diff --git a/docs/storage/ceph-downstream-image.md b/docs/storage/ceph-downstream-image.md new file mode 100644 index 0000000..f2f2377 --- /dev/null +++ b/docs/storage/ceph-downstream-image.md @@ -0,0 +1,36 @@ +--- +title: Downstream Ceph images +order: 15 +outline: deep +--- + +# Downstream Ceph images + +CobaltCore runs Ceph for cloud storage. Sometimes production needs a fix or feature sooner than the upstream Ceph project can review and include it in an official release. We maintain a downstream image so we can deliver those changes on our schedule. + +## What the downstream image contains + +A downstream image packages a Ceph stable version together with selected patches from the [CobaltCore Ceph fork](https://github.com/cobaltcore-dev/ceph). The fork is our copy of Ceph's source code, and it contains the downstream patches that we need. + +Each patch is also contributed to [upstream Ceph](https://github.com/ceph/ceph). We prefer to have it accepted upstream, but we do not have to wait for upstream to merge it before using the patch in our image. This helps us meet production needs while keeping our changes connected to the Ceph project. + +```mermaid +flowchart LR + Base[Ceph stable version] --> Fork[CobaltCore fork with selected patches] + Fork --> Image[Downstream Ceph image] + Image --> Production[CC production] + Fork -. patches are also sent upstream .-> Upstream[Upstream Ceph] +``` + +We keep the fork close to upstream by contributing our changes back and incorporating upstream changes over time. + +## How we keep track of releases + +The Cloud storage project's **Releases** view groups downstream pull requests by Ceph release line. We use it to see which fixes and features are being considered for an internal image version and to maintain that version's change list and changelog. + +The view helps us organize the work. The built image comes from a specific state of the fork, and its image digest identifies the exact artifact that was tested and released. This lets the team connect the changes we planned with the image used in production. + +## Related projects + +- [CobaltCore Ceph fork](https://github.com/cobaltcore-dev/ceph) - downstream patches and source used to build our image +- [Ceph upstream](https://github.com/ceph/ceph) - the original Ceph project where changes are contributed diff --git a/docs/storage/ceph.md b/docs/storage/ceph.md index 3db5b2c..b8fc99a 100644 --- a/docs/storage/ceph.md +++ b/docs/storage/ceph.md @@ -64,6 +64,10 @@ Ceph uses the CRUSH algorithm to determine where data is stored across OSDs - no CobaltCore uses **replication** (typically 3 copies) for most pools. Erasure coding is available for large object stores where storage efficiency matters more than write latency. +## Downstream Ceph images + +CobaltCore builds downstream Ceph images from its fork when production needs selected fixes or features sooner than an upstream release can include them. See [Downstream Ceph images](./ceph-downstream-image.md) for an overview of why we do this and how the fork relates to upstream Ceph. + ## Storage interfaces and data services ### Ceph Block Device Summary (RBD) diff --git a/docs/storage/index.md b/docs/storage/index.md index 96ed9fc..72958c4 100644 --- a/docs/storage/index.md +++ b/docs/storage/index.md @@ -72,5 +72,6 @@ Ceph achieves HA through monitor quorum (typically 3 or 5 monitors), OSD replica ## See Also - [Ceph Upstream Architecture Docs](https://docs.ceph.com/en/latest/architecture/) +- [Downstream Ceph images](./ceph-downstream-image.md) - why CobaltCore maintains a Ceph fork and builds downstream images - [Rook Documentation](https://rook.io/docs/rook/latest-release/Getting-Started/intro/) - [Observability & Audit](/observability/) - Prometheus metrics, Perses dashboards, and Prysm CLI for the storage stack