I’m researching possibility to use Keepalived as a load-balancer for Graylog.
So I will need virtual server(float ip) which will receive all requests and then they will be distributed to real_servers (Graylog nodes)
It seems I will need to create virtual server for each Input (I have syslog and filebeat Inputs)
My doubt is about Input monitoring/health-check.
Is there some native out-of-the box way (API?) to monitor specific Input status on specific Graylog node? (Of course I could use manual parsing of /api/cluster/inputstates, but this is ugly way. And boring)
Node itself can be alive, but specific Input can be dead, so balancing to such Input is meaningless and I’d like to avoid such situation.
From what I see this question was partially touched in
and
but without visible result(and moreover 3 yrs passed, so situation could change)