feat: run update-core on cluster init and restore

core/imageroot/var/lib/nethserver/cluster/actions/create-cluster/00validate_cluster

@@ -9,6 +9,10 @@ import json
 import sys
 import agent
 
+agent.set_weight("00validate_cluster", 0)
+agent.set_weight("10validate_network", 0)
+agent.set_weight("40update_cluster", 3)
+
 request = json.load(sys.stdin)
 
 rdb = agent.redis_connect()

core/imageroot/var/lib/nethserver/cluster/actions/create-cluster/40update_cluster

@@ -0,0 +1,20 @@
+#!/usr/bin/env python3
+
+#
+# Copyright (C) 2026 Nethesis S.r.l.
+# SPDX-License-Identifier: GPL-3.0-or-later
+#
+
+import sys
+import agent
+import os
+
+update_core_response = agent.tasks.run(
+    agent_id='cluster',
+    action='update-core',
+    data={'nodes': [1]}, # always 1 in a new cluster
+    endpoint="redis://cluster-leader",
+)
+if update_core_response['exit_code'] != 0:
+    print(agent.SD_ERR + "update-core failed", file=sys.stderr)
+    sys.exit(1)

core/imageroot/var/lib/nethserver/cluster/actions/create-cluster/50update

@@ -98,6 +98,8 @@ agent.assert_exp(add1_module_failures == 0)
 # Update completed successfully, enable the phonehome.timer
 agent.run_helper(*'systemctl enable --now phonehome.timer'.split(' ')).check_returncode()
 
+agent.unset_env("FIRST_CONFIGURATION")
+
 json.dump({
     "ip_address": ip_address,
 }, fp=sys.stdout)

core/imageroot/var/lib/nethserver/cluster/actions/get-cluster-status/50read

@@ -70,8 +70,9 @@ leader_id = int(rdb.hget("cluster/environment", "NODE_ID"))
 leader_endpoint = rdb.hget(f"node/{leader_id}/vpn", "endpoint") or "leader.invalid:55820"
 local_id = int(os.environ['NODE_ID'])
 local_public_key = rdb.hget(f"node/{local_id}/vpn", "public_key")
+initialized = os.getenv("FIRST_CONFIGURATION") != "1"
 
-ret = { 'initialized': False, 'leader': False, 'nodes': [], 'leader_url': '', 'default_password': False, 'ui_name': cluster_ui_name, }
+ret = { 'initialized': initialized, 'leader': False, 'nodes': [], 'leader_url': '', 'default_password': False, 'ui_name': cluster_ui_name, }
 vpn = {}
 
 default_password = hashlib.sha256(b'Nethesis,1234').hexdigest()
@@ -82,7 +83,6 @@ for p in admin['passwords']:
 
 network = rdb.get('cluster/network')
 if network is not None:
-    ret["initialized"] = True
     # The leader must have a public endpoint:
     leader_hostname, _ = leader_endpoint.rsplit(":", 1)
     ret['leader_url'] = f'https://{leader_hostname}/cluster-admin/'

core/imageroot/var/lib/nethserver/node/install-finalize.sh

@@ -110,6 +110,7 @@ echo "Grant initial permissions:"
 runagent python3 <<'EOF'
 import agent
 import cluster.grants
+agent.set_env("FIRST_CONFIGURATION", "1")
 rdb = agent.redis_connect(privileged=True)
 cluster.grants.grant(rdb, action_clause="*",      to_clause="owner",  on_clause='cluster')
 cluster.grants.grant(rdb, action_clause="list-*", to_clause="reader", on_clause='cluster')

core/ui/src/App.vue

@@ -19,7 +19,6 @@
 import ShellHeader from "@/components/shell/ShellHeader";
 import SideMenu from "@/components/shell/SideMenu";
 import MobileSideMenu from "@/components/shell/MobileSideMenu";
-import axios from "axios";
 import WebSocketService from "@/mixins/websocket";
 import { mapState, mapActions } from "vuex";
 import to from "await-to-js";
@@ -187,15 +186,39 @@ export default {
         }
       });
     },
+    isRetryableAgentRequest(error) {
+      return (
+        // The cluster agent may be temporarily unavailable while restarting.
+        error.config?.method?.toLowerCase() === "post" &&
+        error.response?.status == 404 &&
+        error.response?.data?.message === "client idle check failed" &&
+        error.config?.url?.endsWith("/cluster/tasks")
+      );
+    },
     configureAxiosInterceptors() {
       const context = this;
+      const maxAgentRestartRetries = 5;
 
       // response interceptor
-      axios.interceptors.response.use(
+      this.axios.interceptors.response.use(
         function (response) {
           return response;
         },
-        function (error) {
+        async function (error) {
+          if (context.isRetryableAgentRequest(error)) {
+            const retryConfig = error.config;
+            retryConfig.__agentRestartRetryCount =
+              (retryConfig.__agentRestartRetryCount || 0) + 1;
+
+            if (
+              retryConfig.__agentRestartRetryCount <=
+              maxAgentRestartRetries
+            ) {
+              await new Promise((resolve) => setTimeout(resolve, 2000));
+              return context.axios.request(retryConfig);
+            }
+          }
+
           console.error(error);
 
           // print specific error message, if available