main/server: Add --debug-hang (#14371)

Add an option to debug a hang with ctrl-C, dumping the backtraces to
see where its stuck or slow.

comfy/cli_args.py

@@ -166,6 +166,8 @@ class PerformanceFeature(enum.Enum):
 
 parser.add_argument("--fast", nargs="*", type=PerformanceFeature, help="Enable some untested and potentially quality deteriorating optimizations. This is used to test new features so using it might crash your comfyui. --fast with no arguments enables everything. You can pass a list specific optimizations if you only want to enable specific ones. Current valid optimizations: {}".format(" ".join(map(lambda c: c.value, PerformanceFeature))))
 
+parser.add_argument("--debug-hang", action="store_true", help="Enable stack trace dumps on Ctrl-C for debugging hangs.")
+
 parser.add_argument("--disable-pinned-memory", action="store_true", help="Disable pinned memory use.")
 
 parser.add_argument("--mmap-torch-files", action="store_true", help="Use mmap when loading ckpt/pt files.")

main.py

@@ -26,6 +26,7 @@ import utils.extra_config
 from utils.mime_types import init_mime_types
 import faulthandler
 import logging
+import signal
 import sys
 from comfy_execution.progress import get_progress_state
 from comfy_execution.utils import get_executing_context
@@ -37,7 +38,19 @@ if __name__ == "__main__":
     os.environ['HF_HUB_DISABLE_TELEMETRY'] = '1'
     os.environ['DO_NOT_TRACK'] = '1'
 
-faulthandler.enable(file=sys.stderr, all_threads=False)
+faulthandler.enable(file=sys.stderr, all_threads=args.debug_hang)
+if __name__ == "__main__" and args.debug_hang:
+    dumping_traceback = False
+
+    def dump_traceback_on_sigint(signum, frame):
+        global dumping_traceback
+        if dumping_traceback:
+            raise KeyboardInterrupt
+        dumping_traceback = True
+        faulthandler.dump_traceback(file=sys.stderr, all_threads=True)
+        raise KeyboardInterrupt
+
+    signal.signal(signal.SIGINT, dump_traceback_on_sigint)
 
 import comfy_aimdo.control
 

server.py

@@ -1253,6 +1253,15 @@ class PromptServer():
 
         if verbose:
             logging.info("Starting server\n")
+            if args.debug_hang:
+                logging.info(
+                    f"{'-' * 80}\n"
+                    "ComfyUI has been started in debug-hang mode. Run your workflow as normal up to\n"
+                    "the point of the hang or freeze, then use ctrl-C in the cmd or controlling\n"
+                    "terminal to dump the python backtraces for debugging. Please attach the extra\n"
+                    "debug info to your bug report.\n"
+                    f"{'-' * 80}"
+                )
         for addr in addresses:
             address = addr[0]
             port = addr[1]