ChrisSamo632 commented on pull request #5299:
URL: https://github.com/apache/nifi/pull/5299#issuecomment-912966926


   On the question from @MikeThomsen and response from @eylonronen, I'd say the 
main thing that's missing in this implementation is any error handling within 
the simple round-robin implementation in this PR.
   
   The Elasticsearch Client will (optionally) sniff the Elasticsearch cluster 
for all available hosts/URLs at startup/on connection - this implementation 
here relies upon the NiFi user configuring all the current Elasticsearch hosts 
and keeping them updated if the cluster membership changes.
   
   Also, the Elasticsearch client will temporarily stop trying to connect to a 
host it knows is not responding - this PR will instead result in operations 
failing for any hosts that go offline (for example) and the NiFi user will need 
to cater for that, e.g. with a Retry loop on failure
   
   That's not to say that this approach isn't a potentially useful option for 
Elasticsearch cluster users, it certainly avoids the need for them to introduce 
a load balancer in front of Elasticsearch if they don't already have one, and 
it will mean that if the single host they were able to connect with previously 
were to go offline, then others being configured in the processors means 
operations would work at least some of the time.
   
   Generally I'd recommend people using the Elasticsearch Client-based 
processors where possible, but these OkHttp based processors have some 
functionality that the newer ES Client ones don't currently, so they're still 
relevant and this update feels like it's probably going to be useful.


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to