✨ Use network resources managed outside cluster api #456
Conversation
|
/kind feature |
|
/kind bug |
Joseph94m
force-pushed
the
prevent-resource-deletion
branch
from
60f1429 to
bfe021a
Compare
|
This could be extended to include other resources, we included what we needed. |
Bastion being one of them. But still it it a step in the right direction and we thank you for your contribution. We still need to decide wether we only merge bugfixes (#451 being more a security fix than a feature) for now. We might wait until the next release for new features to be merged. |
jfbus
changed the title
|
Indeed, we did that because we did not want to change the resource spec. We have thought of another implementation that can fix this issue, but it requires a modification to the resource spec. Add a SkipReconcile Flag to every resource (default false so doesn't change the current behavior) This flag, combined with the resourceId field in the yaml spec generates the following use cases: We will propose another PR implementing the behavior soon. This should allow existing clusters to choose future reconciliation behavior simply by updating the yaml files defining the Cluster and applying it to the cluster. Legend:
|
|
I'm not a fan of adding schema changes if we can avoid them. I'm sure there is a way of doing it without impacting the schema using annotations. For each resource, adding an annotation would both set the resource id to use and declare it "not managed by cluster-api". |
Joseph94m
force-pushed
the
prevent-resource-deletion
branch
5 times, most recently
from
e52d7de to
b75c950
Compare
…ag instead of managed by capi
Joseph94m
force-pushed
the
prevent-resource-deletion
branch
from
b75c950 to
93c04ec
Compare
|
We are using this implementation in-house for the time being. |
What type of PR is this?
/kind feature
/kind bug
What this PR does / why we need it:
By default, cluster API outscale provider create all network object: NatGW/PublicIP/IS/Net/Subnet. In our use case, we import all these object in the cluster YAML definition using the “resourceId” field. Nevertheless, when deleting a cluster with resource imported using “resourceId” field, some resources are deleted and it should not be the case. We should fix the outscale provider to prevent that.
Code of the outscale API provider:GitHub - outscale/cluster-api-provider-outscale
Issue is already opened : [Feature]: Allow to reuse parts of existing infrastructure · Issue #381 · outscale/cluster-api-provider-outscale
Controller to fix to not clean resources when a resourceId is given as an input in YAML:
NatGW: cluster-api-provider-outscale/controllers/osccluster_natservice_controller.go at main · outscale/cluster-api-provider-outscale
PublicIP: cluster-api-provider-outscale/controllers/osccluster_publicip_controller.go at main · outscale/cluster-api-provider-outscale
InternetService: cluster-api-provider-outscale/controllers/osccluster_internetservice_controller.go at main · outscale/cluster-api-provider-outscale
Net: cluster-api-provider-outscale/controllers/osccluster_net_controller.go at main · outscale/cluster-api-provider-outscale
Subnet: cluster-api-provider-outscale/controllers/osccluster_subnet_controller.go at main · outscale/cluster-api-provider-outscale
Which issue(s) this PR fixes (optional, in
fixes #<issue number>(, fixes #<issue_number>, ...)format, will close the issue(s) when PR gets merged):Fixes #381
cf @alistarle