pollux (192.168.178.165) wedged at the network level during an
end-to-end install test today — mkinitcpio on a 4 GiB smoke VM +
the cached 1.5 GB ISO + a busy runner container pushed the host into
OOM, taking pveproxy and the SSH path down with it. Recovered by
physical reset.
Smoke VM now defaults to 8192 MiB / 2 vCPU, configurable via
PVE_TEST_VM_MEMORY / PVE_TEST_VM_CORES. Host has 64 GiB, so one
smoke VM at 8 GiB is well within headroom.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The delete branch required Datastore.Allocate (or was hitting a
privilege-separated token ACL edge case) and produced 403s on re-runs
against the same commit SHA. Since the ISO bytes are reproducible for
a given SHA — furtka-<sha>.iso is content-addressed — we can just
reuse whatever is already in PVE storage instead of cycling it.
Fixes the "runs-on-same-sha" re-dispatch case without needing any extra
PVE permission, and shaves ~2 min off repeated smoke runs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
After build-iso, a new smoke-vm job uploads the freshly built ISO to
the test Proxmox at 192.168.178.165 via PVE API token, boots it in a
fresh VM (VMID range 9000-9099, MAC derived from commit SHA so the
runner can find the DHCP IP by scanning the LAN), and curls :5000 to
confirm the webinstaller answers HTTP 200. Last 5 smoke VMs + their
ISOs are kept for post-mortem; older ones are purged. continue-on-error
on the smoke job so a VM-side flake doesn't mark the ISO build red.
Shortens the feedback loop on ISO regressions from "next manual VM
test session" (days) to "next push" (minutes) — the 2026-04-15/16 VM
sessions each found real boot-time bugs that unit tests missed.
Docs at docs/smoke-vm.md. Requires Forgejo secrets PVE_TEST_HOST and
PVE_TEST_TOKEN (dedicated smoke@pve!ci PVE token, privilege-separated).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
