Fixes:

• Add deploy role → moved to Add ansible files for deployment #33
• Add security group for Obsinn during configuration
• Add table partitions during configuration
• Fix postgres data directory (closes Make Postgres use the correct partition for data storage #36)
• Make style conform to ansible-lint
• Figure out how to deal with changed_when and ignore_errors
• Check if lard service should be stopped/restarted during switchover
• Add automatic teardown of instances and volumes
• Upgrade to postgres 17 and Ubuntu Noble
• Add a way to handle ssh on teardown? (remove old VMs from known_hosts)
• Figure out why lard_ingestion can't connect to postgres on the standby node
  "failed to lookup address information: Temporary failure in name resolution"
• Add partitions for flags.kvdata (closes Partition flags.kvdata #58)

I think I want to merge this now, but there are still a couple issues (that I will open):

1. For the last point the problem is that we can't connect to stinfosys, because the standby is not routed through the ipalias interface, therefore it drops most of the external connections. Probably we want to have two floating IPs and switch them if the primary goes down
2. For the deploy, switchover, and failover roles we need to specify both the primary and the standby, which is bothersome and prone to error. This could be avoided by having a first tasks checking that the attached ip alias is the correct one (or not, for the replica(s)). Or check with repmgr like I did for thebigip.
3. The failover playbook probably requires more testing, especially now that we have the ingestor running.

Edit: dropped kdvh and kvalobs labels, they should be part of the migration_fixes branch