Upgrading kuma causes all MeshGateway pods to be restarted

[bartsmykla]

Description

Upgrading Kuma causes change in the image tag for gateway pod. It means it will always be restarted. It might be not ideal as as an operator you might want to explicitly decide when to update/restart gateway pods. Few quick ideas:

• Maybe we should make sure that we configure rollout unavailability
• Maybe we could put in reconciller option to not auto update tag of gateway pods
• The most conservative option would be to add image version toMeshGatewayInstance
• We should solve this for Gateway API and non-Gateway API (when user creates MeshGatewayInstance manually)

[jakubdyszkiewicz]

Triage: if you want to control this - run your own Deployment. Let's make sure we have good defaults (max unavailable instances etc.)

[github-actions[bot]]

This issue was inactive for 90 days. It will be reviewed in the next triage meeting and might be closed. If you think this issue is still relevant, please comment on it or attend the next triage meeting.