Kibana users can unintentionally hose the elasticsearch nodes

[dpb587]

Seems like users can unintentionally (or intentionally) kill an elasticsearch node with complex queries or large date ranges. To workaround this, we should try auto-appending the timeout option for the elasticsearch searches.

For now, it seems easiest to simply do that with nginx query re-writing and appending a timeout to all requests. Something like the following should help:

rewrite ^(.*)$ $1?timeout=15s break;

[dpb587]

@mrdavidlaing, have you been happy enough with the current 5s timeout for me to officially commit it here?

[mrdavidlaing]

@dpb587,

Yes, lets commit

        location ~ ^/[^\/\*]+/_search$ {
            rewrite ^(.*)$ $1?timeout=5s break;
            proxy_pass http://elasticsearch;
        }

[mrdavidlaing]

Lets increase this timeout to 15s

[dpb587]

The comment probably should have been a new issue, given requirements changing and this being an already closed issue. Committed.