FAQ
Run Box FAQ
I can't connect over SSH
Check the items below first. If you still can't connect, add -vvv to your existing SSH command and share a screenshot through the contact link.
ssh -vvv -i (……………..)
- Confirm you can connect via the web console
- Confirm the directory where you run the SSH command matches where the private key is stored
- Confirm the instance was actually created
My SSH connection keeps dropping
If the instance's memory usage is too high, the instance can crash. Clean up processes and memory, then try SSH again.
If the problem persists, please contact us using the link below.
My instance stays in "Pending" and won't start
When On-Demand demand spikes, instance resource allocation can be delayed. If you need stable usage for things like research projects, apply for a Reserved plan via the link below.
The CUDA version shown by nvidia-smi is different
The CUDA version shown by nvidia-smi is the maximum CUDA version supported by the GPU driver. This is different from the CUDA Toolkit (nvcc) version or the CUDA version your framework actually uses. Check the CUDA Toolkit version using the link below.
Am I still charged for storage after stopping an instance?
Yes — stopping the instance does not stop storage charges. To stop storage charges you need to delete the instance. You can review instance and storage charges in your usage details.
Can I expand storage capacity?
Yes — you can expand storage capacity up to 2 TiB from the instance detail page.
Can I shrink storage?
No — once expanded, storage cannot be reduced. However, you are billed for the actual storage you occupy, not the maximum allocated capacity.
You can reduce your storage bill by cleaning up data inside the instance.
How do I enable Data Hub external connections?
To enable "Allow External Connection" on a bucket, go to Bucket Management > Edit Bucket > Per-User Permissions > search for the user > Allow External Connection. For details, see the guide linked below.
rsync between instances is very slow
If you enable "Allow External Connection" on the bucket and use rclone, you can upload to Data Hub quickly. For directories with many files (rather than a single file), uploading in parallel with rclone is much faster.
rclone copy /source/path your-datahub-config:bucket-name --transfers=8 --s3-no-check-bucket --inplace
How do I bulk-download files from a bucket?
You can bulk-download files with commands like the following:
-
rclone sync {Remote Name}:{Bucket Name} -
ex) rclone sync elice-datahub:elice-bucket-bmlh6uyt
For more details on rclone, see the link below.