[bug] Tensorflow serving Timed out waiting for notification

[lminer]

trafficstars

I have tensorflow-serving requests that should take around a minute. Unfortunately, they seem to be timing out before they can finish. I'm currently getting the following error:

Timed out waiting for notification

Is there some way to set the tensorflow-serving timeout within the inference container that I haven't found?

[NeilJonkers]

Can you follow this guidance here by increasing the --rest_api_timeout_in_ms from the default 30 seconds to 180 seconds for example. Timed out waiting for notification

[lminer]

There's no way to do this without making your own custom image. Even then, the change has to occur in serve.py and you need to change the gunicorn timeout and the nginx timeout as well. It would be nice if this were just a simple env var.

[gitrc]

Also experiencing this issue and had to roll our own image to workaround. Seems like a rather simple change to implement. Is there a process to build awareness to github issues through our AWS account team?