dstack
dstack copied to clipboard
dstack is an open-source alternative to Kubernetes, designed to simplify development, training, and deployment of AI across any cloud or on-prem. It supports NVIDIA, AMD, and TPU.
### Problem In some cases, `.gitignore` cannot be used, e.g. when users want to push large local files to Git eventually and don't want to list them in `.gitignore`. At...
### Problem If an SSH instance forbids port forwarding with `AllowTcpForwarding no`, `dstack` does not detect this configuration error at `dstack apply` time. The instance is marked as `idle` and...
### Problem It can be difficult to get run status when there are many runs in the same project ### Solution Implement push notifications to slack (and/or other platforms) for...
### Problem Some runs may need to be run sequentially and depends on a previous run completing successfully. ### Solution Introduce a `depends_on` field, in which a run can be...
### Problem No data persistence when using the `nebius` backend. ### Solution Introduce `dstack` network volumes for the `nebius` backend, implemented using Nebius [shared filesystems](https://docs.nebius.com/compute/storage/types#filesystems). ### Workaround _No response_ ###...
### Problem When I'm trying to develop a task, it is typically a bunch of, 'write yaml', run dstack apply, figure out where it failed. It would be great if...
### Problem A node in a distributed task may need to know its own private IP address, e.g. to bind a socket to it or for logging purposes. Currently, there...
### Problem Currently the only documented way of pulling images from private registries is to use the [registry_auth](https://dstack.ai/docs/reference/dstack.yml/task/#_registry_auth) fields dstack configs. This option requires putting secrets in a file that...
### Problem Hi `dstack` team and community, First off, thanks for creating `dstack`! It's a fantastic tool that has really simplified our inference infrastructure. We use `dstack` for our LLM...
Currently, if I set up an SSH fleet that uses a proxy jump, any users that want to use such a fleet have to update their SSH config and specify...