flink-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From jaxbihani <jagad...@helpshift.com>
Subject Re: Problem with CEPPatternOperator when taskmanager is killed
Date Tue, 20 Sep 2016 04:54:35 GMT

@Frank : Thanks.
I verified changing that and it works fine.  (I still need to copy the user
jar to the lib dir which ideally shouldn't be required but I think that is
because of userCodeClassLoader() related problem. Will look into that)
About implementation I feel instead of initializing priority queue with 1
when numberPriorityQueueEntries=0. We should have a  condition check that if
numberPriorityQueueEntries > 1 then only perform the next processing
(creating priority queue object and deserialization calls in a loop) as
numberPriorityQueueEntries=0 means nothing needs to be done in restoreState.

I can add a patch for it but as Frank found the resolution it is his call to

View this message in context: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Problem-with-CEPPatternOperator-when-taskmanager-is-killed-tp9024p9074.html
Sent from the Apache Flink User Mailing List archive. mailing list archive at Nabble.com.

View raw message