Closed dmartinol closed 2 weeks ago
@tchughesiv @tmihalac Note: validations will be added to the CRD types as the other options will be integrated.
requested a couple changes. also, i'm noticing we're missing new FeatureStore CR checks in featurestore_controller_test.go for these new status fields and defaults after reconcile
one more commit to add the missing checks
requested a couple changes. also, i'm noticing we're missing new FeatureStore CR checks in featurestore_controller_test.go for these new status fields and defaults after reconcile
one more commit to add the missing checks
Latest commit includes new controller tests. Please note that I had to change the resource name because some resources (svc?) are probably not deleted by previous tests, so it would fail due to an attempt to change ownership:
Object default/feast-services-offline is already owned by another Service controller feast-services-offline
I did not investigate further. I think we can handle this issue in a separate PR.
2024-11-07T17:30:30+01:00 INFO Successfully reconciled {"Service": "feast-services-offline", "operation": "updated"}
2024-11-07T17:30:30+01:00 ERROR Error deploying the FeatureStore client service {"error": "Object default/feast-services-offline is already owned by another Service controller feast-services-offline"}
github.com/feast-dev/feast/infra/feast-operator/internal/controller/services.(*FeastServices).setFeastServiceCondition
/Users/dmartino/projects/AI/feast/feast/infra/feast-operator/internal/controller/services/services.go:290
github.com/feast-dev/feast/infra/feast-operator/internal/controller/services.(*FeastServices).deployFeastServiceByType
/Users/dmartino/projects/AI/feast/feast/infra/feast-operator/internal/controller/services/services.go:95
github.com/feast-dev/feast/infra/feast-operator/internal/controller/services.(*FeastServices).Deploy
I did not investigate further. I think we can handle this issue in a separate PR.
it's not an "issue" per say ... it's got to do w/ another unit test from a prior run where i've changed object ownership so we can test that ownership conflicts/failures are working properly. that said, we prob should manually cleanup that object after the fact.
so, yea ... i guess it is an issue :)
What this PR does / why we need it:
Add ephemeral persistence configuration to the Feast services deployed with the operator. A sample manifest is also committed and reported here for your convenience:
Next PR will add persistent storage by using mounted volumed (Kubernetes PVC)
Which issue(s) this PR fixes:
Relates to https://github.com/feast-dev/feast/issues/4561