Open simonostendorf opened 3 weeks ago
Yes there is. The status in Kubernetes is something that can be lost and the reproduced. What you see in spec.status cannot be reproduced in some parts and therefore cannot be put into the Kubernetes status, at least parts.
This is why we had the options of putting it under spec (what we did) or in the annotations in some compact json string. We decided to put it in the spec because it is more stable than the annotations that have a size limit.
If you have another perspective how to achieve this (to not lose this data because we cannot just re-create it in any reconcile loop), I'm open for it!
Thank you for your quick and detailed reply @janiskemper!
I think the decision is understandable and I have no further perspective or questions.
The whole decision (or this issue) could be written/referenced as code comment to avoid future questions and to document the decision process.
makes sense! will do!
/kind bug
What steps did you take and what happened:
If you describe a
HetznerBareMetalHost
you can see, that the.status
fields are placed under.spec.status
.What did you expect to happen:
Placing status as
.status
and not as.spec.status
is the "normal" way. Are there any reasons why this is placed under.spec.status
here?Anything else you would like to add:
You can see the same structure in the code here.
Environment:
v1.0.0-beta.34
v1.30.0