For further context, as this is a huge scope. I think at a first step only 1 or 2 of those need to be automated. I think it is more important that there is a script that identifies the "category" and provides even just a link to docs than 100% automation.

E.g.:

db-depool masterdb1001.eqiad.wmnet
ABORTED: Depool not done- this is a primary mediawiki server so use this procedure instead: <link to master switchover>

The idea is, as more categories and automations are created, having an easy entry point for non-DBAs.

In a ideal scenario I'd love this to be:
depooling_cookbook $HOST and all the logic and checks are done underneath. So the operator wouldn't need to care whether it is safe to depool a host or not (or a section), if it belongs to MW or misc etc.

depooling_cookbook $HOST and all the logic and checks are done underneath. So the operator wouldn't need to care whether it is safe to depool a host or not (or a section), if it belongs to MW or misc etc.

+1

The start of this is the recently added sre.mysql.pool and sre.mysql.depool cookbooks for core DBs that could be expanded to cover the additional use cases.

Btw, I am going to use the same idea for backups, with a single command to $ recover db1001 and abstract the different methods into an easy to use script wrapper (but that is out of scope of this).

Change #1215575 had a related patch set uploaded (by Federico Ceratto; author: Federico Ceratto):

[operations/cookbooks@master] sre.mysql.pool: [de]pool various section kinds

https://gerrit.wikimedia.org/r/1215575