fix: wait for SSH readiness after VM provisioning

Freshly recreated VMs can take a few minutes before cloud-init users and SSH are available. Retry SSH authentication in the bootstrap controller before failing so rebuild/bootstrap does not abort immediately on new hosts.

nixos/kubeadm/bootstrap/controller.py

@@ -121,6 +121,8 @@ class Controller:
         self.fast_mode = self.env.get("FAST_MODE", "1")
         self.skip_rebuild = self.env.get("SKIP_REBUILD", "0") == "1"
         self.force_reinit = True
+        self.ssh_ready_retries = int(self.env.get("SSH_READY_RETRIES", "20"))
+        self.ssh_ready_delay = int(self.env.get("SSH_READY_DELAY_SEC", "15"))
 
     def log(self, msg):
         print(f"==> {msg}")
@@ -130,12 +132,19 @@ class Controller:
         return run_local(full, check=check, capture=True)
 
     def detect_user(self, ip):
-        for user in self.ssh_candidates:
-            proc = self._ssh(user, ip, "true", check=False)
-            if proc.returncode == 0:
-                self.active_ssh_user = user
-                self.log(f"Using SSH user '{user}' for {ip}")
-                return
+        for attempt in range(1, self.ssh_ready_retries + 1):
+            for user in self.ssh_candidates:
+                proc = self._ssh(user, ip, "true", check=False)
+                if proc.returncode == 0:
+                    self.active_ssh_user = user
+                    self.log(f"Using SSH user '{user}' for {ip}")
+                    return
+            if attempt < self.ssh_ready_retries:
+                self.log(
+                    f"SSH not ready on {ip} yet; retrying in {self.ssh_ready_delay}s "
+                    f"({attempt}/{self.ssh_ready_retries})"
+                )
+                time.sleep(self.ssh_ready_delay)
         raise RuntimeError(f"Unable to authenticate to {ip} with users: {', '.join(self.ssh_candidates)}")
 
     def remote(self, ip, cmd, check=True):