Runpod•3mo ago

error creating container: nvidia-smi: parsing output of line 0: failed to parse (pcie.link.gen.max)

I've been testing the cold start, and it works 3-4 times, but then I get the above error. I'm using the serverless loadbalancer endpoints.

emilwallnerOP•10/2/25, 5:33 PM

Similiar error: error creating container: nvidia-smi: parsing output of line 0: failed to parse (pcie.link.gen.max) into int: strconv.Atoi: parsing "": invalid syntax

WeamonZ•10/2/25, 5:59 PM

check if it only affects one worker

WeamonZ•10/2/25, 5:59 PM

I had this error on a worker, I deleted it

Dj•10/2/25, 7:52 PM

That helps, but next time if you wanna grab the endpoint or worker ID it'd help a lot. Not a requirement though!

WWeamonZ Mhhh I've seen a worker with the following system error: ``` error creating cont...

WeamonZ•10/2/25, 9:10 PM

WeamonZ•10/2/25, 9:12 PM

@Dj endpoint : 7sgrvwwxr3nbzz

WeamonZ•10/2/25, 9:13 PM

But I had to delete the worker since it broke my production environment...

emilwallnerOP•10/3/25, 9:08 AM

I had two workers with the error, not ideal if a large amount of workers don't work, makes prod risky

emilwallnerOP•10/3/25, 9:12 AM

@Dj also, it's not just that it fails, it keeps trying and getting the same error and not launching a new worker

WeamonZ•10/3/25, 4:47 PM

@Dj absolutely still not working

Dj•10/3/25, 4:48 PM

One sec, let me find this worker

Dj•10/3/25, 4:49 PM

For you, is it the worker with id aj0eezu3r5o7e9aj0eezu3r5o7e9?

WeamonZ•10/3/25, 4:53 PM

@Dj I'm filling a ticket, it happens on other endpoints with an other worker

WeamonZ•10/3/25, 4:53 PM

Dj•10/3/25, 4:53 PM

It's the specific machine you're being balanced to

Dj•10/3/25, 4:53 PM

Support can find the machine a little more delicately than I can, but their time to first response may be a little longer

WeamonZ•10/3/25, 4:53 PM

it's been booting for 3mins. I ran the same request 1 min before, it handled it correctly

Dj•10/3/25, 4:54 PM

Oh i see, you're still pullng the image from the Runpod Registry

DDj Support can find the machine a little more delicately than I can, but their time...

WeamonZ•10/3/25, 4:54 PM

I'm filling a ticket for more things anyway, bad experiences are adding up recently

DDj Oh i see, you're still pullng the image from the Runpod Registry

WeamonZ•10/3/25, 4:54 PM

yet the worker was working 1min ago

WeamonZ•10/3/25, 4:54 PM

am I charged for it ?

WeamonZ•10/6/25, 4:38 PM

@Dj ah yes

now an other endpoint hit by this

WeamonZ•10/6/25, 4:38 PM

error creating container: nvidia-smi: parsing output of line 0: failed to parse (pcie.link.gen.max)

Similar Threads

Similar Threads

Similar Threads