[
https://issues.apache.org/jira/browse/HADOOP-13001?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15226007#comment-15226007
]
Steve Loughran commented on HADOOP-13001:
-----------------------------------------
that would be good. What I wanted was to set the RM client up to have a policy
of try once then fail —no retry—, but as the retry policy is hard coded, all I
can do is change the timeouts. Setting interval=0 triggers this.
Setting maxwait=0 and interval=1 works, though you get warned that as maxwait <
interval, you don't get any retries.
I think I'd like there special case of "maxwait=0" to be recognised as "you
don't want retries".
If you do have a maxwait >0, and an interval ==0, well, you do have a problem
in your configuration, don't you? That could be be rejected more gracefully
than a division by zero exception.
> RetryPolicies$RetryUpToMaximumTimeWithFixedSleep raises division by zero
> exception if the sleep time is 0, even if max wait == 0
> --------------------------------------------------------------------------------------------------------------------------------
>
> Key: HADOOP-13001
> URL: https://issues.apache.org/jira/browse/HADOOP-13001
> Project: Hadoop Common
> Issue Type: Bug
> Components: ipc
> Affects Versions: 2.7.2
> Reporter: Steve Loughran
>
> set YARN RM max wait and retry intervals to 0, try to talk to an RM, get an
> arithmetic exception
> {code}
> Caused by: java.lang.ArithmeticException: / by zero
> at
> org.apache.hadoop.io.retry.RetryPolicies$RetryUpToMaximumTimeWithFixedSleep.<init>(RetryPolicies.java:265)
> at
> org.apache.hadoop.io.retry.RetryPolicies.retryUpToMaximumTimeWithFixedSleep(RetryPolicies.java:89)
> at
> org.apache.hadoop.yarn.client.RMProxy.createRetryPolicy(RMProxy.java:237)
> at
> org.apache.hadoop.yarn.client.RMProxy.createRMProxy(RMProxy.java:91)
> at
> org.apache.hadoop.yarn.client.ClientRMProxy.createRMProxy(ClientRMProxy.java:72)
> at
> org.apache.hadoop.yarn.client.api.impl.YarnClientImpl.serviceStart(YarnClientImpl.java:188)
> at
> org.apache.hadoop.service.AbstractService.start(AbstractService.java:193)
> at
> org.apache.spark.deploy.history.yarn.server.YarnHistoryProvider.initYarnClient(YarnHistoryProvider.scala:931)
> at
> org.apache.spark.deploy.history.yarn.server.YarnHistoryProvider.init(YarnHistoryProvider.scala:296)
> {code}
> I'd have expected the code to recognise the caller is saying "don't retry",
> but clearly not
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
