As cluster-admin\
I want to run the webhook server in its own pod \
So that** that crashloops of the operator doesn't affect the ability to edit instances.
Context
Currently (v0.1.1) both operator and webhook are in the same container and pod.
If the controller encounters an unrecoverable error (such as a panic) then the chances are high that the container ends up in a crashloop. This means that the webhook server also gets constantly restarted.
Consider this scenario:
Instance object is in a broken state
Operator tries to reconcile instance, but panics and crashloops
A user tries to fix the instance by editing the broken spec -> user can't do this since kubernetes can't reach the webhook server
Out of Scope
No response
Further links
No response
Acceptance Criteria
No response
Implementation Ideas
Run the webhook part as a separate CLI subcommand in its own deployment
Summary
As cluster-admin\ I want to run the webhook server in its own pod \ So that** that crashloops of the operator doesn't affect the ability to edit instances.
Context
Currently (
v0.1.1
) both operator and webhook are in the same container and pod. If the controller encounters an unrecoverable error (such as a panic) then the chances are high that the container ends up in a crashloop. This means that the webhook server also gets constantly restarted.Consider this scenario:
Out of Scope
No response
Further links
No response
Acceptance Criteria
No response
Implementation Ideas