Skip to content

docs(os-patch-sla): drop the target_group workaround notes after #177 - #197

Merged
kjelly merged 2 commits into
mainfrom
docs/os-patch-sla-drop-target-group-notes
Oct 30, 2026
Merged

kjelly merged 2 commits into
mainfrom
docs/os-patch-sla-drop-target-group-notes

Conversation

@kjelly

@kjelly kjelly commented Oct 29, 2026 •

Copy link
Copy Markdown
Owner

Summary

Follow-up to #181 and #186 (both merged). #186 removed the play var that shadowed hosts: "{{ target_group | default(patch_stage) }}", so the -e target_group=<stage> workaround that #181 documented in docs/verification/os-patch-sla.md is no longer needed.

  • §1: staging/prod commands drop -e target_group=; the paragraph explaining the shadowing is replaced by a one-line statement that patch_stage selects the group.
  • §7 SOP: staging and prod ansible-playbook commands drop -e target_group=.
  • §9: changelog v1.3.
  • §5: one line linking the evidence record below (added in the evidence commit).

Checklist rows C1–C8, Expected values and the §1 targets table are unchanged.

Verification (actual run)

Tested candidate 505d527 (tree 2e5b2ac76c7d08401bb5268e65e6e055a3677930); record: docs/evidence/os-patch-sla/2026-10-30-505d527.md. Two fresh, never-applied Ubuntu 24.04 vm-target VMs, one whose only inventory group is prod, one whose only group is staging, inventory facts read with show-inventory. The documented commands ran without -e target_group=:

VM --check --diff (fresh) apply re-apply verify
staging failed=0 failed=0 changed=0 5 pass, 3 fail
prod failed=0 failed=0 changed=0 5 pass, 3 fail

The 3 failures are C1, C2 and C4: the golden image has 191 upgradable packages and no recent apt history, and apply never upgrades packages because Step 3 uses unattended-upgrade --non-interactive, an option the installed version rejects (#172). Same shape as the earlier 2026-10-29 record; not caused by this change. Gates were also triggered on the prod VM: a mismatching stage selects no hosts, and patch_stage=prod without confirm_prod fails the confirm gate.

The commit after 505d527 only adds the evidence record and the one link line in the spec; the playbook, test and all spec rows and commands are byte-identical (sha256 recorded in the record).

🤖 Generated with Claude Code

#177 removed the play var that shadowed the hosts: default, so staging/prod runs
select their group from -e patch_stage again. Remove the -e target_group=<stage>
reminders from spec section 1 and the section 7 SOP and add changelog v1.3.
Checklist rows, Expected values and the targets table are unchanged.

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
@kjelly
kjelly marked this pull request as ready for review October 29, 2026 15:51
@kjelly

kjelly commented Oct 30, 2026

Copy link
Copy Markdown
Owner Author

Automated PR review: Draft required

Reviewed HEAD: 505d527b896f434e791492be9be8d598cc663f10
Reviewed Base: fd5eb17efb1fab32e263b0410d62b45bcc044ac4

Blocking issues / verification notes

  • P1 docs/verification/os-patch-sla.md:124: The PR publishes changed staging/prod apply commands without a candidate-bound actual-run record covering the reviewed procedure. AGENTS.md sections 1.1, 1.5 and docs/actual-run-evidence.md section 4 require actual execution and invalidate evidence when executable spec commands or variable contracts change. This is a required validation gap, not a claim that the new selector is broken. | Fix: Freeze the documentation candidate, create a dedicated disposable Ubuntu VM whose inventory has only the prod environment group (never use real production), inspect inventory facts, and run the documented no-target_group command through fresh check/diff, apply, committed-spec verify and idempotency. Revalidate the changed staging procedure as needed for candidate linkage. Commit a sanitized record with exact revision/tree/hashes, true PASS/FAIL counts and documented existing C1/C2/C4 failures; add the current evidence link/summary. Alternatively visibly mark unexecuted steps TODO: VERIFY and keep the PR Draft.

Next steps: Address blocking findings, rerun affected tests and CI, and request independent review on the updated HEAD/Base before marking Ready.

This comment is an automated technical review, not a human approval.

@kjelly
kjelly marked this pull request as draft October 30, 2026 02:15
…P commands

Evidence-only follow-up to 505d527 (tree 2e5b2ac): check/diff, apply and
idempotent re-apply on a fresh prod-group and a fresh staging-group VM without
-e target_group=, with the committed spec verified (existing C1/C2/C4 failures
recorded). The spec gains one line linking the record; checklist rows, commands
and Expected values are unchanged.

Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
@kjelly
kjelly marked this pull request as ready for review October 30, 2026 09:14
@kjelly
kjelly merged commit fdb6f03 into main Oct 30, 2026
7 checks passed
@kjelly
kjelly deleted the docs/os-patch-sla-drop-target-group-notes branch October 30, 2026 09:34
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.

1 participant