Troubleshooting¶
Reporting issues¶
When you encounter a problem and need help, it's essential to report it as a GitHub issue . Please avoid brining up the issue to Discord before reporting it.
Steps to reproduce
Make sure to provide clear, detailed steps to reproduce the issue. This will allow us to troubleshoot it and request any additional information. Include server logs, CLI outputs, and configuration samples. Avoid using screenshots for logs or errors—use text instead.
See these examples for well-reported issues: this and this .
Typical issues¶
Provisioning fails¶
In certain cases, running dstack apply may produce the following output:
wet-mangust-1 provisioning completed (failed)
All provisioning attempts failed. This is likely due to cloud providers not having enough capacity. Check CLI and server logs for more details.
Backend configuration¶
If runs consistently fail to provision due to insufficient capacity, it’s likely there is a backend configuration issue. Ensure that your backends are configured correctly and check the server logs for any errors.
Service quotas¶
If some runs fail to provision, it may be due to an insufficient service quota. For cloud providers like AWS, GCP, Azure, and OCI, you often need to request an increased service quota before you can use specific instances.
Resources¶
Another possible cause of the insufficient capacity error is that dstack cannot find an instance that meets the
requirements specified in resources.
GPU
The gpu property allows you to specify the GPU name, memory, and quantity. Examples include A100 (one GPU), A100:40GB (
one GPU with exact memory), A100:4 (four GPUs), etc. If you specify a GPU name without a quantity, it defaults to 1.
If you request one GPU but only instances with eight GPUs are available, dstack won’t be able to provide it. Use range
syntax to specify a range, such as A100:1..8 (one to eight GPUs) or A100:1.. (one or more GPUs).
Disk
If you don't specify the disk property, dstack defaults it to 100GB.
In case there is no such instance available, dstack won’t be able to provide it.
Use range syntax to specify a range, such as 50GB..100GB (from fifty GBs to one hundred GBs) or 50GB..
(fifty GBs or more).
Run fails¶
There could be several reasons for a run failing after successful provisioning.
Termination reason
To find out why, use -v (stands for --verbose) with dstack ps.
This will show the run's status and any failure reasons.
Spot interruption¶
If a run fails after provisioning with the termination reason INTERRUPTED_BY_NO_CAPACITY, it is likely that the run
was using spot instances and was interrupted. To address this, you can either set the
spot_policy to on-demand or specify the
retry property.
Can't run a service¶
Gateway configuration¶
The most common reason a service fails to start is either because you haven’t created a gateway or haven’t set up the correct DNS record pointing to the gateway's hostname.
Service endpoint doesn't work¶
Authorization¶
If the service endpoint returns a 403 error, it is likely because the Authorization
header with the correct dstack token was not provided.
SSH fleets¶
If you attempt to run a service on an SSH fleet, it won't work due to a known issue that is expected to be fixed soon.
Cannot access a dev environment or a task's ports¶
When running a dev environment or task with configured ports, dstack apply
automatically forwards remote ports to localhost via SSH for easy and secure access.
If you interrupt the command, the port forwarding will be disconnected. To reattach, use dstack logs --attach <run name.
Windows¶
If you're using the CLI on Windows, make sure to run it through WSL by following these instructions. Native support will be available soon.
An SSH fleet doesn't provision¶
If you set up an SSH fleet and it fails to provision after a long wait, first check the server logs.
Also, review the /root/.dstack/shim.log file on each host used to create the fleet.
Questions¶
Community
If you have a question, please feel free to ask it in our Discord server.