Major refactoring and bug fixing #54
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
"grep ${CLUSTER_NAME}" also matches gitops-playground, gitops-playground2, etc.
Remove confirmation for init. It's not necessary. If the cluster exists, deletion has to be confirmed. Otherwise just create.
DEBUG=1 scripts/init-cluster.sh will print each command (set -x)
It's nice to save resources. But it also makes the setup more complicated. Is it worth this complexity or is it premature optimization? Remove for now!
Good practices: exit on first error, fail on unset vars and fail on first error in pipe. Turns out: SKIP_KUBECTL is never set. And never used: Deleted!
And hopefully more constant ones.
"/var/run/docker.sock permission denied" because queryDockerGroupOfJenkinsNode never had a change to get to actual docker group and so jenkins agents pod run with default group, which is not allowed to access docker socket.
Change up/downgrade query into an INFO. The users should decide for themselves and not be bugged by our potentially older k3ds version.
The default depends on k3d so it might change implicitly when upgrading k3d. We want deterministic behavior!
The name is misguided and confusing (at least to me) and the script can rather easily determine if it runs inside kubernetes for itself. This should make the whole process much simpler for the end user and requires less documentation.
It's much more convenient for end user than checking out the repo, installing all the tools and applying the script. Also, we hopefully get rid of the bash hell in the long run.
Also docs for destroy.sh.
We dropped the "k8s-" prefix for simplicity a while ago. Still, some resource inside the repo still used this prefix. Not anymore.
For example if --argocd is passed, install only argo example app jobs. Fixes #17
This should make the whole process much simpler for the end user and requires less documentation. The script can query the node's ip via kubectl, which is true for local k3d cluster at least
This makes testing in remote cluster, reviews much more easy. Only pushes git revision as tag, never latest!
Also use a newer cluster version.
…lure. There is some discussion going on about the kubectl run --serviceaccount flag. It is deprecated but still works: kubernetes/kubernetes#99732 So for now, use the "overrides" workaround. --restart=OnFailure - leaves the pod in status completed, if no errors occur. Otherwise it would restart and apply again, even on success. Why don't we use a job instead of a pod? It's likely that we would have to use YAML, because "kubectl create job" does not support --override. This makes handling it much more complicate for the user. For example, passing flags such as --remote, --argocd, etc. Also interactive mode (""-i --tty") is not possible. This makes it more complicate to wait for success and to view output. This would likely require a combination of "kubectl wait --for=condition=failure job/myjob" and "kubectl logs -f". So for now, "kubectl run" seems to be the easiest way here.
Less complexity and attack surface for the price of a little more redundancy (confirm functions is also defined in utils.sh).
Should simplify maintenance a bit.
Same meanings in apply.sh, destroy.sh and init-cluster.sh: DEBUG: print debug log (no spinner) TRACE: print each line (set -x)
Technically we could read the k8s version from a single source. But this is rather complicated to implement (in terraform, Dockerfile and shell script) and would probably cause much more effort than changing the redundant version in three places.
schnatterer
force-pushed
the
feature/refactoring
branch
2 times, most recently
from
f03d84f
to
aeade4e
Compare
schnatterer
force-pushed
the
feature/refactoring
branch
from
aeade4e
to
42069e0
Compare
schnatterer
force-pushed
the
feature/refactoring
branch
from
42069e0
to
615bbba
Compare
Signed-off-by: Marek Markiewka <marek.markiewka@cloudogu.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hopefully the playground is now much more intuitive to start from the docs.
And it should also work locally and remote again.