If a fully deployed site/namespace is deleted directly and the database record still exists, the management controller errors and crashes due to the manage accesspoint DNS being unresolvable. This becomes a larger problem when the colocated site sync module tries to redeploy the site due to the namespace not existing. The router module starts to error and stops the sync from properly deploying.
To replicate:
- Fully deploy an sk2 site (create manage AP and fully bootstrap to 'deployed' state)
- delete the namespace the site is deployed to in kubernetes
- observe the logs from the management controller
This needs to be handled more gracefully so the router module doesn't keep trying to reach a deleted accesspoint, blocking the redeployment of that same accesspoint.
If a fully deployed site/namespace is deleted directly and the database record still exists, the management controller errors and crashes due to the manage accesspoint DNS being unresolvable. This becomes a larger problem when the colocated site sync module tries to redeploy the site due to the namespace not existing. The router module starts to error and stops the sync from properly deploying.
To replicate:
This needs to be handled more gracefully so the router module doesn't keep trying to reach a deleted accesspoint, blocking the redeployment of that same accesspoint.