You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
Occasionally for long-running workflows triggered via API, the websocket connection gets lost.
To Reproduce
Difficult to reproduce because it only happens occasionally, with long running workflows in particular.
Expected behavior
I would expect much more durable performance from comfy run --workflow. This command also seems fairly undocumented so would be nice to know how mature this feature is.
Nice to have
This issue is well documented in our repo here.
Additional context
I work at Modal, a serverless compute / gpu platform. We have a lot of inbound customers searching for "comfyui api" that land on our ComfyUI example, which uses comfy-cli. Even before comfy-cli, we've anecdotally had many customers run into this websocket timeout issue for long-running workflows triggered via API. It would be great to understand if this is a known issue and/or if you have any official workarounds. For instance, many of our customers have just reverted back to using the basic HTTP api example and avoid web sockets entirely.
The text was updated successfully, but these errors were encountered:
Describe the bug
Occasionally for long-running workflows triggered via API, the websocket connection gets lost.
To Reproduce
Difficult to reproduce because it only happens occasionally, with long running workflows in particular.
Expected behavior
I would expect much more durable performance from
comfy run --workflow
. This command also seems fairly undocumented so would be nice to know how mature this feature is.Nice to have
This issue is well documented in our repo here.
Additional context
I work at Modal, a serverless compute / gpu platform. We have a lot of inbound customers searching for "comfyui api" that land on our ComfyUI example, which uses
comfy-cli
. Even before comfy-cli, we've anecdotally had many customers run into this websocket timeout issue for long-running workflows triggered via API. It would be great to understand if this is a known issue and/or if you have any official workarounds. For instance, many of our customers have just reverted back to using the basic HTTP api example and avoid web sockets entirely.The text was updated successfully, but these errors were encountered: