Reduce the operator bundle sizes/1 bundle for the product #1185
Conversation
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: dprince The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
Build failed (check pipeline). Post https://softwarefactory-project.io/zuul/t/rdoproject.org/buildset/b38fc0f49ef14e869d3e407b5e0dc55d ✔️ openstack-k8s-operators-content-provider SUCCESS in 2h 55m 23s |
|
Build failed (check pipeline). Post https://softwarefactory-project.io/zuul/t/rdoproject.org/buildset/6e091c5f71e84cc3addb4875b8075781 ✔️ openstack-k8s-operators-content-provider SUCCESS in 2h 34m 54s |
|
Build failed (check pipeline). Post https://softwarefactory-project.io/zuul/t/rdoproject.org/buildset/21a654ede32742b580a01cbe43a0e3ae ❌ openstack-k8s-operators-content-provider FAILURE in 8m 02s |
|
Build failed (check pipeline). Post https://softwarefactory-project.io/zuul/t/rdoproject.org/buildset/7828a64b64024513997c9b44d8e99b66 ❌ openstack-k8s-operators-content-provider FAILURE in 4m 18s |
|
Could we think of alternatives? One of I can think about is removing operators-specific data from oscp CRD leaving only top scope and nested top-scope, like Nova, spec data there. And hiding everything specific to services operators into data structs and go libraries. That would let us to rely on OLM, still, and avoid custom solutions around it. That is, by the price of no more separate deployments of services operators possible by their own CRs (but we don't support that anyway?) |
dprince
force-pushed
the
respin
branch
3 times, most recently
from
c30c403
to
1722fa0
Compare
|
Build failed (check pipeline). Post https://softwarefactory-project.io/zuul/t/rdoproject.org/buildset/d02a9d8fa37c440bb90bb4e77621079b ✔️ openstack-k8s-operators-content-provider SUCCESS in 1h 35m 57s |
|
I'm all for this. I think OLM is great, but it's really designed for simpler operators. We have a lot of complexity, so the more of that complexity that we can handle on our own, independent of OLM, the better I think the overall experience will be. For users, but also for developers. |
complexity is pain, indeed. |
operator-sdk create api --group operator --version v1beta1 --kind OpenStack Jira: OSPRH-11244
Jira: OSPRH-11244
Dockerfile is updated to include new 'operator' binary and bindata directory. Bindata will contain all the CRDs, RBAC, and deployment files needed to deploy all the operators. Add Makefile targets for bindata, and run-operator. Forklift bindata impl from old CNOSP compute_node_operator. Implement new controllers/operator/openstack_controller.go which process files in /bindata. Calling 'make bindata' during a Renovate sync should keep things in sync. Jira: OSPRH-11244
This gets passed to the the operator-manager and is used to run the controller-manager
|
@dprince: The following test failed, say
Full PR test history. Your PR dashboard. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. I understand the commands that are listed here. |
|
Build failed (check pipeline). Post https://softwarefactory-project.io/zuul/t/rdoproject.org/buildset/33706df2396a45c9be1c54c083349fd7 ✔️ openstack-k8s-operators-content-provider SUCCESS in 54m 43s |
|
PR needs rebase. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
This PR creates a new "initialization resource" called.... OpenStack (name TBD)
Dockerfile is updated to include new 'operator' binary and bindata directory. Bindata will contain all the CRDs, RBAC, and deployment files needed to deploy all the operators.
Add Makefile targets for bindata, and run-operator.
Forklift bindata impl from old CNOSP compute_node_operator.
Implement new controllers/operator/openstack_controller.go which process files in /bindata using bindata impl (uses k8s unstructured client).
Calling 'make bindata' during a Renovate sync should keep things in sync.
When compared to the old mechanism (i.e. using OLM) this new approach:
-only installs webhooks on the top level openstack-operator (may need them for infra/test operators too though)
-uses Certmanager directly for webhook certs (vs. having OLM create them)
-requires no other Openstack bundle dependencies (simpler)
-no bundle size limit as we store all things in the openstack-operator container
Jira: OSPRH-11244