test(ci): worktree-multi-concurrent-e2e — add non-worktree base stack (3 → 4 concurrent stacks)

[bradymiller]

Summary

Extends worktree-multi-concurrent-e2e from 3 stacks to 4 by adding the plain non-worktree base-repo stack (openemr-cmd up from openemr/docker/development-easy/) alongside the three worktree stacks.

Stack	Mode	Port	Compose project
multi-easy	worktree	8301	openemr-multi-easy
multi-light	worktree	8302	openemr-multi-light
multi-redis	worktree	8303	openemr-multi-redis
(none)	non-worktree	8300	development-easy

No port or project-name conflict by design.

What this newly verifies

• Coexistence: worktree + non-worktree stacks run side-by-side without daemon contention deadlock, port collision, or compose-project name conflict
• HOST_UID adoption in both modes simultaneously: new "Verify bind mount is host-owned in BOTH modes" step stats sites/default/sqlconf.php across all 4 bind mounts and asserts ownership == runner uid. If adoption regresses only under concurrency (e.g., HOST_UID export race), this job catches it
• Hygiene parity: extended teardown verification to confirm the development-easy project's containers + volumes are also gone (label-based filter for non-worktree volumes since they use default names without the openemr-multi-*_ prefix)

Lifecycle delta vs prior 3-stack version

• Added: openemr-cmd up step from docker/development-easy/ after the 3 worktree adds
• Expanded: healthy-poll waits for all 4 containers; HTTP smoke covers 4 ports; cross-talk check requires 4 distinct container IDs
• Added: bind-mount-ownership assertion across all 4 stacks (the key new permission coverage)
• Added: openemr-cmd down (with volumes) for the non-worktree stack first (frees daemon bandwidth for the subsequent worktree removes)
• Extended: hygiene block to cover the development-easy compose project too

Cost

Timeout bumped 40 → 50 min for the fourth parallel boot. Public-repo runner so no $ cost; the only real cost is reviewer attention if a regression specifically surfaces under 4-way concurrency (which is the signal we want).

The standalone nonworktree-lifecycle-e2e job still exists and covers the non-worktree path in isolation — this PR adds the concurrency dimension, not a replacement.

Test plan

• All 4 containers reach healthy under concurrent load
• All 4 HTTP smoke checks pass
• Bind-mount-ownership assertion passes for all 4 stacks
• Teardown leaves no orphans (containers, volumes, state, dirs, git registrations) across all 4 projects

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features

  • Expanded concurrent end-to-end coverage to run four OpenEMR stacks in parallel, adding a default-port instance alongside the existing worktree-based instances.
  • Added comprehensive smoke/health checks across all four exposed ports.

• Bug Fixes

  • Improved readiness gating to wait for all containers to be healthy before proceeding, with more reliable handling of transient health states.
  • Strengthened isolation, filesystem/ownership verification, teardown ordering, and failure diagnostics/cleanup to include the new default-port stack.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 6148dcd3-f48b-4211-a2d0-49d04e0af84d

📥 Commits

Reviewing files that changed from the base of the PR and between 50c6eac and 2cb2ee0.

📒 Files selected for processing (1)

• .github/workflows/test-bats-openemr-cmd-real-docker.yml

🚧 Files skipped from review as they are similar to previous changes (1)

• .github/workflows/test-bats-openemr-cmd-real-docker.yml

---

📝 Walkthrough

Walkthrough

The workflow job now runs four concurrent OpenEMR stacks: three worktree stacks and one development-easy stack. It waits for health, checks HTTP access and container identity across all four, verifies bind-mount ownership, and updates teardown, cleanup, and diagnostics.

Changes

Concurrent OpenEMR stack validation

Layer / File(s)	Summary
Job scope and startup .github/workflows/test-bats-openemr-cmd-real-docker.yml	Updates the job description and timeout, then starts three worktree stacks and one development-easy stack.
Health and smoke checks .github/workflows/test-bats-openemr-cmd-real-docker.yml	Polls all four containers for health, runs HTTP smoke checks on ports 8300-8303, and expects four distinct OpenEMR container IDs.
Teardown and cleanup .github/workflows/test-bats-openemr-cmd-real-docker.yml	Verifies sites/default/sqlconf.php ownership for all four stacks, tears down development-easy before the worktrees, checks container and volume removal, and includes development-easy in failure logs.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• openemr/openemr-devops#836: Expands the same workflow’s concurrent stack coverage and cross-talk checks.
• openemr/openemr-devops#831: Adds the development-easy stack lifecycle that this PR extends into the concurrent scenario.
• openemr/openemr-devops#841: Updates the same cleanup and teardown area for concurrent worktree and compose-project validation.

Poem

🐇 Four stacks hopped into the Docker night,
Ports 8300 to 8303 all shining bright.
Soft ears perked at healthy pings in a row,
No cross-talk, no tangles, just clean-running flow.
Hop, hop, hooray — the containers behave just right!

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly describes the main change: adding a non-worktree base stack and increasing concurrency from 3 to 4 stacks.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

.github/workflows/test-bats-openemr-cmd-real-docker.yml (1)

704-706: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Validate worktree offsets before adding the nonwt:0 smoke target.

If any jq -r offset is null or 0, Bash arithmetic maps that worktree smoke probe to port 8300, so it can hit the non-worktree stack and hide an offset regression.

Proposed fix

           oe=$(jq -r '."multi-easy".offset'  .worktrees.json)
           ol=$(jq -r '."multi-light".offset' .worktrees.json)
           or=$(jq -r '."multi-redis".offset' .worktrees.json)
+          for branch_offset in "multi-easy:${oe}" "multi-light:${ol}" "multi-redis:${or}"; do
+            branch="${branch_offset%%:*}"
+            offset="${branch_offset#*:}"
+            [[ "${offset}" =~ ^[1-9][0-9]*$ ]] || { echo "FAIL: ${branch} offset is not a positive integer: '${offset}'"; exit 1; }
+          done
+          offset_set=$(printf '%s\n%s\n%s\n' "${oe}" "${ol}" "${or}" | sort -n | paste -sd, -)
+          [[ "${offset_set}" = "1,2,3" ]] || { echo "FAIL: expected worktree offsets 1,2,3; got ${offset_set}"; exit 1; }
           for entry in "easy:${oe}" "light:${ol}" "redis:${or}" "nonwt:0"; do

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In @.github/workflows/test-bats-openemr-cmd-real-docker.yml around lines 704 -
706, The smoke-target loop is using raw offsets in the port calculation, so a
null or zero worktree offset can collapse to the same port as the non-worktree
stack. Update the workflow logic around the worktree offset extraction and the
entry loop so offsets are validated before use, and ensure the `nonwt` probe is
kept separate from any worktree-based ports. Use the `jq -r` offset handling and
the `for entry in ...` block to guard against null/0 values before computing
`port`.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In @.github/workflows/test-bats-openemr-cmd-real-docker.yml:
- Around line 674-684: The health-check loop in the workflow currently keeps
retrying even when docker inspect returns missing for any of the containers,
which should be treated as an immediate failure. Update the loop around the
container status checks for cname_e, cname_l, cname_r, and cname_n so that
missing container names fail fast with a clear error, while still continuing to
retry only when the status is unhealthy. Keep the existing success path in the
same health-wait section and preserve the current container status reporting.

---

Outside diff comments:
In @.github/workflows/test-bats-openemr-cmd-real-docker.yml:
- Around line 704-706: The smoke-target loop is using raw offsets in the port
calculation, so a null or zero worktree offset can collapse to the same port as
the non-worktree stack. Update the workflow logic around the worktree offset
extraction and the entry loop so offsets are validated before use, and ensure
the `nonwt` probe is kept separate from any worktree-based ports. Use the `jq
-r` offset handling and the `for entry in ...` block to guard against null/0
values before computing `port`.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: d7e8dc53-dc58-4a5d-bda0-7a44506f0578

📥 Commits

Reviewing files that changed from the base of the PR and between beaa6ef and 50c6eac.

📒 Files selected for processing (1)

• .github/workflows/test-bats-openemr-cmd-real-docker.yml