Nathan Gough created NIFI-10062:
-----------------------------------
Summary: Improve error handling and logging for NiFi's Zookeeper
client
Key: NIFI-10062
URL: https://issues.apache.org/jira/browse/NIFI-10062
Project: Apache NiFi
Issue Type: Improvement
Components: Core Framework
Reporter: Nathan Gough
Assignee: Nathan Gough
Currently Zookeeper/Curator client spam the NiFi app log when failing to
connect, and will infinitely retry to connect regardless of the issue (eg.
failed X509 authentication, unlikely to fix itself without intervention, will
infinitely attempt to reconnect).
Cases where I've seen Zookeeper client spam logs:
* Fails to connect at all (eg. misconfigured port or the Zookeeper server is
not running)
* Server is shut down
* Fails to authenticate
We should improve the logging being output by Zookeeper, and better handle
retry attempts.
The cluster itself also does not seem to work well with a failing Zookeeper
server, which may be something we can improve? Some ideas include transferring
state management to _local_ if we fail to reconnect to Zookeeper and the
ability to transfer state if Zookeeper is later reconnected. Reducing frequency
of reconnect attempts, reduce logging for each reconnect attempt/use background
attempts that are less visible without debugging.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
