Add Podman Compatibility and Enhance Kind cluster setup #11086
Conversation
Signed-off-by: see-quick <[email protected]>
|
Why most of the already existing commands are now started via |
Mainly because we have to run kind of as |
Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
…on required. Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
|
/azp run acceptance |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
| } | ||
|
|
||
| function label_node { | ||
| # It should work for all clusters | ||
| for nodeName in $(kubectl get nodes -o custom-columns=:.metadata.name --no-headers); | ||
| for nodeName in $($SUDO kubectl get nodes -o custom-columns=:.metadata.name --no-headers); |
There was a problem hiding this comment.
should $SUDO be used everywhere?
There was a problem hiding this comment.
Yeah, because if do not execute with sudo you will not see kind cluster as we previously started kind as sudo
|
/packit test --labels acceptance |
| // Kaniko pushes the image to the local registry, but Podman stores it in an internal storage | ||
| // that is not directly accessible to Kind. Podman must first pull the image from the registry | ||
| // so that it is available in the Podman daemon's local storage before "kind load" can work. | ||
| Exec.exec("sudo", "podman", "pull", image); |
There was a problem hiding this comment.
what if I will use kind with docker?
There was a problem hiding this comment.
I think I should add some wrapper to check your container runtime and base on that just use podman/docker.
Signed-off-by: see-quick <[email protected]>
|
/packit test --labels acceptance |
Signed-off-by: see-quick <[email protected]>
|
/packit test --labels acceptance |
Signed-off-by: see-quick <[email protected]>
systemtest/src/main/java/io/strimzi/systemtest/utils/specific/ContainerRuntimeUtils.java
Outdated
Show resolved
Hide resolved
| Exec.exec("sudo", ContainerRuntimeUtils.getRuntime(), "pull", image); | ||
| } | ||
| // if container-runtime is docker we do not need to pull | ||
| Exec.exec("sudo", "kind", "load", "docker-image", image, "--name", "kind-cluster"); |
There was a problem hiding this comment.
I guess we will have to think about a different way how to load it, or take the name of the Kind cluster first. Because not all of the Kind clusters are named kind-cluster.
There was a problem hiding this comment.
wouldn't be easier just skip this test with podman runtime + kind ?
There was a problem hiding this comment.
I guess that's not desired, or?
I mean, yeah for AZPs we would run it anyway, as we are running the tests on Minikube.
For TF, I think that we are using podman there (if not, then fine).
The only issue is on Jenkins, where we are running the nightly builds and I guess the skip would be bad.
Another thing I was thinking was to build the image before running the tests, similarly as we do it with Connect image. But that would add more conditions and setup just for running the one test 🤷 .
There was a problem hiding this comment.
I had discussion with @Frawless
I agree that having another kind of conditions around that just for one test doesn't make sense.
For AZP, it will run without issues. For other like OCP as well. So let's skip the test for kind + podman @see-quick
There was a problem hiding this comment.
Thanks, I will update this accordingly.
systemtest/src/main/java/io/strimzi/systemtest/utils/specific/ContainerRuntimeUtils.java
Outdated
Show resolved
Hide resolved
Signed-off-by: see-quick <[email protected]>
Signed-off-by: see-quick <[email protected]>
|
/packit test --labels acceptance |
Signed-off-by: see-quick <[email protected]>
|
/packit test --labels acceptance |
Signed-off-by: see-quick <[email protected]>
|
/packit test --labels acceptance |
| @@ -118,7 +118,7 @@ prepare: | |||
| # Setup local registry for building images and connect image | |||
| export DOCKER_REGISTRY="${HOST}:5001" | |||
| export DOCKER_ORG="strimzi" | |||
| export DOCKER_TAG="test" | |||
| export DOCKER_TAG="latest" | |||
There was a problem hiding this comment.
Because with test KafkaBridge tests were failing. With the
sed -i "s#quay.io/strimzi/kafka#${DOCKER_REGISTRY}/${DOCKER_ORG}/kafka#g" packaging/install/cluster-operator/*-Deployment-strimzi-cluster-operator.yaml
we were also modifying the content of the KafkaBridge image instead of just replacing Kafka images. So in other words when I fixed the image issue then
quay.io/strimzi/kafka-bridge:testwas another problem. So I have changed the test docker tag to the latest and we would use:
- Kafka/operator etc. images with replaced DOCKER_REGISTRY (e.g.,
10.2.4.1:5001/strimzi/kafka:latest-kafka-3.9.0) - but KafkaBridge will be used the latest (e.g.,
quay.io/strimzi/kafka-bridge:latest)
There was a problem hiding this comment.
wouldn't be enought just specify env var for bridge image?
Type of change
Description
This PR introduces Podman compatibility to the KIND cluster setup script. It seamlessly allows users to use Docker or Podman as their container runtime. The changes ensure proper configuration and support for Kubernetes cluster provisioning.
Checklist