Open raphaeltm opened 1 month ago
This could be solved by adding an intermediate status/state when CD pulumi up
is done, something like UPDATE_DONE
, and have CD publish that state update to Loki, similar to Fabric. Until we have a real pub-sub, it would have to fake the job=fabric
so it gets properly picked up by the subscriptions.
Let's use SNS to receive the events from AWS and translate these in Fabric to high level HEALTHY or FAILED events.
Use CloudWatch alarm, publish to SNS (http), handle in Fabric and translate to high level state. This will work for future managed services as well.
Managed Redis stays stuck with status
UPDATE_QUEUED