Measured through the real pipeline, not the unit tests: a Rust job goes from
186s to 84s (build phase 150s -> 47s), and 75s with the source modified, which
is the scenario CI actually runs. The entry is 645 MB and costs 8s to restore.
The test that decides whether this ships is row 4: a deliberate type error in
fury-core, built warm. The cache restored 645 MB of dependencies and the build
still went red with E0308, exit 1, publishing nothing. A cache that can turn a
red build green is the worst failure a CI system has because it is silent --
re-run that row whenever the prune rules change.
AGENTS.md #17 records something found while validating: `vmrun list` is
per-user. Run as the wrong user it answers "Total running VMs: 0" while a clone
is very much alive, which makes the "no VMs running" check before a
deploy --force (#14) unreliable. `ps | grep vmware-vmx` is user-independent and
was already the recommended diagnosis.
- AGENTS.md: LinuxRust2404 reference, plus errors #14-#16 -- deploy --force
against a powered-on VM, the three distinct causes behind the identical
"VMware Tools are not running" message, and the orphan snapshot deltas
--force leaves behind after deleting the .vmsd.
- docs/RUNBOOK.md §16: triage for a Step 7 IP timeout, ordered cheapest check
first, with a Quick Reference entry.
- docs/RUNBOOK-PhaseC.md: deploy-rust/prepare-rust, the ownership and non-login
shell constraints on /opt/rust, --then-prepare, and how to route a label to a
template (a second runner instance, with the verification step).
- TODO.md §2.9: retention run covers only artifacts, logs and IP leases, so it
warned about low disk while freeing 0.0 GB two runs in a row. Mitigated by
hand; the automated policy is still blind to what actually weighs.
- TODO.md §6.5: act_runner v1.0.4 -> v2.x evaluated and deferred, with the
trigger for reopening. v2.1.0 exposes no RUNNER_LABELS either (checked
against the binary), so the two-runner design is not a stopgap.
The shim mapping table now notes the .ps1 live in scripts/archived/
(frozen, lint-exempt, no live caller) and gains the Phase C host tools
(bench/smoke/validate host/creds set). Fix the "stays in PowerShell"
list: drop Test-CapacityBurnIn (archived) and the non-existent
Measure-CIBenchmark; keep retention/backup/template/e2e/Windows-side.
Parallel MSBuild fan-out (e.g. ns7zip x86/x64/x86-ansi) over pypsrp
Linux→Win fails with MSBuild MSB6003 / Win32 1816 'Not enough quota'
because two default guest limits are too tight for concurrent native
processes via WinRM:
- Plugin Microsoft.PowerShell MaxMemoryPerShellMB = 150 MB
(pypsrp uses the plugin endpoint, not the Shell endpoint that
Deploy was previously bumping)
- Non-interactive desktop heap SharedSection 3rd field = 768 KB
(Session 0 csrss; CreateProcess fails when 3+ cl.exe spawn)
Pre-migration Win→Win went through Invoke-Command on PSSession with
quotas inherited from the local process, so the limits never bit.
Changes:
- Deploy-WinBuild2025.ps1 / Deploy-WinBuild2022.ps1: also set the
plugin-level MaxMemoryPerShellMB=2048, bump MaxProcessesPerShell
25→100, patch SharedSection 3rd field 768→4096 (reboot needed for
csrss to pick up, applied by post-install shutdown).
- Install-CIToolchain-WinBuild2025/2022.ps1: add Assert-Step for
plugin quota, MaxProcessesPerShell≥100, SharedSection≥4096; fail
hard if any are under threshold.
- docs/WINDOWS-TEMPLATE-SETUP.md: new "WinRM quotas e desktop heap"
section explaining the diagnosis + one-shot patch for legacy
templates; updated Step 3 validation row + troubleshooting entry.
- AGENTS.md: error #13 with symptom, root cause, and pointer to docs.
Reconcile docs with the end-to-end pipeline fixes:
- upload/download-artifact @v4 -> @v3 (Gitea GHES) in WORKFLOW-AUTHORING
and workflow-example.yml; add Common Mistakes rows (v4, action ref
form + DEFAULT_ACTIONS_URL, public action repo).
- BEST-PRACTICES / README / HOST-SETUP: guest credential must live in
the LocalSystem vault with a host-qualified username; document
Set-CIGuestCredential.ps1 / Test-CIGuestWinRM.ps1 and auth=ntlm.
- README / AGENTS / HOST-SETUP: production venv install is NON-editable
(LocalSystem); no CI workflow may install into it.
- HOST-SETUP: add DEFAULT_ACTIONS_URL=github + full-URL uses: + public
repo + @main requirements discovered during validation.
- Correct stale repo name local-ci-system -> local-ci-cd-system.
Update --cov-fail-under in lint workflow + docs (AGENTS, README, ARCHITECTURE). Note: A5 closeout reported 80.10%% so this run will likely fail until additional tests are added.
Configure system Python path once in runner/config.yaml (runner.envs) instead of probing N candidate locations in the workflow. Workflow now fails fast with a clear error if the env is unset or points to a missing file. Document the new env in AGENTS.md.
AGENTS.md: new 'Python development' section (venv paths, install/lint/typecheck/test commands, conventions, full PS->Python sub-command mapping, list of PS scripts that stay PowerShell, link to phase closeouts). README.md: expand 'Python orchestrator' with examples for all 10 sub-commands (wait-ready, vm new/remove/cleanup, build run, artifacts collect, monitor disk/runner, report job, job), local validation gates with 80% coverage gate, link to test_agents_errors.py. docs/ARCHITECTURE.md: new 'Python orchestrator (Phase A)' section with package layout tree, VmBackend Protocol contract, load_backend(config) factory, Phase C ESXi extension recipe, transport selection note, and CI validation gates.
Root cause: ssh-keygen -R prints 'Host X not found in F:\CI\State\known_hosts'
to stderr when the IP has no prior entry. In PS5.1 with ErrorActionPreference=Stop,
native stderr is converted to ErrorRecords before 2>$null can discard them,
making the command a terminating error caught by the job's catch block.
This caused all 4 concurrent burn-in jobs to fail after ~15s with:
Error: Host 192.168.79.XXX not found in F:\CI\State\known_hosts
Additionally, 4 concurrent SSH connections writing to the same known_hosts file
with no advisory locking risk file corruption under concurrent load.
Fix: set SshKnownHostsFile default to '' in Invoke-CIJob, Invoke-RemoteBuild,
and Get-BuildArtifacts. When KnownHostsFile='' _Transport.psm1 uses
StrictHostKeyChecking=no + UserKnownHostsFile=NUL (permissive mode), consistent
with Wait-VMReady and Prepare-LinuxBuild2404.ps1. Ephemeral CI VMs on a
local NAT network have no meaningful MITM risk.
Remove the ssh-keygen purge block from Invoke-CIJob.ps1 (no longer needed).
AGENTS.md: document PS5.1 native stderr / 2>$null interaction (error #12).
AGENTS.md (new):
- Environment reference for AI agents: host specs, PS 5.1 constraints,
dir structure, VM templates, transport layer, style rules
- Lessons #9 (snapshot cold state) and #10 (vmrun list vs getGuestIPAddress)
ARCHITECTURE.md:
- Add SSH/SCP transport section (Linux VMs)
- Update system diagram to include _Transport.psm1, SSH transport path
- Network layout: add SSH port 22 for Linux VMs
- Timeline: Linux readiness check phases documented
CI-FLOW.md:
- Prerequisites table: add Template VM (Linux), SSH key row
- Step 6 (readiness): document Windows vs Linux phases
- Step 7 (build) and Step 8 (artifacts): document Linux/SSH paths
- ci-report-ip guestinfo mechanism referenced
HOST-SETUP.md:
- Directory tree: add LinuxBuild2404 template dir and keys/
- Post-setup checklist: add step 5 (SSH key generation + Linux template)
- Fix duplicate step numbering (56, 67)
- Reference Paths table: add Linux template VMX, snapshot, SSH key
LINUX-TEMPLATE-SETUP.md:
- Rewritten as implementation record (no longer a plan)
- Fase A user-data: minimal (user+SSH key only, no packages/runcmd)
- All phases marked complete with actual paths and snapshot names
WINDOWS-TEMPLATE-SETUP.md: minor updates for consistency
TODO.md:
- Summary: update last-updated to 2026-05-11, Sprint 9 description
- Linux Build VM row: all 5 items complete
- Prossimi passi: Test Plan + §6.6 Tier-2 as next priorities
Simone
Simone