When we retire a Duffy pool, the information about that pool persists in the database. This means Zabbix never stops monitoring old pools, because we connect to the DB to retrieve pool info.
We should write a small SQL script we can run against the DB when retiring a pool to clean up the data. Chatting with @nphilipp that would seem to be simply deleting from the sessions_nodes join-table, and then from sessions and nodes, all where pool = 'thing-we-are-removing.
sessions_nodes
sessions
nodes
pool = 'thing-we-are-removing
We can add this to the Duffy Ansible role and have it available on-disk for use.
Metadata Update from @gwmngilfen: - Issue assigned to gwmngilfen
Metadata Update from @arrfab: - Issue tagged with: centos-ci-infra
https://github.com/CentOS/ansible-role-duffy/pull/17 creates and deploys the required script. I can't see a good way to automate it - we'd have to:
That feels brittle, so I'm leaving it as manual, since we'd manually have to delete entries from 70_nodepools anyway...
https://github.com/CentOS/ansible-role-duffy/pull/17 was merged, and docs are in progress at https://git.centos.org/centos/centos-infra-docs/pull-request/12 - closing as done.
Metadata Update from @gwmngilfen: - Issue close_status updated to: Fixed - Issue status updated to: Closed (was: Open)
Log in to comment on this ticket.