Don't allow two DRPCs with conflicting selectors in the same namespace #1081
Conversation
raghavendra-talur
force-pushed
the
rtalur-drpc-conflicting-pvc-labelselector
branch
from
bfc6742
to
30ce48c
Compare
raghavendra-talur
force-pushed
the
rtalur-drpc-conflicting-pvc-labelselector
branch
3 times, most recently
from
07f4b2e
to
cb4a719
Compare
raghavendra-talur
changed the title
|
@raghavendra-talur can you describe the problem you are trying to solve? Is there a bz that this PR is fixing? |
| rmn "github.com/ramendr/ramen/api/v1alpha1" | ||
| plrv1 "github.com/stolostron/multicloud-operators-placementrule/pkg/apis/apps/v1" | ||
| ) | ||
|
|
There was a problem hiding this comment.
Is this test related to overlapping selector? how?
| shouldMatch: false, | ||
| }, | ||
| { | ||
| name: "Test with cache and not cache", |
There was a problem hiding this comment.
using "cache" is confusing, why not "value"?
| }, | ||
| shouldMatch: true, | ||
| }, | ||
| } |
There was a problem hiding this comment.
We can have obviously matching selector (e.g. key = value, key = value), but we can have some like:
drpc1: key1 = value1
drpc2: key2 = value2
This looks like valid selectors, but the user may add both labels to some PVC and this check cannot detect the issue.
What can work is to add an annotation to the PVC when we protect it with the name of the drpc (like placement). If a PVC is already protected by another drpc the new drpc should fail or report an error that some PVCs could not be handled due to conflict.
Signed-off-by: Raghavendra Talur <raghavendra.talur@gmail.com>
Signed-off-by: Raghavendra Talur <raghavendra.talur@gmail.com>
Signed-off-by: Raghavendra Talur <raghavendra.talur@gmail.com>
Signed-off-by: Raghavendra Talur <raghavendra.talur@gmail.com>
raghavendra-talur
force-pushed
the
rtalur-drpc-conflicting-pvc-labelselector
branch
from
cb4a719
to
9b6a2fa
Compare
raghavendra-talur
deleted the
rtalur-drpc-conflicting-pvc-labelselector
branch
No description provided.