Continually try to resolve a host that is unresolved

[slaupster]

This is a basic fix for the issue described in logstash-plugins/logstash-output-elasticsearch#131

I'm not sure this behaviour is desirable as a default or whether it would make sense to honor Java's networkaddress.cache.ttl and networkaddress.cache.negative.ttl or make ES specific properties.

[slaupster]

Not sure if there is a delay but I've signed the CLA and had what appears to be confirmation.

[clintongormley]

Hi @slaupster

Thanks for the PR (and I can see that you have signed the CLA, thanks). I wonder if this fix is sufficient. As I understand it, Elasticsearch will never re-resolve DNS entries, so if the IP for a node changes, the only way to make the transport client resolve the IP address again is to restart it. Wondering if we can improve this logic somehow.

@spinscale what do you think?

[spinscale]

@clint you're right, we already had a couple of discussions about this, see here. Especially, once an IP address is part of the cluster state (like as part of a client node), dynamic changes are more tricky, as we do not serialize the current ip address over the wire.

In the case of using a TransportClient though, I could imagine the above change.

[mirekingr]

This bug fix looks quite useful, any plans to release it?

[dakrone]

@spinscale revisiting this, do you still think it's a good idea? Does it need another review?

[spinscale]

@dakrone yes, I still think the change is useful for transport clients, however logstash is not affected by this due to using HTTP now. It does need another review, I did not follow any recent changes about networking code

[dakrone]

@spinscale any update on this?

[rjernst]

We should make a decision on this PR as it has been open for over 2 years. I have marked it for discussion.

[elasticmachine]

Since this is a community submitted pull request, a Jenkins build has not been kicked off automatically. Can an Elastic organization member please verify the contents of this patch and then kick off a build manually?

[dakrone]

I'm going to close this since there has been no update and Logstash no longer uses the Transport client (the underlying issues has been closed as well)