GitOps Tools¶
⚠️ This could be outdated! ⚠️
Last update: Mar 14, 2019
Weave Flux¶
- One Flux Controller per repo
- One repo could be scanned from multiple flux controllern
- It's possible to handle multiple envs in one repo (FAQ)
- Removing resources are still an experimental feature which has to be enabled manually
- Helm Charts will not be removed if the
HelmReleaseCRD is deleted from the repo. TheHelmReleaseCRD-Object has be removed manually in the cluster. (FAQ)
- Can be extended via operators (e.g. Helm Operator)
- Is working with real Helm Releases
- Use CRD for defining a
HelmRelease - Docker Registry scanning for new Images
- Scanning via Regex (e.g.
master-*,dev-*)
- Scanning via Regex (e.g.
- Namespace whitelisting is possible (FAQ)
- Events could be thrown to slack and webhooks (GitHub Repo)
FluxTag in Repository to mark the latest deployed commit- NEGATIVE Needs write access on the repo
- No sync back from cluster to repo
- Integrates with Weave Cloud
- Flux Operator to easily install multiple Flux instances
argo¶
- Running as a k8s controller
- Able to deploy to multiple clusters
- Automated or manual syncing of applications to its desired state
- Web UI which provides real-time view of application activity
- CLI for automation and CI integration
- Webhook integration
- Prometheus metrics
- SSO Integration
- NEGATIVE deployes helm charts as normal k8s resources
- -> No helm rollback mechanism or other helm benefits
- NEGATIVE deployment via helm chart possible but not very well documented
autoapply¶
- Deploys K8s-Resourcen and Helm Charts
- Simply pull repo and run
kubectl applyor what every you want - Runs in the cluster
GitLab mit der K8s Integration¶
- NEGATIVE Needs own installation of Tiller
- NEGATIVE The Prometheus installation doesn't include the Metric-Server
- No really benefits
Ship¶
- Deploys Helm Charts
- Needs for every Helm Chart an seperated process
- No centered instance -- is living in the ci pipeline