build: bump RQ to latest (#31952)

perf fix for busy pubsub thread

frappe/utils/background_jobs.py

@@ -355,26 +355,6 @@ class FrappeWorker(Worker):
 
 		Thread(target=start_scheduler, daemon=True).start()
 
-	def subscribe(self):
-		"""Subscribe to this worker's channel"""
-		# This function is overridden to increase the timeout of pubsub thread. Default is 0.2
-		# second which is too frequent for us, this change sets it to 2s which is 10x the default.
-		# ref: https://github.com/frappe/caffeine/issues/46
-
-		# The pubsub thread is responsible for handling three commands from master process:
-		# 1. shutdown
-		# 2. stop current job
-		# 3. Kill forked horse (~ force stop the job)
-
-		# Impact of increasing timeout: shutdown might have up to 2s before background thread
-		# times out and is joined with main thread. Ideally, we should not have to do this at all.
-		# But the code that handles blocking socket behaviour is deep inside redis-py/hiredis.
-
-		self.log.info("Subscribing to channel %s", self.pubsub_channel_name)
-		self.pubsub = self.connection.pubsub()
-		self.pubsub.subscribe(**{self.pubsub_channel_name: self.handle_payload})
-		self.pubsub_thread = self.pubsub.run_in_thread(sleep_time=2, daemon=True)
-
 
 class FrappeWorkerNoFork(FrappeWorker):
 	def __init__(self, *args, **kwargs):

pyproject.toml

@@ -66,7 +66,7 @@ dependencies = [
     # We depend on internal attributes of RQ.
     # Do NOT add loose requirements on RQ versions.
     # Audit the code changes w.r.t. background_jobs.py before updating.
-    "rq==2.1.0",
+    "rq==2.2.0",
     "rsa>=4.1",
     "semantic-version~=2.10.0",
     "sentry-sdk~=1.37.1",