I am quite curious.
Let’s say I have this Lagom service called “customer service”.
For some reason, customer service is not starting, let’s say it is unable to form a cluster and printing some cassandra related error messages.
And so we decided to drop the customer keyspace from the cassandra (assuming we have backup and this is happening during the announced downtime!), but we DO NOT delete the customer svc deployment from kubernetes. And we trigger the customer svc deployment.
In the above scenario, what do you think will happen and why? Do you think the customer svc will be able to deploy as normal with the customer keyspace being created also?
I’ve tried this in my testing environment for a couple of times and so far, I have not been able to start the service successfully.
It succeeded only after I dropped the keyspace AND the deployment also.
I am really wondering what causes this symptom.