Is there a method we can use to effect ES configuration changes without having to restart ES?

(Kevin Soucy) #1

We've been tinkering with Watcher and Shield and its been a hassle to have to restart ES on all nodes to see the effects every time we make an ES cfg change. While we do disable shard allocation before shutdown and re-enable after cluster status is yellow again, it takes well over an hour for replica shard allocation to complete. Before we can test and attempt another config tweak.

Using ES 2.2, Watcher 2.2, Shield 2.2, etc.


(Mark Walkom) #2

Nope. Shield is dynamic, ES is not.

What are you tweaking, and why.

(Kevin Soucy) #3

Things like setting "watcher.shield.encrypt_sensitive_data" to mask the id/pwd required in watches (cuz of shield), "shield.audit.enabled" to see who's doing what, and setting up email ability for watches. We don't like to make all of these changes at once as it makes troubleshooting more difficult if ES has probs after making these changes.

What would be nice is a way to force ES to re-read its config file (or perhaps a separate cfg file) after making changes related to plugins. I understand we wouldn't want ES to re-read "core" type config detail in mid-stream as that could be catastrophic.


(Mark Walkom) #4

Ahh ok, I guess that makes sense.

Let me raise an internal issue (cause the Watcher+Shield repo's are private) and see what the team reckons!

(system) #5