[ 
https://issues.apache.org/jira/browse/YARN-11399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17651256#comment-17651256
 ] 

ASF GitHub Bot commented on YARN-11399:
---------------------------------------

slfan1989 commented on PR #5234:
URL: https://github.com/apache/hadoop/pull/5234#issuecomment-1362774408

   @Daniel-009497 Thank you very much for your contribution, I read this part 
of code and I think this variable(`delegationTokenRenewerPoolTrackerFlag`) 
should always be true.
   
   This optimization comes from JIRA: YARN-9768. RM Renew Delegation token 
thread should timeout and retry.
   The description information explains the reason for the optimization:
   ```
   This call is made to an underlying HDFS NN or Router Node (which has exact 
APIs as HDFS NN). 
   If one of the nodes is bad and the renew call is stuck the thread remains 
stuck indefinitely. 
   The thread should ideally timeout the renewToken and retry from the client's 
perspective.
   ```
   If this variable is set to `false`, once the HDFS NN Or Router Node fails, 
the RM refresh thread will be stuck.
   
   We can read YARN-9768 carefully. The author at that time should have set 
this variable for compatibility with previous unit tests. We can see the 
following code.
   
   
org.apache.hadoop.yarn.server.resourcemanager.security#DelegationTokenRenewer 
L939
   ``` 
    @VisibleForTesting
     public void setDelegationTokenRenewerPoolTracker(boolean flag) {
       delegationTokenRenewerPoolTrackerFlag = flag;
     }
   ```




> Make ResourceManager DelegationTokenRenewer timeout feature configurable
> ------------------------------------------------------------------------
>
>                 Key: YARN-11399
>                 URL: https://issues.apache.org/jira/browse/YARN-11399
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: resourcemanager
>    Affects Versions: 3.3.1, 3.3.2, 3.3.3, 3.3.4
>            Reporter: Daniel Ma
>            Priority: Minor
>              Labels: pull-request-available
>
> ResourceManager Delegation token renew timeout and retry mechanism is 
> introduced in Hadoop 3.3.1, 
> The switch of this feature is already there but it is true and can not be 
> disabled.
> So this patch introduce a configuration to make it configurable.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to