fix(host): bring up VMware host stack at boot via native systemd units

Build VMs stopped starting after a reboot into a new kernel: vmmon/vmnet were left unloaded and vmnet1/vmnet8 (DHCP/NAT) never came up, so vmware-vmx died (status=237/KEYRING) and clones got no IP. Root cause: the LSB->systemd sysv-generated vmware.service intermittently fails to generate at early boot ("systemd-sysv-generator: Cannot find unit vmware.service"), and the existing module-rebuild guard only builds the modules -- it never loads them or starts vmnet networking. Add a native vmware-host.service that runs `/etc/init.d/vmware start` (loads vmmon/vmci + brings up vmnet), ordered after the rebuild guard, and mask the flaky sysv vmware.service so the native unit is the sole owner. Provisioned by scripts/install-vmware-host-services.sh and wired into setup-host-linux.sh (new step 8/9). VMware ships no native unit and no kernel postinst hook, so this native wrapper around VMware's own tools is the reliable boot path. Verified: linked clone boots and gets a vmnet8 DHCP IP.
2026-07-19 19:47:16 +02:00
parent 92bea69f19
commit 19186981a5
7 changed files with 234 additions and 14 deletions
@@ -84,3 +84,50 @@ sudo systemctl daemon-reload

 A questo punto il fallback è ri-attivare i task su Windows con
 `scripts/Register-CIScheduledTasks.ps1`.
+
+---
+
+## Servizi VMware host (prerequisito avvio VM)
+
+Distinti dai timer periodici sopra: due unit **native** che portano su lo stack
+host di VMware Workstation a ogni boot. Servono perché VMware su Linux **non**
+fornisce unit systemd native, **non** registra `vmmon`/`vmnet` con DKMS e **non**
+installa hook di post-install del kernel.
+
+| unit | ruolo |
+| ---- | ----- |
+| `vmware-modules-rebuild.service` | ricostruisce `vmmon`/`vmnet` via `vmware-modconfig` se il `.ko` per il kernel in esecuzione manca (guard *build-if-missing*). Ordinata `Before=vmware-host.service`. Script: `/usr/local/sbin/vmware-modules-rebuild.sh`. |
+| `vmware-host.service` | esegue `/etc/init.d/vmware start` → carica `vmmon`/`vmci` e avvia la rete `vmnet1`/`vmnet8` (DHCP/NAT). Ordinata `After=vmware-modules-rebuild.service`. |
+
+Il sysv `vmware.service` (LSB) viene **mascherato**: si è dimostrato inaffidabile
+da generare al boot (`systemd-sysv-generator: Cannot find unit vmware.service`),
+lasciando `vmmon` non caricato e la rete VM giù → le build VM non partono.
+`vmware-host.service` ne diventa l'unico owner.
+
+### Installazione
+
+```bash
+sudo ./scripts/install-vmware-host-services.sh   # installa + enable + start (idempotente)
+```
+
+Installato anche automaticamente da `setup-host-linux.sh` (Step 8/9).
+
+### Verifica
+
+```bash
+systemctl is-enabled vmware-host.service vmware-modules-rebuild.service
+lsmod | grep -E 'vmmon|vmnet'
+ip -o link show | grep vmnet          # attese: vmnet1, vmnet8
+```
+
+### Recovery (stack giù, VM non partono)
+
+```bash
+sudo systemctl start vmware-host.service
+```
+
+### Rollback
+
+```bash
+sudo ./scripts/install-vmware-host-services.sh --uninstall
+```
@@ -0,0 +1,21 @@
+[Unit]
+Description=Start VMware Workstation host stack (vmmon/vmci + vmnet networking)
+Documentation=man:vmware(1)
+# Native replacement for the flaky LSB->systemd sysv-generated vmware.service,
+# which intermittently fails to generate at early boot ("Cannot find unit
+# vmware.service"), leaving vmmon unloaded and vmnet DHCP/NAT down so VMs
+# cannot start. The rebuild guard ensures the modules are built first; this
+# unit loads them and brings up vmnet1/vmnet8 via VMware's own init script.
+Wants=vmware-modules-rebuild.service
+After=vmware-modules-rebuild.service network-pre.target local-fs.target
+ConditionPathExists=/etc/init.d/vmware
+
+[Service]
+Type=oneshot
+RemainAfterExit=yes
+ExecStart=/etc/init.d/vmware start
+ExecStop=/etc/init.d/vmware stop
+TimeoutStartSec=120
+
+[Install]
+WantedBy=multi-user.target
@@ -0,0 +1,17 @@
+[Unit]
+Description=Rebuild VMware host modules (vmmon/vmnet) for running kernel if missing
+Documentation=man:vmware-modconfig(1)
+# VMware modules are not DKMS-managed; ensure they exist before vmware-host starts.
+After=local-fs.target
+Before=vmware-host.service
+ConditionPathExists=/usr/bin/vmware-modconfig
+
+[Service]
+Type=oneshot
+RemainAfterExit=yes
+ExecStart=/usr/local/sbin/vmware-modules-rebuild.sh
+# Building modules can take a little while on a cold kernel.
+TimeoutStartSec=600
+
+[Install]
+WantedBy=multi-user.target
@@ -0,0 +1,28 @@
+#!/bin/sh
+# Rebuild VMware host modules (vmmon/vmnet) for the running kernel if absent.
+# VMware Workstation does not register these with DKMS, so a kernel upgrade
+# leaves them unbuilt until the GUI or vmware-modconfig is run. This runs at
+# boot, before vmware-host.service, and self-heals using VMware's own builder.
+set -eu
+
+KVER="$(uname -r)"
+
+if /sbin/modinfo vmmon >/dev/null 2>&1 && /sbin/modinfo vmnet >/dev/null 2>&1; then
+    # Module objects already present for the running kernel — nothing to do.
+    exit 0
+fi
+
+logger -t vmware-modules-rebuild "vmmon/vmnet missing for kernel ${KVER} — rebuilding via vmware-modconfig"
+
+if ! command -v vmware-modconfig >/dev/null 2>&1; then
+    logger -t vmware-modules-rebuild "ERROR: vmware-modconfig not found"
+    exit 1
+fi
+
+# vmware-modconfig builds against the running kernel's headers.
+if /usr/bin/vmware-modconfig --console --install-all; then
+    logger -t vmware-modules-rebuild "rebuild complete for kernel ${KVER}"
+else
+    logger -t vmware-modules-rebuild "ERROR: vmware-modconfig failed (headers for ${KVER} installed?)"
+    exit 1
+fi