Load WAL checkpointer outside Zeitwerk reload; harden fork stop

Keep the module under lib/rails_ext, use before_worker_boot, shorten the
SQLite busy timeout under stop's join, and PASSIVE-checkpoint on exit for
short-lived writers.

Co-authored-by: Thomas Klemm <github@tklemm.eu>

config/initializers/sqlite_wal_checkpoint.rb

@@ -1,5 +1,8 @@
-# Start a checkpoint contender in every non-test process. Puma and Resque pool
-# stop before fork and start again in the child so the flock is never inherited.
+# Loaded from lib/rails_ext (autoload_lib ignore list) so reloads do not orphan
+# the contender thread. Non-test processes start here; Puma/Resque stop before
+# fork and start again in the child.
+require Rails.root.join("lib/rails_ext/sqlite_wal_checkpoint")
+
 Rails.application.config.after_initialize do
   SqliteWalCheckpoint.start unless Rails.env.test?
 end

config/puma.rb

@@ -54,7 +54,7 @@ Membership.disconnect_all
 # do not inherit the flock; each worker starts its own contender. Single-process
 # mode never forks, so the initializer's contender keeps running.
 before_fork { SqliteWalCheckpoint.stop }
-on_worker_boot { SqliteWalCheckpoint.start }
+before_worker_boot { SqliteWalCheckpoint.start }
 
 Signal.trap :SIGPROF do
   Thread.list.each do |t|

lib/rails_ext/sqlite_wal_checkpoint.rb

@@ -21,6 +21,7 @@ module SqliteWalCheckpoint
         return if @thread&.alive?
 
         @stop = false
+        install_exit_checkpoint
         @thread = Thread.new { run(interval) }
         @thread.report_on_exception = false
       end
@@ -33,6 +34,7 @@ module SqliteWalCheckpoint
     def stop
       @stop = true
       thread = @mutex&.synchronize { @thread }
+      # Join longer than busy_handler_timeout so before_fork does not race a PRAGMA.
       thread&.join(2)
       @mutex&.synchronize { @thread = nil if @thread && !@thread.alive? }
     end
@@ -63,6 +65,7 @@ module SqliteWalCheckpoint
       @lock_path = nil
       @database_path_override = nil
       @stop = false
+      @exit_checkpoint_installed = false
     end
 
     private
@@ -113,7 +116,8 @@ module SqliteWalCheckpoint
 
         result = nil
         SQLite3::Database.new(path) do |database|
-          database.busy_handler_timeout = 5_000
+          # Keep below stop's join timeout so before_fork can finish cleanly.
+          database.busy_handler_timeout = 1_000
           result = yield database
         end
         result
@@ -148,5 +152,14 @@ module SqliteWalCheckpoint
       ensure
         @lock_file = nil
       end
+
+      # Best-effort PASSIVE for short-lived console/rake writers that exit before
+      # the contender acquires the flock. Does not touch lock ownership.
+      def install_exit_checkpoint
+        return if @exit_checkpoint_installed
+
+        @exit_checkpoint_installed = true
+        at_exit { checkpoint rescue nil }
+      end
   end
 end

test/lib/sqlite_wal_checkpoint_test.rb

@@ -35,6 +35,24 @@ class SqliteWalCheckpointTest < ActiveSupport::TestCase
     assert_empty checkpoint_threads
   end
 
+  test "stop before start again mimics a fork-safe worker boot" do
+    db_path = build_wal_database(rows: 10)
+    SqliteWalCheckpoint.database_path_override = db_path
+
+    first = SqliteWalCheckpoint.start(interval: 0.05, enabled: true)
+    wait_until { first.name == "sqlite-wal-checkpoint" }
+    SqliteWalCheckpoint.stop
+    assert_not first.alive?
+
+    second = SqliteWalCheckpoint.start(interval: 0.05, enabled: true)
+    wait_until { second.name == "sqlite-wal-checkpoint" }
+    assert second.alive?
+
+    SqliteWalCheckpoint.stop
+    assert_not second.alive?
+    assert_equal :checkpointed, SqliteWalCheckpoint.tick
+  end
+
   test "tick checkpoints through the elected lock holder" do
     db_path = build_wal_database(rows: 50)
     SqliteWalCheckpoint.database_path_override = db_path