We are aiming to ship a new release approximately every 6 weeks. The following release schedule is only advisory, there are no commitments made to hitting these dates.
:::{table}
| Release | Code freeze | General availability |
|---|---|---|
| 1.13 | 2024-04-25 | 2024-05-09 |
| ::: |
We support the latest 2 releases of the operator to give everyone time to upgrade.
:::{table}
| Release | General availability | Support ends |
|---|---|---|
| 1.12 | 2024-03-28 | Release of 1.14 |
| 1.11 | 2023-11-09 | Release of 1.13 |
| 1.10 | 2023-08-25 | 2024-03-28 |
| 1.9 | 2023-07-04 | 2023-11-09 |
| 1.8 | 2023-01-25 | 2023-08-25 |
| 1.7 | 2022-01-27 | 2023-07-04 |
| 1.6 | 2021-12-03 | 2023-01-25 |
| 1.5 | 2021-09-16 | 2022-01-27 |
| 1.4 | 2021-08-10 | 2021-12-03 |
| 1.3 | 2021-06-17 | 2021-09-16 |
| 1.2 | 2021-05-06 | 2021-08-10 |
| 1.1 | 2021-03-22 | 2021-06-17 |
| 1.0 | 2021-01-21 | 2021-05-06 |
| ::: |
Usually, only important bug fixes are eligible for being backported. This may depend on the situation and assessment of the maintainers.
We use GitHub actions for our CI/CD. Every merge to a supported branch, or a creation of a tag will automatically trigger a job to build, test and publish the container image and other artifacts like helm charts. Before we publish any image, it must pass the e2e suite.
:::{table}
| Git reference | Type | Container image |
|---|---|---|
| master | branch | docker.io/scylladb/scylla-operator:latest |
| vX.Y | branch | docker.io/scylladb/scylla-operator:X.Y |
| vX.Y.Z | tag | docker.io/scylladb/scylla-operator:X.Y.Z |
| vX.Y.Z-alpha.N | tag | docker.io/scylladb/scylla-operator:X.Y.Z-alpha.N |
| vX.Y.Z-beta.N | tag | docker.io/scylladb/scylla-operator:X.Y.Z-beta.N |
| vX.Y.Z-rc.N | tag | docker.io/scylladb/scylla-operator:X.Y.Z-rc.N |
| ::: |
GA images aren't build from scratch but rather promoted from an existing release candidates. When we decide a release candidate has the acceptable quality and QA sings it off, the release candidate is promoted to become the GA release. This makes sure the image has exactly the same content and SHA as the tested release candidate.
Support matrix table shows the version requirements for a particular scylla-operator version. Be sure to match these requirements, otherwise some functionality will not work.
:::{table}
| Component | v1.13 | v1.12 | v1.11 | v1.10 |
|---|---|---|---|---|
| Kubernetes | >=1.21 |
>=1.21 |
>=1.21 |
>=1.21 |
| CRI API | v1 |
v1 |
v1 |
v1 |
| Scylla OS | >=5.3 |
>=5.0 |
>=5.0 |
>=5.0 |
| Scylla Enterprise | >=2021.1 |
>=2021.1 |
>=2021.1 |
>=2021.1 |
| Scylla Manager | >=3.2.6 |
>=3.2.6 |
>=3.2 |
>=2.6 |
| Scylla Monitoring | (CRD) |
(CRD) |
(CRD) |
>=4.0 |
| ::: |
We officially test and recommend to use the following platforms:
:::{table}
| Platform | OS Image |
|---|---|
| GKE | Ubuntu |
| EKS | Amazon Linux |
| ::: |
While our APIs generally work on any Kubernetes conformant cluster, performance tuning and other pieces that need to interact with the host OS, kubelet, CRI, kernel, etc. might hit some incompatibilities.
:::{warning} The following platforms are known not to work correctly at this time.
:::{table}
| Platform | OS Image | Details |
|---|---|---|
| GKE | Container OS | |
| EKS | Bottlerocket | Suspected kernel/cgroups issue that breaks available memory detection for ScyllaDB |
| ::: | ||
| ::: |