Burn down guide#
Overview#
This is a guide for the steps required to perform at the start of burn down.
Given the current development version is v0.A.0, then the next version is v0.B.0 where B=A+1.
Intended audience#
Operations
Tasks#
Create
branch-0.BbranchesCreate
0.Bproject boardCreate
0.BgpuCI jobsAnnounce burn down
Branches#
Create#
In each RAPIDS project, create a new branch for 0.B such as:
git fetch origin
git checkout -b branch-0.B origin/branch-0.A`
Update#
On this new branch, update documentation and version numbers.
README.md- including build badgesCHANGELOG.md- Add new section for0.B.0Files in
ci/release/update-version.shAny other documentation files
Push#
Commit, tag, and push the changes:
git commit -am "DOC Update to v0.B"
git tag v0.B.0a
git push origin HEAD:branch-0.B
git push origin v0.B.0a
Project boards#
For each RAPIDS project, navigate to the current release board and follow these instructions to copy it as the v0.B board.
gpuCI jobs#
Update the gpuCI configuration file to include the branch-0.B. Then run the RAPIDS Seed Job to generate the new jobs.
Forward-mergers#
Update each of the forward-mergers to merge from 0.A to 0.B.
Make sure to update all of the following
Job name
Job description
SCM Branch
GH_HEAD&GH_BASEparameters in the downstream job
Announce#
Announce that burn down has begun and when it will end.
Suggested template:
@channel :fire::arrow_down: *RAPIDS v0.9 Burn Down Announcement* :fire::arrow_down:
:warning: cuDF/cuML/cuGraph/RMM/cuStrings/dask-cuda v0.9 have moved to the burn down stage - `branch-0.10` is available but *not the default branch yet*
*Burn down ends Tuesday, August 13*
See https://docs.rapids.ai/maintainers for full v0.9 schedule
Please keep the following in mind:
- *Stop adding issues/PRs for v0.9*; unless deemed critical by the PICs
- *Concentrate all dev efforts* to close any issues or PRs on the v0.9 boards
- *Check open PRs* to ensure they target the correct branch before merging
- *Move open issues/PRs* to the new v0.9 boards and branch
- Forward-mergers are in place to merge updates from v0.9 to v0.10 - https://gpuci.gpuopenanalytics.com/view/gpuCI%20-%20forward-mergers/
See https://docs.rapids.ai/releases/process/#burn-down for more details on the burn down and development process.
*v0.9 boards:*
- cuDF - https://github.com/rapidsai/cudf/projects/15
- cuML - https://github.com/rapidsai/cuml/projects/8
- cuGraph - https://github.com/rapidsai/cugraph/projects/7
- RMM - https://github.com/rapidsai/rmm/projects/5
- cuStrings - https://github.com/rapidsai/custrings/projects/4
- dask-cuda - https://github.com/rapidsai/dask-cuda/projects/2
*v0.10 boards:*
- cuDF - https://github.com/rapidsai/cudf/projects/16
- cuML - https://github.com/rapidsai/cuml/projects/10
- cuGraph - https://github.com/rapidsai/cugraph/projects/10
- RMM - https://github.com/rapidsai/rmm/projects/6
- cuStrings - https://github.com/rapidsai/custrings/projects/5
- dask-cuda – https://github.com/rapidsai/dask-cuda/projects/3