Update Ideal State API creating new ideal state #3077
Description
Describe the bug
When the resource gets deleted midway by customer scripts/externally after (!_zkClient.exists(zkPath)) check and another process is trying to update the deleted resource's idealstate. The resource gets created with just the updated parameters in the API payload.
This can lead to rebalance failure as well when the updated parameters does not contain the num of partitions. Rebalancer considers num_partition for this resource as -1 leading to rebalance failure.
To Reproduce
- Start helix rest in debug mode. Trigger update ideal state API for a resource.
- After (!_zkClient.exists(zkPath)) check point, manually delete the resource.
- New idealstate will be created using the update ideal state API.
Expected behavior
- Update Ideal State API should not create new ideal state resource.
Additional context
Add any other context about the problem here.