diff --git a/modules/ha/cluster-config.nix b/modules/ha/cluster-config.nix
index 76ad56b..0c7003b 100644
--- a/modules/ha/cluster-config.nix
+++ b/modules/ha/cluster-config.nix
@@ -21,7 +21,26 @@
 #   STONITH device during the testing phase.  Switch to resource-and-stonith
 #   once the fence_pve_ssh STONITH resource is active (see
 #   scripts/ha/cluster-enable-stonith.sh).
-{ lib, vars, ... }:
+#
+#   PATH wrapper: when the DRBD kernel module invokes the fence-peer handler
+#   via the UMH (User Mode Helper) mechanism it provides a minimal PATH that
+#   omits /run/current-system/sw/bin.  crm-fence-peer.sh calls cibadmin,
+#   crm_mon etc.; if those aren't found a pipeline in the script breaks with
+#   SIGPIPE.  A process killed by signal has WEXITSTATUS() == 0, so the kernel
+#   sees exit code 0 and logs "fence-peer helper broken, returned 0", looping
+#   forever.  The writeShellScript wrappers below prepend the NixOS sw path
+#   before exec-ing the real handler, giving it a working Pacemaker toolchain.
+{ lib, pkgs, vars, ... }:
+let
+  fencePeerWrapper = pkgs.writeShellScript "drbd-fence-peer" ''
+    export PATH="/run/current-system/sw/bin:/run/current-system/sw/sbin:$PATH"
+    exec /run/current-system/sw/lib/drbd/crm-fence-peer.sh "$@"
+  '';
+  unfencePeerWrapper = pkgs.writeShellScript "drbd-unfence-peer" ''
+    export PATH="/run/current-system/sw/bin:/run/current-system/sw/sbin:$PATH"
+    exec /run/current-system/sw/lib/drbd/crm-unfence-peer.sh "$@"
+  '';
+in
 {
   # Root SSH access — same key set as nixos user so all admin keys can reach root.
   users.users.root.openssh.authorizedKeys.keys = [
@@ -62,8 +81,8 @@
           fencing resource-only;
         }
         handlers {
-          fence-peer   "/run/current-system/sw/lib/drbd/crm-fence-peer.sh";
-          unfence-peer "/run/current-system/sw/lib/drbd/crm-unfence-peer.sh";
+          fence-peer   "${fencePeerWrapper}";
+          unfence-peer "${unfencePeerWrapper}";
         }
       }