public static final String CONTAINER_RETRY_COUNT =
"yarn.container.retry.count";
public static final String CLUSTER_MANAGER_CONTAINER_RETRY_COUNT =
"cluster-manager.container.retry.count";
- private static final int DEFAULT_CONTAINER_RETRY_COUNT = 8;
+ public static final int DEFAULT_CONTAINER_RETRY_COUNT = 8;
+
+ public static final String CLUSTER_MANAGER_CONTAINER_RETRY_MAX_DELAY_MS =
"cluster-manager.container.host-affinity-retry.max.delay.ms";
+ public static final long DEFAULT_CONTAINER_RETRY_MAX_DELAY_MS =
Duration.ofSeconds(120).toMillis();
Review comment:
This should be in greater than the max amount of time it takes for the RM to
mark a dead NM unhealthy,
which in YARN 2.9.2 is 5 minutes?
So perhaps not configurable and set to a higher value.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
