You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@pulsar.apache.org by GitBox <gi...@apache.org> on 2020/07/10 20:18:39 UTC

[GitHub] [pulsar] devinbost removed a comment on issue #7489: Every function/connector exception results in a new metric

devinbost removed a comment on issue #7489:
URL: https://github.com/apache/pulsar/issues/7489#issuecomment-656871407


   @matthewfollegot 
   Since we're using the Prometheus Counter's internal state to keep track of our exceptions, every time an exception is thrown, we're still adding the exception to the Counter, aren't we?
   e.g. https://github.com/apache/pulsar/blob/02bf9a0b770e53f5a6f3810e9602ccc9a4c05050/pulsar-functions/instance/src/main/java/org/apache/pulsar/functions/instance/stats/FunctionStatsManager.java#L293
   
   If I'm not mistaken, even if we omit the timestamp, wouldn't we still need to check if that exception has been thrown recently and filter new writes (unless we change how we're tracking that information) to prevent the new exception from perpetuating the same issue?
   If that's true and we need to filter new writes, then unless we periodically reset the exceptions (which we might already do), we could end up with new exceptions being suppressed inadvertently because they match old exceptions. There's a difference between an exception that occurs once a day and an exception that occurs with every method call, so we'd need a way to distinguish between them. 


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org