Add stability debugging for beefy lockups.

- Add netconsole receiver on zippy to capture kernel messages
- Configure beefy as netconsole sender to zippy (192.168.1.2)
- Enable kdump with 256M reserved memory for crash analysis
- Add lockup detectors (softlockup_panic, hung_task_panic, nmi_watchdog)
- Add consoleblank=300 for greeter display sleep
- Persist crash dumps and add analysis tools (crash, makedumpfile)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

common/netconsole-receiver.nix

@@ -0,0 +1,32 @@
+{
+  config,
+  lib,
+  pkgs,
+  ...
+}:
+{
+  options.services.netconsoleReceiver = {
+    enable = lib.mkEnableOption "netconsole UDP receiver";
+    port = lib.mkOption {
+      type = lib.types.port;
+      default = 6666;
+      description = "UDP port to listen on for netconsole messages";
+    };
+  };
+
+  config = lib.mkIf config.services.netconsoleReceiver.enable {
+    systemd.services.netconsole-receiver = {
+      description = "Netconsole UDP receiver";
+      wantedBy = [ "multi-user.target" ];
+      after = [ "network.target" ];
+      serviceConfig = {
+        ExecStart = "${pkgs.socat}/bin/socat -u UDP-LISTEN:${toString config.services.netconsoleReceiver.port},fork STDOUT";
+        StandardOutput = "journal";
+        StandardError = "journal";
+        SyslogIdentifier = "netconsole";
+        Restart = "always";
+        RestartSec = "5s";
+      };
+    };
+  };
+}

hosts/beefy/default.nix

@@ -1,4 +1,4 @@
-{ pkgs, inputs, ... }:
+{ pkgs, inputs, config, ... }:
 {
   imports = [
     ../../common/encrypted-btrfs-layout.nix
@@ -19,6 +19,38 @@
   networking.cluster.primaryInterface = "enp1s0";
   services.tailscaleAutoconnect.authkey = "tskey-auth-k79UsDTw2v11CNTRL-oYqji35BE9c7CqM89Dzs9cBF14PmqYsi";
 
+  # Console blanking after 5 minutes (for greeter display sleep)
+  # NMI watchdog for hardlockup detection
+  boot.kernelParams = [ "consoleblank=300" "nmi_watchdog=1" ];
+
+  # Netconsole - stream kernel messages to zippy (192.168.1.2)
+  boot.kernelModules = [ "netconsole" ];
+  boot.extraModprobeConfig = ''
+    options netconsole netconsole=@/enp1s0,6666@192.168.1.2/c0:3f:d5:62:55:bb
+  '';
+
+  # Kdump for kernel crash analysis
+  boot.crashDump = {
+    enable = true;
+    reservedMemory = "256M";
+  };
+
+  # Lockup detectors - panic on detection so kdump captures state
+  boot.kernel.sysctl = {
     # Enable all SysRq functions for debugging hangs
-  boot.kernel.sysctl."kernel.sysrq" = 1;
+    "kernel.sysrq" = 1;
+    # Panic on soft lockup (CPU not scheduling for >20s)
+    "kernel.softlockup_panic" = 1;
+    # Panic on hung tasks (blocked >120s)
+    "kernel.hung_task_panic" = 1;
+    "kernel.hung_task_timeout_secs" = 120;
+  };
+
+  # Persist crash dumps
+  environment.persistence.${config.custom.impermanence.persistPath}.directories = [
+    "/var/crash"
+  ];
+
+  # Crash analysis tools
+  environment.systemPackages = with pkgs; [ crash makedumpfile ];
 }

hosts/zippy/default.nix

@@ -5,9 +5,13 @@
     ../../common/global
     ../../common/cluster-member.nix  # Consul + storage clients
     ../../common/nomad-worker.nix     # Nomad client (runs jobs)
+    ../../common/netconsole-receiver.nix
     ./hardware.nix
   ];
 
+  # Receive kernel messages from beefy via netconsole
+  services.netconsoleReceiver.enable = true;
+
   diskLayout = {
     mainDiskDevice = "/dev/disk/by-id/ata-KINGSTON_SKC600MS1024G_50026B7785AE0A92";
     #keyDiskDevice = "/dev/disk/by-id/usb-Intenso_Micro_Line_22080777660702-0:0";