feat: update the existing selectResources related func to support new API #390
Conversation
zhiying-lin
force-pushed
the
crp-res
branch
4 times, most recently
from
8f61e55
to
61381b8
Compare
| @@ -33,25 +34,56 @@ var ( | |||
| // There should be something wrong with the system and cannot be recovered by itself. | |||
| ErrUnexpectedBehavior = errors.New("unexpected behavior which cannot be handled by the controller") | |||
|
|
|||
| // ErrExpectedBehavior indicates the current situation is expected, which can be recovered by itself after retries. | |||
| ErrExpectedBehavior = errors.New("expected behavior which can be recovered by itself") | |||
There was a problem hiding this comment.
I am not 100% sure if I understand this part, and a little bit clarification would be greatly appreciated.
What would be if there's a resource version conflict (API error)? Is this an API error or an expected error? Similarly, what would be if an error is expected with sporadic occurrences and unexpected if repeated consistently?
There was a problem hiding this comment.
I think one way to interpret this is we can alert on too many ExpectedBehaviorError which will be treated the same as "UnExpectedError" while APIServerError will clearly say that there is something wrong with the api-server. We will look more into cluster status.
There was a problem hiding this comment.
For now there are api server errors, user errors and the rest are expected/unexpected logic errors. They should have no overlap.
agree with Ryan, the alert could be set based on the occurrences or status code returned by the API server.
| } | ||
| if !r.InformerManager.IsInformerSynced(gvr) { | ||
| return nil, fmt.Errorf("informer cache for %+v is not synced yet", restMapping.Resource) | ||
| return nil, controller.NewExpectedBehaviorError(fmt.Errorf("informer cache for %+v is not synced yet", restMapping.Resource)) |
There was a problem hiding this comment.
This is an example of my confusion specified in an earlier comment. It could be, at this point, that the cache is simply not working fast enough and will get sync'd soon; or it could also be that there is a connection/RBAC issue that will never recover without intervention and needs attention.
Description of your changes
Differentiate the error types
I have:
make reviewableto ensure this PR is ready for review.How has this code been tested
Need to add more UT in the future
Special notes for your reviewer