Runpod•4mo ago

Delay time of 120,000 ms?

Runpod is advertising <250ms cold start times. I am running a custom ASR model that isnt more than a couple of gigabytes. Total docker image is 11gb.

For some reason, the delay time is inifinite and the request never goes through. Any ideas?

Dj•8/26/25, 4:01 PM

We need to download your Docker image before we can start it, this contributes to your delay time. If your handler is misconfigured we can't assign jobs to your workers, if you want to share your endpoint id I can take a look at what's happening for you.

KiqueBoatOP•8/26/25, 8:45 PM

Thanks DJ. I DM'd you the endpoint

KiqueBoatOP•8/26/25, 8:45 PM

I fixed the configuration with the ports and now getting a response. However, I am seeing 23s start times. What can I do to optiimize this?

flash-singh•8/27/25, 2:39 PM

you can't do much other than optimize and put models into container image, flashboot itself is more of a lottery at this point and depends on availability, more available hardware equates to higher chance of flashboot which gets you 200ms coldstarts, the best bet is to pick hardware that is more available than others, 4090s is a good example

Delay time of 120,000 ms?

Similar Threads

Similar Threads

Similar Threads