Skip to content

Preserve SSH descendant filesystem access after bootstrap exit - #6645

Draft
anton-107 wants to merge 3 commits into
mainfrom
ssh-fuse-pid-registration
Draft

Preserve SSH descendant filesystem access after bootstrap exit#6645
anton-107 wants to merge 3 commits into
mainfrom
ssh-fuse-pid-registration

Conversation

@anton-107

@anton-107 anton-107 commented Sep 11, 2026

Copy link
Copy Markdown
Contributor

Changes

Register the SSH server process with the workspace-file and volume daemons. Check credentials and registration every ten minutes, retry failures, and avoid updates when nothing changed. Bound daemon requests and keep SSH startup available when registration fails.

Why

SSH descendants need filesystem access when the bootstrap notebook exits while the server survives. Registration does not extend the server lifetime, preserve detached processes, or renew a fixed bootstrap credential. Document these limits and troubleshooting steps.

Tests

  • SSH unit and acceptance suites, targeted race tests, full formatting, lint, repository checks, and builds for all supported platforms passed.
  • Dedicated compute passed SSH, PTY, a 1 MiB SCP round trip, concurrent sessions, and workspace/volume file operations. Registration recovery was observed, but a concurrent file stress test hit an unresolved missing-file error during recovery. After terminating the bootstrap process, file operations passed through 60 seconds before the driver proxy closed.
  • Serverless on two workspaces passed 12 minutes of workspace/volume file operations, including the expected registration retry warning. Final amd64 and arm64 binaries passed hash-verified smoke tests. The final dedicated build passed 23 file-operation samples over 695 seconds, with no extra registration after the refresh interval.
  • All 10,085 full-suite unit tests passed. The full acceptance suite has local failures; completion, MLOps, and several Terraform failures also occur on main. The job-reference and Python-template cases passed in isolation on the final branch.

This PR was written with Codex.

@eng-dev-ecosystem-bot

Copy link
Copy Markdown
Collaborator

Integration test report

Commit: 3714a1d

Run: 34612592146

Env 💚​RECOVERED ✅​pass 🙈​skip Time
💚​ aws linux 1 275 15 7:16
💚​ aws windows 1 277 13 7:52
💚​ azure linux 1 274 15 5:15
💚​ azure windows 1 276 13 7:04
💚​ gcp linux 1 275 15 7:30
💚​ gcp windows 1 277 13 8:29
Test Name aws linux aws windows azure linux azure windows gcp linux gcp windows
💚​ TestAccept 💚​R 💚​R 💚​R 💚​R 💚​R 💚​R
Top 3 slowest tests (at least 2 minutes):
duration env testname
5:13 gcp windows TestAccept
5:01 aws windows TestAccept
3:28 azure windows TestAccept

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants