caican00 commented on PR #5443:
URL: https://github.com/apache/gravitino/pull/5443#issuecomment-2458665058

   > > > Hi @caican00 will there be any potential UGI issue when we reuse the 
client instance? CC @yuqi1129 @jerqi to also take a look.
   > > 
   > > 
   > > When Impersonation is disabled, we use a super user to access hms, 
reusing the clients is reasonable.
   > > When Impersonation is enabled , we get a HiveClientPool instance from 
the clientPoolCache through a `Key` instance. If the ugi or user_name of the 
current thread is the same, the `Key` instance should be the same and the 
connection in the HiveClientPool should be taken instead of recreating a new 
HiveClientPool.
   > > Now each request will create a HiveClientPool instance, the 
HiveClientPool is meaningless.
   > > Is my understanding correct? If there is any problem, please help point 
out. Thank you very much cc @yuqi1129
   > 
   > Another user has reported that `Impersonation` does not work when using 
Spark to access Iceberg with Kerberoized Hive due to the reuse to pool, I have 
not confirmed it as the environment is not accessible to me. they claim that 
"Abandon the cache and just create a new pool works fine" and I'm not sure 
whether it's related to this PR.
   
   > > > they claim that "Abandon the cache and just create a new pool works 
fine" and I'm not sure whether it's related to this PR
   
   Hi @yuqi1129 May i ask how the user abandoned the cache?


-- 
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]

Reply via email to