Internal server error due to '_msearch' endpoint

[hazim]

Hi,

I have a logstash-* index, and when I open kibana discover or dashboards I USUALLY see an internal server error on the UI.

Upon looking at the browser console or network, I see that this occurs when the call to '_msearch' endpoint fails with 500.

/kibana/elasticsearch/_msearch?timeout=0&ignore_unavailable=true&preference=1523626984691

I get the following response for the call above:

net/http: timeout awaiting response headers

Looking at other related issues people were facing I think it might be due to time out? I'm not sure though. If yes How do I increase this timeout? Is that the solution?

I've tried hitting the _msearch endpoint directly on elasticsearch with the same request payload manually after this happens and that request is successful, I don't get 500.

Moreover, I get this error most of the time but after several reloads and once I feel the data has been loaded I stop seeing this error for a while.

Versions:
Elasticsearch: 2.3.4
Kibana: 4.5.4

[hazim]

Ok so I am running kibana behind a proxy, and that proxy timed out on the _msearch endpoint.

So the response net/http: timeout awaiting response headers was from the proxy and not from kibana or ES.

Increasing the timeout in proxy solved the issue. It wasn't related to Kibana or Elasticsearch but updating here if any one else falls in the trap.