Image Pull Hanging
My image pull has been hanging on
create container: still fetching image ...
レUsually it tells me the progress, but currently it hasn't been showing me anything.
P@レム
Escalated To Zendesk
The thread has been escalated to Zendesk!
Ticket ID: #26901
Jhmm any pod / endpoint id?
DI have the same for me. It pulls very long time. And also it were trying to download 2.4GB of container related files for 14minutes. Are there network problems on pods right no?
AIt appears that there are some network issues occurring today.
Our production environment had been running stably for several weeks, but the network seems to be unstable today.
A large number of workers are being throttled, so it might be the usual Runpod event where a single user consumes a large amount of resources and causes network congestion.
Our production environment had been running stably for several weeks, but the network seems to be unstable today.
A large number of workers are being throttled, so it might be the usual Runpod event where a single user consumes a large amount of resources and causes network congestion.
HIf you get a pod with no visible download progress and its just stuck on creating for longer than 2 minutes then just delete it. It will never stop and its a waste of time. The new pod will then almost always work.
Vsame here
Vcan't start the pod with GPU, I'm trying third time now, over 2 hours waiting for fetching container, first there were network issues and now everything is very slow like 500 KB/s and i need to download 10 GB, before it downloads everything the whole pod restarts, everything worked normally before
VI'm already waiting for over 40 minutes to start the pod
DI have a call with Docker later today to discuss this very issue. Thank you for reporting it, I need cases like this to build my point.
HThe usual dockerhub throttle?
HI'm so glad my image is only 200mb lol
DYes :fbslightsmile: Sending you a message
レI went to bed sorry, but when I got up it was running
AI am using GCP's Artifact Registry without Docker, but I often run into issues with slow image pulls. I believe that the problem is with the data center where the GPUs are deployed, rather than with other companies.
DIt is a little bit of both. The datacenters send a lot of traffic to the container registry, most of them react by tuning the download speed during peak hours. But at the same time the disks were writing to aren't breaking any speed records.
DWe're working on modernizing our storage solution to assist that too, it's live for employees in one single datacenter.
