Hey there,
we did some analysis on our log files to get a better insight of the cluster whats happening. We found out that the log message "WARN org.apache.zookeeper.Login: TGT renewal thread has been interrupted and will exit." is written to the log over 1000 times per minute on a single RS. We've got these messages on all worker nodes in the cluster, but with a much lower count (10 - 100). Only when the usecase restarts the applications - RS changes until the next restart. Does anyone have an idea by what this excessive logging is triggered?
Distribution over the workernodes over a 24h timeframe:
first chart: counting all messages of loglevel warn per Host;
second chart: count all messages of loglevel warn grouped by the java class
Regards,
Timo
