From 37c9ac79ed238f0295a4af7794d54f0c1ec246a9 Mon Sep 17 00:00:00 2001
From: Graham McIntire <graham@mcintire.me>
Date: Fri, 25 Jul 2025 12:54:13 -0500
Subject: [PATCH] Fix Process.alive? crash with remote PIDs
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The cleanup_stale_registrations function was calling Process.alive?
on remote PIDs, which causes an ArgumentError. Fixed by:

- Using Process.alive? only for local PIDs
- Using :rpc.call for remote PID liveness checks
- Adding proper error handling for RPC failures
- Cleaning up registrations for disconnected nodes

This prevents GenServer crashes when checking leader status.

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

Co-Authored-By: Claude <noreply@anthropic.com>
---
 lib/aprsme/cluster/leader_election.ex | 43 +++++++++++++++++++++++----
 1 file changed, 37 insertions(+), 6 deletions(-)

diff --git a/lib/aprsme/cluster/leader_election.ex b/lib/aprsme/cluster/leader_election.ex
index 19cfa73..a1796fa 100644
--- a/lib/aprsme/cluster/leader_election.ex
+++ b/lib/aprsme/cluster/leader_election.ex
@@ -136,13 +136,44 @@ defmodule Aprsme.Cluster.LeaderElection do
         :ok
 
       pid when is_pid(pid) ->
-        # Check if the registered process is alive and on a connected node
-        if Process.alive?(pid) and node(pid) in [node() | Node.list()] do
-          # Process is alive and on a connected node - leave it alone
-          :ok
+        pid_node = node(pid)
+        connected_nodes = [node() | Node.list()]
+
+        # Check if the PID's node is still connected
+        if pid_node in connected_nodes do
+          # Node is connected, try to check if process is alive
+          try do
+            if pid_node == node() do
+              # Local PID - use Process.alive?
+              if Process.alive?(pid) do
+                :ok
+              else
+                Logger.info("Cleaning up stale leader registration for dead local process #{inspect(pid)}")
+                :global.unregister_name(@election_key)
+              end
+            else
+              # Remote PID - use RPC to check if alive
+              case :rpc.call(pid_node, Process, :alive?, [pid]) do
+                true ->
+                  :ok
+
+                false ->
+                  Logger.info("Cleaning up stale leader registration for dead remote process #{inspect(pid)}")
+                  :global.unregister_name(@election_key)
+
+                {:badrpc, _reason} ->
+                  Logger.info("Cleaning up stale leader registration for unreachable process #{inspect(pid)}")
+                  :global.unregister_name(@election_key)
+              end
+            end
+          rescue
+            _error ->
+              Logger.info("Cleaning up stale leader registration for problematic process #{inspect(pid)}")
+              :global.unregister_name(@election_key)
+          end
         else
-          # Process is dead or on a disconnected node - unregister it
-          Logger.info("Cleaning up stale leader registration for dead/disconnected process #{inspect(pid)}")
+          # Node is disconnected - clean up the registration
+          Logger.info("Cleaning up stale leader registration for disconnected node #{pid_node}")
           :global.unregister_name(@election_key)
         end
     end