Runpod•7mo ago

ComfyUI server (127.0.0.1:8188) not reachable after multiple retries

My serverless endpoint worker that ran fine for weeks suddenly fails every request

{
  "delayTime": 7929,
  "error": "ComfyUI server (127.0.0.1:8188) not reachable after multiple retries.",
  "executionTime": 26801,
  "id": "0462b632-56d1-46f5-bcad-fae3a2cb82d0-e2",
  "status": "FAILED",
  "workerId": "65qj1ggc3jj97e"
}

{
  "delayTime": 7929,
  "error": "ComfyUI server (127.0.0.1:8188) not reachable after multiple retries.",
  "executionTime": 26801,
  "id": "0462b632-56d1-46f5-bcad-fae3a2cb82d0-e2",
  "status": "FAILED",
  "workerId": "65qj1ggc3jj97e"
}

Also provided the text logs from the worker

logs-2.txt6.12KB

© 2026 Hedgehog Software, LLC

More

Communities Docs About Terms Privacy

{
  "delayTime": 7929,
  "error": "ComfyUI server (127.0.0.1:8188) not reachable after multiple retries.",
  "executionTime": 26801,
  "id": "0462b632-56d1-46f5-bcad-fae3a2cb82d0-e2",
  "status": "FAILED",
  "workerId": "65qj1ggc3jj97e"
}

{
  "delayTime": 7929,
  "error": "ComfyUI server (127.0.0.1:8188) not reachable after multiple retries.",
  "executionTime": 26801,
  "id": "0462b632-56d1-46f5-bcad-fae3a2cb82d0-e2",
  "status": "FAILED",
  "workerId": "65qj1ggc3jj97e"
}

Jason

J

Jason•6/23/25, 11:41 AM

Which cuda version did you use?

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:41 AM

12.8

Jason

J

Jason•6/23/25, 11:41 AM

thats because comfyui failed to launch, thats why the request failed

Jason

J

Jason•6/23/25, 11:41 AM

hmm can you filter only 12.8 cuda in your endpoint

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:42 AM

It is already set to 12.8 in runpod endpoint settings

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:42 AM

But it's working already

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:42 AM

Terminated the worker and pulled the container again

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:42 AM

Strangely that seems to work

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:42 AM

Yesterday I terminated a couple times as well which didn't work

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:43 AM

But this morning I did it again one last time and that did the trick

gokuvonlange

Ggokuvonlange Terminated the worker and pulled the container again

Jason

J

Jason•6/23/25, 11:43 AM

Oh so you had set cuda version filter in the endpoint
and the fix was to terminate the worker when it happened?

Jason

J

Jason•6/23/25, 11:43 AM

just curious too which gpu are you using?

gokuvonlange

Ggokuvonlange My serverless endpoint worker that ran fine for weeks suddenly fails every reque...

Poddy

P

PoddyAPP•6/23/25, 11:43 AM

@gokuvonlange

Escalated To Zendesk

The thread has been escalated to Zendesk!

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:44 AM

It's been like this for a month. My endpoint worked fine for a full month

Screenshot_2025-06-23_at_13.43.44.png

Jason

J

Jason•6/23/25, 11:44 AM

Might have to report this to a ticket too

Jason

JJason just curious too which gpu are you using?

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:44 AM

80GB Pro (H100's)

Jason

J

Jason•6/23/25, 11:44 AM

any specific region that you're using?

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:44 AM

Right now I'm on CA-MTL-1 which works for me

Jason

JJason any specific region that you're using?

Jason

J

Jason•6/23/25, 11:44 AM

Who knows other user might be experiencing this, i'll be able to know

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:44 AM

I don't remember the problematic regions before I terminated them

Jason

J

Jason•6/23/25, 11:45 AM

np

gokuvonlange

G

gokuvonlangeOP•6/23/25, 11:46 AM

Thanks for support

Jason

J

Jason•6/23/25, 11:54 AM

Your welcome!