Elastic search behind nginx proxy - Node updates

Elastic Stack Elasticsearch
1

I have a setup such that I'm proxying requests over a custom domain that points to an nginx server to Elasticsearch cluster.

I have noticed that nginx is caching the data node ips as per the ttl. Now, when a node goes down or when blue green deployments happen, until the ttl is expired nginx is routing the requests to nodes that aren't active anymore.

Has anyone else faced this issue ? How do we solve this ?

2

I think this is more a question of your Nginx config than anything particularly to do with Elasticsearch. See in particular its documentation on HTTP health checks and TCP health checks. If you need more help, I'd recommend finding a forum dedicated to Nginx questions rather than this Elasticsearch-focussed one.