whats going on with the pods today?
literally been trying to start up a pod since the moment the day started, waited many rounds of 30 min to an hour and never loads, one time only did it load only to direct me towards a broken http link that didnt go up for another hour. Is there some sort of maintenance thing going on? just so i know whether or not to keep throwing money
13 Replies
please answer instead of just passing this by. I really need some help here.
another question, how do I fix the bad gateway page always popping up?
ive eaten through like 5 dollars just trying to even get pods to start up.
fuck yall bro im never using this service again 😠why have these "support" channels if nobody bothers checking on them.
like look at the rest of the threads here literally 0 answers in any of them
if ur reading this as a potential client and considering buying runpod lemme tell u something since these guys wont read this anyway: dont. not worth it. get a better alternative.
You can always email
help@runpod.io for access to our Support Team. This Discord is primarily run by myself and I am lucky to be supported by members of our Engineering and Product teams.
To answer your question:
Whatever your Pod has to do at startup takes a very long time (or, slightly less commonly your Docker Registry is very slow). The "bad gateway" after your Pod starts indicates the HTTP service on the Pod isn't started. Without any Pod or Endpoint ID I can only speculate or define the errors you've described.
Nothing about our platform has changed or will change without an announcement.Not to hijack this thread, but the user is correct, something is off with RunPod. And it has been for a week or so.
Statements like these again without any symptoms or an account identifier ultimately mean nothing. At the end of the day we sell hardware, we do not manufacture GPUs and we have not replaced the entire fleet with faulty components overnight.
I'd like to help you, really I do, but you have to understand how vague that statement is.
I've already DM'd Max about it on twitter, the other day, so fingers crossed somethign will come of it
@Dj ok so here's my newest pod i just opened zlp7dr53nacws4
today ive had 3 attempts at making pods and all of them have sent me to broken links
itd be cool if you could take a look and maybe tell me whats going on with the pod or what i can do to fix this before it keeps eating up more money
dear and lovely support, i am having the same issues lately - 3rd pod shows no signs of any http activity...
@Daniel / @tedi.ted are these official templates u are able to have the same problem with? or these custom tempaltes
asking b/c i launched a pod on my personal account and works with the pytorch templates? so im thinking is it a template specific issue?
(Or could be region issue)
@Daniel if you are using a custom templates, sometimes can be helpful to provide logs.
What could be happening is that a template is trying to install a lot of services, and not launchign the application on the http port till it is done. (hard to say without seeing the template)
@justin (New) [Staff Not Staff] I am using Automatic1111 Stable Diffusion WebUI, Kohya SS and ComfyUI in RO region (that's where my network storage is) - I can open Jupyter, but the usual Filezilla xml config file re-generated on startup is already 0 bytes, other HTTP services don't respond, web terminal works though. Just testing it in CET working hours, will kepp you posted here...
why are we pretending like runpod doesn't have network issues reasonably frequently? im seeing it now on US-CA-2 where docker pulls are failing cuz of TLS handshake, and before that i had to terminate mypod because it was constantly disconnecting for about 60 seconds only to reconnect for about 3.
it doesnt sound like a network issue vs something else then.
if its a network issue then jupyter labs / web terminal would not work as web terminal is based off ssh connection establishing properly, jupyter labs is proving ur http connection is fine + network proxy.
means a file is corrupted somehow or configuration with the template for ur other services
Totally understand the frustration. It's a battle between many data centers, hardware, and a dozen other things in between. (and especially with custom templates, can always be hard to pinpoint what is going on). Always trying to do better, and if you ever encounter issues that is on us:
https://contact.runpod.io/hc/en-us/requests/new
Support can always help issue refunds. Just screenshot, pod id, timestamps, all help for us to trace the event down.