max-melentyev commented on issue #1869: URL: https://github.com/apache/cassandra-gocql-driver/issues/1869#issuecomment-2784293597
I'm not sure how this can be added to the existing code without adding a new observer because every attempt is observed independently and they are not aware if there will be other retries in the case of failure, i.e. for failed attempt should it report its duration as total, or should it wait for a retry attempt and add abserve total latency there. > That said, I'm curious what the total latency across all attempts is really measuring. We want to observe the real latency that client code sees. Gocql provides a speculative retries feature, so single Query.Iter() call can result in multiple queries that are observed independently. Consider this case: ``` Attempt 1 |-------- 150 ms ---------------------| Attempt 2 .. 10 ms ...|---- 10 ms ---| // speculative retry Query |---- 20 ms ---------------| ``` Observers are receiving 2 items: 1) error: nil, duration: 10ms (attempt 2) 2) error: context canceled, duration: >20ms (attempt 1) What client code actually observed: error: nil, duration: 20ms Or this case: ``` Attempt 1 |--- 10 ms ---| Fail Attempt 2 |---- 10 ms ---| // normal retry Query |---- 20 ms -----------------| ``` Observers are receiving 2 items: 1) error: failure, duration: 10ms (attempt 1) 2) error: nil, duration: 10ms (attempt 2) What client code actually observed: error: nil, duration: 20ms -- 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: commits-unsubscr...@cassandra.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
