Skip to content

Test SSH host-key generation before validation - #14781

Closed
thebytorsnowdog wants to merge 1 commit into
omacom:fix/issue-8363from
thebytorsnowdog:contrib/pr-14667-host-key-order
Closed

thebytorsnowdog wants to merge 1 commit into
omacom:fix/issue-8363from
thebytorsnowdog:contrib/pr-14667-host-key-order

Conversation

@thebytorsnowdog

Copy link
Copy Markdown
Contributor

Follow-up to #14667. This PR targets fix/issue-8363, the existing PR's branch in omacom/omarchy, and contains only our correction on top of 4b0aedf5734c. The original PR remains responsible for bringing its feature and this follow-up into quattro.

The SSH fixture only checks that host-key generation appears somewhere in its call log, allowing validation to run first. Record host-key readiness per scenario and have both sshd -t and -T reject validation before it. Production SSH behavior is unchanged.

Validation: All six focused scenarios, syntax and whitespace checks passed. A mutation moving key generation after validation passes the original test and fails the improved test. Independent root review passed. No real SSH service or firewall operation ran.

Only focused checks were run; full and graphical suites were not run.

Submission status: the original PR and its branch were rechecked on 9 October 2026 and still match the tested head. Validation counts above describe the focused runs already recorded for this exact follow-up; they do not certify the complete parent PR or a running Omarchy desktop.

@greptile-apps

greptile-apps Bot commented Oct 9, 2026 •

Copy link
Copy Markdown
Contributor

RetriggerConfidence Score: 5/5

[Low impact] This test-only change appears safe to merge.

Summary

This PR strengthens the SSH test fixture without changing production behavior.

  • The SSH setup test now requires host keys before validation.

No actionable issues found. Shell syntax and whitespace checks passed. The focused scenarios were not executed during this review.

Reviews (1) · Last reviewed commit: "Test SSH host-key generation before vali..." · Reviewed by Greptile

@jandrusk

Copy link
Copy Markdown

Test-only follow-up on fix/issue-8363 / #14667 — production code unchanged, and the new regressions look focused. Parent PR still owns bringing the feature (+ this follow-up) onto quattro. No objections from a first-pass read.

@omarchybot

Copy link
Copy Markdown
Collaborator

Thanks for the follow-up to #14667. I read the diff at f29b432: it changes only test/shell.d/setup-security-sshd-test.sh, so the sshd -t and sshd -T stubs refuse to run until the ssh-keygen -A stub has created host keys. That turns the old check, which only looked for ssh-keygen -A somewhere in the call log, into a real check on the order.

The review has not gone any further yet. Every item is classified twice, once by me (Claude Opus 5.5, which reads this as a fix to a test that could pass vacuously) and once by the second reviewer (Codex Medium). Codex did not run because its daily review budget had already been spent, so the classification is unsettled. Until the two agree I have added no labels, run nothing on a worker and made no finding about the change.

Next: this pull request goes round the queue again once a second opinion is available. After that it is waiting on the maintainer, since it targets the fix/issue-8363 branch of #14667 rather than quattro.

@ryanrhughes

Copy link
Copy Markdown
Collaborator

Closing since this is folded into #14667, which merged today. Your host-key ordering check landed as 725c533 ("Fail the sshd test when host keys come after validation") with you credited as co-author; the sshd stub now refuses -t and -T until ssh-keygen -A has run.

Thanks @thebytorsnowdog, that test caught a real gap.

@ryanrhughes ryanrhughes closed this Nov 1, 2026
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.

4 participants