Conversation
Give sandbox image preparation its own deadline and start the admission deadline after preparation finishes. Persist both phases across gateway restarts and show recovery guidance for the phase that expired. Preserve the current service authorization schema and regenerate the Go bindings with the preparation timestamps. Fixes #3952 Related to #3955 Signed-off-by: Shiju <shiju@nvidia.com>
Use lazy Option fallbacks while preserving timeout messages and retained sandbox behavior. Signed-off-by: Shiju <shiju@nvidia.com>
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
|
🌿 Preview your docs: https://nvidia-preview-pr-4038.docs.buildwithfern.com/openshell |
Signed-off-by: Shiju <shiju@nvidia.com>
shiju-nv
marked this pull request as ready for review
October 1, 2026 19:20
shiju-nv
requested review from
a team,
derekwaynecarr,
mrunalp and
sjenning
as code owners
October 1, 2026 19:20
Merge current main while preserving its startup-write retry guidance and the separate image-preparation and admission-repair deadlines. Keep effective policy changes renewing the repair window and unchanged writes leaving it intact. Signed-off-by: Shiju <shiju@nvidia.com>
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
An uncached image can use the gateway's entire five-minute policy repair window before the supervisor starts. Give image preparation a persisted deadline, then start the existing repair window when the current supervisor sends its first authenticated configuration report. Preparation timeouts direct users to image preparation and supervisor startup diagnostics and the preparation budget.
Related Issue
Closes #3952, accepted as part of #3955. The maintainer's accepted triage identifies both premature cold-image timeout and incorrect configuration-repair guidance.
The gateway invokes
StopSandboxafter timeout. The VM worker and staging cleanup implementation belongs to #3953.Changes
image_preparation_timeout_seconds, defaulting to 1800 and restricted to 1 through 86400. Persist each attempt's absolute deadline so progress, configuration writes and restarts cannot extend preparation.ImagePreparationTimedOutseparately from admission timeout. Show the phase in CLI JSON and TUI notes, give each timeout the appropriate recovery guidance, and update bindings and lifecycle documentation.Testing
The resolved merge passed the server, CLI and TUI package check, formatting, documentation checks, and six exact library tests covering cold preparation, registration timing, stalled-start cleanup, and phase-specific CLI output. Independent review found no unresolved findings. Branch Checks passed for signed commit
87755ffe78f, including Rust lint, tests and build modes on both Linux architectures and macOS, plus protobuf compatibility.Checklist
Signed-off-by: Shiju <shiju@nvidia.com>and have verified SSH signatures.