pralabhkumar commented on pull request #33917:
URL: https://github.com/apache/spark/pull/33917#issuecomment-914408440


   @gaborgsomogyi  Thx for your valuable feedback . It really helps to 
understand things better. 
   
   If multiple history servers are running then they need different config 
because they handle different clusters, right? If 2 history servers are sitting 
in the same directory then they corrupt the data inside. A good example is when 
HS does compaction in streaming use-cases.
   
   => Not exactly , multiple history servers are running on same cluster , its 
for the purpose of not sending all the users request to one instance of  
history server (url). Ambari +HDP distribution  provides the way to launch 
multiple Spark history on different machines in same cluster . Note in this 
case spark.yarn.historyServer.address is still point to one machine only.   I 
think it helps when user query history server (load can be distributed across 
machines) .
   
   ==> Thx for providing information regarding headless keytab . In our case , 
i think  we must have host/node in the principal . However I am not much aware 
about the keytab without host or with host.    Since earlier we had keytab with 
host , I am trying to replicate the same behavior in new cluster (principal 
with host).  As stated , the same process we used for hiveserver2  and 
livyserver (princial +host) 
   
   
   Again Thx for your feedback  and time.  
   
   


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]



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

Reply via email to