Facing a Serverless run error that is not encountered in ComfyUI web UI.

[white edge]

Hi all, my Serverless Endpoint has the same custom nodes and models as my persistent pod. However, it encountered the following error when I tried to run my custom workflow on the Serverless Endpoint. Please see logs attached😢 and help me to fix.

[slate mountain]

2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | pytorch version: 2.6.0+cu124\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Total VRAM 24210 MB, total RAM 127981 MB\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Could not load ImpactPack nodes Could not find ImpactPack nodes\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Could not find VideoHelperSuite nodes\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Could not find IPAdapter nodes\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Could not find AnimateDiff nodes\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Could not find AdvancedControlNet nodes\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Could not find ControlNetPreprocessors nodes\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Could not find efficiency nodes\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | Adding /comfyui/custom_nodes to sys.path\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 |   warnings.warn(f"Importing from {__name__} is deprecated, please import via timm.layers", FutureWarning)\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | /usr/local/lib/python3.10/dist-packages/timm/models/layers/__init__.py:48: FutureWarning: Importing from timm.models.layers is deprecated, please import via timm.layers\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | ------------------------------------------\n
2025-04-18  15:39:45.581 | info | vve30u1f4uh6x3 | ** For help, please see the wiki at https://github.com/Suzie1/ComfyUI_Comfyroll_CustomNodes/wiki\n
2025-04-18  15:39:45.580 | info | vve30u1f4uh6x3 | ** For changes, please see patch notes at https://github.com/Suzie1/ComfyUI_Comfyroll_CustomNodes/blob/main/Patch_Notes.md\n

this error?, just wondering too, are you adding yourcustom nodes inside network storage or inside your image (via dockerfile / from your builds step)

[white edge]

I'm adding custom nodes to my Docker image via snapshot as instructed in the following repo.

https://github.com/blib-la/runpod-worker-comfy

[slate mountain]

I see maybe try debugging if the custom nodes and pip packages is correctly installed

By using 1 active worker and then connect to the worker via web terminal

[white edge]

ok I'll try doing that

[white edge]

my prompt did finish execution after 3mins+ but my request showed as FAILED after 2mins.

[slate mountain]

No errors after that?

[slate mountain]

What's the job result? Or response from /status from the failed job

[white edge]

hey i managed to solve it, thanks

[slate mountain]

Oh just wondering what's the problem?

[white edge]

the issue is due to the ComfyUI worker's polling timeout, which defaults to about 2 minutes, causing the request to fail before my 3-minute workflow completes.

[white edge]

I've set COMFY_POLLING_MAX_RETRIES to 1000 and it works

[slate mountain]

I see