Skip to content

machine424/cicdbox

Repository files navigation

cicdbox

CI/CD magic (or hacks) without spaghetti shell code on your end.

Intendend for Gitlab CI/CD (for now), and hosted here (for now) for accessibility reasons. You can find a Gitlab mirror here.

Available tools

collapse-helm-diff

No one would check a 2000 lines diff, no one would even browse them!

This tool makes helm diff more readable by collapsing diff lines which provide no real information, they could be: docker images's tags relying on the git sha, checksums etc.

It takes advantage of Gitlab collapsed job log sections.

Example:

Diff without collapse-helm-diff:

without_collapse

After hiding image: docker.foo.fr... diffs:

  # Don't forget to set -o pipefail (in bash e.g.) to pervent helm diff errors from being masked.
  - helm diff | ./cicdbox collapse-helm-diff '^\s+-?image: docker\.foo\.fr' 

The diff will look like:

with_collapse

One can then expand sections if needed:

expand_collapse

This works well with --context set to 1 for helm diff.

./cicdbox collapse-helm-diff regex1 regex2 ... (Regexes to match diffs to hide, RE2 syntax as described here)

Supposes echo is available.

annotate-helmfile-releases

Adds GITLAB_USER_LOGIN and CI_PIPELINE_URL as annotations to the sh.helm.release.v1.RELEASE.vN secrets corresponding to releases you sync with helmfile.

Example:

  - helmfile --selector foo=bar sync
  - ./cicdbox annotate-helmfile-releases --selector foo=bar

will add cicdbox/* annotations like:

kind: Secret
apiVersion: v1
metadata:
  name: sh.helm.release.v1.SELECTED-RELEASE.v1
  ...
  annotations:
    cicdbox/releaser: username
    cicdbox/gitlab-ci-pipeline-url: https://gitlab.foo.bar/foo/bar/foo/-/pipelines/173171
...

This can help understanding how a release ended up there, how to fix it, whom to blame :) etc. (one can expose these metadata as Prometheus metrics and create alerts about old/abndonned releases.)

./cicdbox annotate-helmfile-releases --help for more details.

Supposes helmfile and kubectl are available.

listen-helm-debug

Get Helm relevant live --debug logs and hide the rest in a collapsed section, these logs can tell you for which resources Helm is waiting and provide users a fast feedback. No more Help! my deploy job is stuck.

Example:

  - set -o pipefail
  - helmfile --debug sync |& ./cicdbox listen-helm-debug

The output would look like:

listen_helm_output

About

CI/CD magic (or hacks) without spaghetti shell code on your end.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages