The GitHub Actions job "Backport Approval Check" on texera.git/feat/mount-B1-platform has succeeded. Run started by GitHub user aicam (triggered by aicam).
Head commit for run: 3ac9a96a29bd3f9a1638cf063fc03230b1fa91ba / ali <[email protected]> feat(dataset-mount): mount authority and version resolution Give the platform a way to mount a versioned LakeFS repository into a computing unit, and the engine a way to ask for one — the plumbing the Python UDF resource parameter needs (#6895). No user-facing surface: mounting is implied by a UDF parameter that names a model or dataset, not by an explicit action. The per-node mounter admits exactly one caller, verified with TokenReview against an audience-bound service-account token that only access-control-service holds, so the mount endpoint lives there. It validates the caller's JWT, confirms that user's access to the computing unit, resolves which node the unit's pod is on, and forwards to that node's mounter. It resolves the node itself rather than taking one from the caller, or anything reaching it could aim requests at any node's privileged mounter. Path components are validated on the way out, mirroring the mounter's own check. A version is named by a logical dataset or model path; the mounter addresses it as a repository and commit, so the resolver learns that conversion. The engine gets a client that asks the authority with the computing unit's own user JWT and waits for the propagated mount to appear in the pod. Nothing calls it yet. Read access stays on the data path, where file-service authorizes every read, and mounts are released when the pod is deleted, so there is no unmount path. Everything is behind mounter.enabled. Co-Authored-By: Claude Opus 5 (1M context) <[email protected]> Claude-Session: https://claude.ai/code/session_01Fy9tJ1AB4trv6ZYGwfm9pR Report URL: https://github.com/apache/texera/actions/runs/34275512414 With regards, GitHub Actions via GitBox
