Usage in usage view is not updated within 15 min

[jldec]

New usage takes much longer than 15min to appear in the usage view of a team with UBP billing configured.
This will be a problem for users, and also for us internally because it slows down our testing.

See internal discussion - this may be related to missing DB indexes.

[easyCZ]

This is a known issue at the moment. To fix this, we need the following changes:

• Wipe deleted Workspace (+ their WorkspaceInstances) at some point #11378
• [db] Add index to d_b_workspace_instance on startedTime & stoppingTime #12706

This has been currently blocked due to new indices causing downtime.

For now, we could change the wording to say (refreshes every 1h) which is roughly how often we do it now. See dashboard for details.

[easyCZ]

US migrations now applied. We still need to apply the stoppingTime migration in EU

[easyCZ]

To apply the EU indices, execute the following:

# your kubecontext name may vary
kubectx webapp-eu2

# Disable usage controller
# Remove the following config line to disable the controller:
#       "controllerSchedule": "15m",
kubectl edit configmap usage

# Delete existing pod to make it restart with the new config
kubectl delete pods -l component=usage

kubectl port-forward svc/cloudsqlproxy 3306:3306

# Connect to the DB on port 3306
# ...
# Run the following query:
CREATE INDEX IDX_workspace_instance__stopping_time ON d_b_workspace_instance (stoppingTime);

# Re-add controller schedule config 
# Add   "controllerSchedule": "15m",  to the config
kubectl edit configmap usage

# Delete existing pod to make it restart with the new config
kubectl delete pods -l component=usage

# Check there are now running pods  of `usage` again
kubectl get pod 

Expected run time is ~25 minutes.

[geropl]

Done with the steps above ✔️