Best practice for restarting stack?

Elastic Observability APM
1

I'm running the elasticstack on debian Stretch. (the only thing we use it for is monitoring some nodejs apps with apm-server, though we'd like to get into using it for more.) Currently version 6.7.0 of everything, but for many versions since i started using it about a year ago, restarting it (after a server reboot or whatever) has always been rather fragile.
Often 1 or more of the 4 services (ES, LS, Kibana, apm-server) does not come up automatically on boot.
I find that I usually need to start or restart 2 - 4 of them, sometimes 2 or 3 times in a few different combinations to finally get them all going and stable.

What's the current best practice for how it should work? What's the order of when each should be started? I haven't seen anything in the docs about this. thanks.

2

Welcome to the forums @steevhise. Given full control, services should be started in dependency order - ES first, Kibana and LS next, apm-server last to minimize any error logging noise.

That said, all of these services should retry connections so I'm surprised to hear that things don't stabilize on their own. Can you provide any additional information / config / logs that demonstrate the issues you're experiencing?

1 Like
3

This topic was automatically closed 20 days after the last reply. New replies are no longer allowed.