Thanos Helm Chart¶
A community-maintained build of the Thanos Helm chart, hosted
in the ALASCA GitLab and published as alasca-thanos. It is rebuilt from the
upstream Bitnami Thanos chart sources but wired to upstream, Apache 2.0 container images,
so it replaces the Bitnami chart that is no longer freely usable. It is the shared,
lasting solution for ALASCA projects and others that depend on Thanos for long-term
metrics storage.
| Source project | alasca.cloud/cai/charts/thanos |
| Published chart | alasca-thanos via https://charts.alasca.cloud/thanos |
| License | Apache 2.0 |
| Built from | Bitnami Thanos chart (Apache 2.0) |
| Images used | docker.io/thanosio/thanos, docker.io/bitnami/os-shell |
| Created | 22 May 2026 |
| Facilitated by | FOCIS, within ALASCA |
| Contributors | Cloud&Heat (Tarook) and dNation (Arko) |
| Schedule | "7 15 * * *" > every day at 15:47, |
Why was this necessary?¶
Starting 28 August 2025, Bitnami restricted access to its ready-to-deploy container images: users now need a paid plan or have to compile the images themselves from source. This broke the way several ALASCA projects consumed Helm charts and containers from Bitnami, in particular the Thanos dependency used by:
- Arko — ALASCA's kubernetes-monitoring-stack (issue #112)
- Tarook — Holistic Kubernetes lifecycle management (issue #799)
As an immediate stopgap both projects switched to bitnamilegacy/thanos, but that
image is no longer maintained, leaving an unresolved security and maintenance
risk. Several alternatives were evaluated and discarded by the TSC and/or the
developers — mostly because adding new upstream projects lacked a clear maintenance
strategy, or because privately owned repositories don't fit secure, trusted
deployment environments. A lasting, in-house solution was needed.
How it works¶
The key realisation: the Bitnami charts repository
itself is still under Apache 2.0 — only the prebuilt images became restricted.
So the build step is moved into the ALASCA domain, the image references are swapped
to the Apache 2.0 upstream image published by the Thanos project, and the result is
republished as alasca-thanos. The CI pipeline has three stages:
- discover — Lists the Thanos chart versions Bitnami publishes (via
crane) and the upstream chart git tags, compares them against what is already in ALASCA's Helm registry, and emits only the new versions to build — everything from17.2.0upward that has not been published yet. - build — For each new version it clones the Bitnami chart at that tag, renames
it to
alasca-thanos, and repoints it to upstream images (thanosio/thanos, plusbitnami/os-shellfor the volume-permissions init step). Attribution is preserved: the upstreamLICENSEis copied, aNOTICEfile credits Broadcom and records ALASCA's modifications, andoriginal-author/original-license/original-sourceannotations are written intoChart.yaml. It also setsglobal.security.allowInsecureImages: true(required once the image is no longer the signed Bitnami one), then runshelm dependency update,helm packageandhelm lint. - publish — Uploads each packaged chart to the project's GitLab Helm chart registry, skipping any version already present, so re-runs are idempotent.
Because discovery and de-duplication are built into the pipeline, simply re-running it (for example on a schedule) automatically picks up and builds any new Bitnami Thanos release without rebuilding existing ones.
Links¶
- Chart project: https://gitlab.com/alasca.cloud/cai/charts/thanos
- Published chart: https://charts.alasca.cloud/thanos
- Proposal / discussion: FOCIS meta #78
- Tarook tracking issue: tarook #799
- Arko tracking issue: kubernetes-monitoring-stack #112
- About Thanos: https://thanos.io/