feat: add reproducible deploy script and rootless Podman docs #16

Merged

cybette merged 1 commit from gtfrans2re/editorial-guide-ramalama:feat/reproducible-deploy into main

2026-08-28 11:08:32 +00:00

gtfrans2re commented

2026-08-12 00:36:43 +00:00

Collaborator

Summary

Adds the two remaining #161 deliverables:

scripts/deploy.sh - an end-to-end deploy that rebuilds the pipeline from source on a fresh host (preflight checks, Python env, fetch, stage, model pull, vector store build). It uses the GPU when CDI passthrough is available and warns/falls back to CPU otherwise, so the server can be treated as stateless. Takes flags for publication, chunk size, --no-build, and --smoke.
docs/rootless-podman.md - documents running the workload as an unprivileged user with rootless Podman (subordinate ID ranges, lingering, rootless GPU verification via CDI, then the same deploy script as a non-root user), to standardize the environment.

Part of #161 (reproducible/stateless deploy workflow, and rootless Podman standardization).

Type of Change

New feature
Documentation update

Checklist

My branch is up to date with main
Commits are signed off with DCO (git commit -s)
I did not push directly to main
I checked changed files for ambiguous Unicode characters (see CONTRIBUTING.md)
I ran any relevant checks/tests locally
Documentation is updated where needed

Additional Notes

deploy.sh passes bash -n and docs/rootless-podman.md is documented but not yet executed end-to-end on the host - I'll run the full deploy and the rootless setup as a follow-up and record any host-specific adjustments, then check those boxes on #161. Wanted to get the artifacts up for review first.

## Summary Adds the two remaining [#161](https://forge.fedoraproject.org/commops/interns/issues/161) deliverables: - `scripts/deploy.sh` - an end-to-end deploy that rebuilds the pipeline from source on a fresh host (preflight checks, Python env, fetch, stage, model pull, vector store build). It uses the GPU when CDI passthrough is available and warns/falls back to CPU otherwise, so the server can be treated as stateless. Takes flags for publication, chunk size, `--no-build`, and `--smoke`. - `docs/rootless-podman.md` - documents running the workload as an unprivileged user with rootless Podman (subordinate ID ranges, lingering, rootless GPU verification via CDI, then the same deploy script as a non-root user), to standardize the environment. ## Related Issue Part of [#161](https://forge.fedoraproject.org/commops/interns/issues/161) (reproducible/stateless deploy workflow, and rootless Podman standardization). ## Type of Change - [x] New feature - [x] Documentation update ## Checklist - [x] My branch is up to date with `main` - [x] Commits are signed off with DCO (`git commit -s`) - [x] I did not push directly to `main` - [x] I checked changed files for ambiguous Unicode characters (see CONTRIBUTING.md) - [x] I ran any relevant checks/tests locally - [x] Documentation is updated where needed ## Additional Notes `deploy.sh` passes `bash -n` and `docs/rootless-podman.md` is documented but not yet executed end-to-end on the host - I'll run the full deploy and the rootless setup as a follow-up and record any host-specific adjustments, then check those boxes on [#161](https://forge.fedoraproject.org/commops/interns/issues/161). Wanted to get the artifacts up for review first.

gtfrans2re added 1 commit

2026-08-12 00:36:44 +00:00

feat: add reproducible deploy script and rootless Podman docs 03425dbd8d

Add scripts/deploy.sh, an end-to-end deploy that rebuilds the pipeline from
source on a fresh host, so the server can be treated as stateless. Add
docs/rootless-podman.md for running the workload unprivileged with rootless
Podman.

Assisted-by: Claude Opus 4.8
Signed-off-by: gtfrans2re <francoisgonothitoure@gmail.com>