dang-stripe commented on issue #15057:
URL: https://github.com/apache/pinot/issues/15057#issuecomment-3142179852

   > Query failures [without 
metrics](https://github.com/apache/pinot/blob/d691befa84ea425f0190bb7322545146dbf8c98b/pinot-broker/src/main/java/org/apache/pinot/broker/requesthandler/MultiStageBrokerRequestHandler.java#L495-L500)
   
   this seems to have mostly been covered in 
https://github.com/apache/pinot/pull/15277 and some of the other work in 
https://github.com/apache/pinot/issues/14950. @Jackie-Jiang confirmed that the 
catch all query execution block emits a processing exception metric in the 
onFailedRequest hook: 
https://github.com/apache/pinot/blob/62c75763944aeacfce1bd8eef7ce9a2087c5d612/pinot-broker/src/main/java/org/apache/pinot/broker/requesthandler/MultiStageBrokerRequestHandler.java#L200
 which should cover the example.
   
   > No logs or stats in response metadata that can be used to identify slow 
instances in any stage of executing a query. Or no way of correlating broker 
request IDs to the logs or stats.
   
   this is now covered by https://github.com/apache/pinot/pull/15180
   
   > Timeouts are difficult to diagnose without taking an approach such as 
increasing the timeout, rerunning, and then profiling the query but once again 
that does not enable retrospective debugging
   
   i think @Jackie-Jiang and @gortiz have mentioned that the stage stats are 
the primary debugging mechanism for this. we're working on archiving them now 
after query processing to enable retrospective debugging.
   


-- 
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