dstack icon indicating copy to clipboard operation
dstack copied to clipboard

An open-source container orchestration engine for running AI workloads in any cloud or data center. https://discord.gg/u8SmfwPpMd

Results 124 dstack issues
Sort by recently updated
recently updated
newest added

Many places in the "_process_job" function require error handling. - [ ] `RunnerClient.pull()` inside the `_process_running` function can raise `RequestException` - [ ] `ShimClient.submit() inside the `_process_provisioning_with_shim` function can raise...

stale

### dstack version 0.16.5 ### Python version 3.11 ### Host OS Arch Linux ### Host Arch x86_64 ### What happened? When the job is in the pulling state, the container...

bug
stale

Since SSH keys are added to the instance only during creation, other users cannot add their keys later. dstack server must update SSH keys on job submission (docker image submission,...

bug
stale

Steps to reproduce: 1. Configure the CLI with dstack Sky 2. Run `dstack pool add --gpu 1` It says: ``` There are no offers with these criteria. Exiting... ``` The...

stale
sky
major

### dstack version 0.16.5 ### Python version 3.11 ### Host OS macOS ### Host Arch arm64 ### What happened? When running `dstack run` on a file with the content below,...

bug
stale

Gateway creation is currently a synchronous request, which can potentially take a long time. This is especially relevant in the context of Kubernetes gateways (#861) deployed behind a Load balancer...

internals
stale

Current Configurator interface and Configurator-related models have lots of components that allow for flexible backend configuration. The complexity that comes with Configurator flexibility makes it hard to implement new backends,...

internals
enhancement

Steps to reproduce: 1. Run `dstack run . -b azure --gpu A10 --spot-auto` - The run hangs for 20-ish minutes and then fails - The server throws [an error](https://github.com/dstackai/dstack/files/14599069/error.txt) -...

major

ProjectModel loads PoolModel, InstanceModel, JobModel in a non-lazy way by default. This causes n+1 queries and slows down request processing. Fix by loading PoolModel, InstanceModel, JobModel only when it's required.

enhancement
major

### dstack version master ### Steps to reproduce 1. Install dstack and run `dstack server` on a Python version not listed in [`PythonVersion`](https://github.com/dstackai/dstack/blob/5da808e7949c1df0bdf246a016ca1fdf2edc8f54/src/dstack/_internal/core/models/configurations.py#L25) 1. Run a job without specifying `python`...

bug