SHC load balancer health monitoring, manual detention

[brettwilliams]

So we put a couple peers into manual detention to let them "cool off" before doing some maintenance.  Problem is, people were still getting sent to those members despite manual detention, and as a result, couldn't search.

Is there a single unauthenticated call I can make via REST from my load balancer to remove manual detention members from the pool?  If so, what is it?  Probably gonna dig through the REST API tonight and (hopefully) find it.  Someone...  beat me to it!

[isoutamo]

Hi

how you have configured your LB to detect that node is down? Here is example of status check from LB side to notice that SH node is down (and probably detention is working same way?).

https://community.splunk.com/t5/Deployment-Architecture/Looking-for-a-URL-API-to-configure-on-load-b...

r. Ismo

[brettwilliams]

Detention doesn't take down the web interface though, so a generic HTTPS monitor won't reflect manual detention.

I found the off/on answer for manual detention at https://<hostname or ip>:8089/services/shcluster/config. 

On an F5 LTM, HTTPS monitors support regex receive strings, so coupled with a user and role with capability list_search_head_clustering and a long-lived auth token, I think it can be done.

[isoutamo]

Nice, if you get it to work with F5 I would like to know the exact solution!
r. Ismo