Rolling update takes down all nodes

[ShuaiShao93]

Describe the bug

When rolling update happens and one node is down, the other node is stopped and fails the readiness probe:

To Reproduce

Steps to reproduce the behavior:

1. Use operator to run a cluster of 2 nodes
2. Kill the pod of one of the nodes
3. Check the other node
4. See it fails the readiness probe

Expected behavior
The other node shouldn't be impacted when one node is down

Screenshots

Version and environment information

• RabbitMQ Cluster Operator: v2.10.0

[lukebakken]

Use 3 nodes, not 2.

[ShuaiShao93]

Nice! Verified this works!

Is this documented anywhere?

[lukebakken]

https://www.google.com/search?q=site%3Arabbitmq.com+cluster+size