In reference to a previous post I made:

https://station.railway.com/questions/memory-per-replica-graphql-query-d094cd3a

Our team would really appreciate a feature that allows restarting a specific replica without restarting all replicas of a service.

Currently, restarting a deployment restarts all replicas, which can disrupt other in-progress work.

We occasionally need to restart individual replicas (e.g., due to high memory usage), so being able to target only the affected replica would be much more efficient—especially as we scale to many replicas.