[GitHub] gaborgsomogyi opened a new pull request #23906: [SPARK-27002][SS] Get kafka delegation tokens right before consumer/producer created

[GitBox <gi...@apache.org>]

gaborgsomogyi opened a new pull request #23906: [SPARK-27002][SS] Get kafka delegation tokens right before consumer/producer created
URL: https://github.com/apache/spark/pull/23906
 
 
   ## What changes were proposed in this pull request?
   
   Spark not always picking up the latest Kafka delegation tokens even if a new one properly obtained.
   In the PR I'm setting delegation tokens right before `KafkaConsumer` and `KafkaProducer` creation to be on the safe side.
   
   ## How was this patch tested?
   
   Long running Kafka to Kafka tests on 4 node cluster with randomly thrown artificial exceptions.
   
   Test scenario:
   * 4 node cluster
   * Yarn
   * Kafka broker version 2.1.0
   * security.protocol = SASL_SSL
   * sasl.mechanism = SCRAM-SHA-512
   
   Kafka broker settings:
   * delegation.token.expiry.time.ms=600000 (10 min)
   * delegation.token.max.lifetime.ms=1200000 (20 min)
   * delegation.token.expiry.check.interval.ms=300000 (5 min)
   
   After each 7.5 minutes new delegation token obtained from Kafka broker (10 min * 0.75).
   But when token expired after 10 minutes (Spark obtains new one the doesn't renew the old), the brokers expiring thread comes after each 5 minutes (invalidates expired tokens) and artificial exception has been thrown inside the Spark application (such case Spark closes connection), then the latest delegation token not always picked up.
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on 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


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org