Skip to content
Permalink

Comparing changes

Choose two branches to see what’s changed or to start a new pull request. If you need to, you can also or learn more about diff comparisons.

Open a pull request

Create a new pull request by comparing changes across two branches. If you need to, you can also . Learn more about diff comparisons here.
base repository: dstackai/dstack
Failed to load repositories. Confirm that selected base ref is valid, then try again.
Loading
base: 0.19.17
Choose a base ref
...
head repository: dstackai/dstack
Failed to load repositories. Confirm that selected head ref is valid, then try again.
Loading
compare: 0.19.18
Choose a head ref
  • 11 commits
  • 103 files changed
  • 6 contributors

Commits on Jul 2, 2025

  1. Configuration menu
    Copy the full SHA
    7d999e1 View commit details
    Browse the repository at this point in the history

Commits on Jul 3, 2025

  1. Ignore SSH keys when calculating fleet conf diff (#2869)

    Since `/api/project/{project_name}/fleets/get_plan` returns the plan
    with sensitive info (SSH keys) removed, the configurations are never
    equal. To work around this issue, we exclude (strictly speaking, not
    exclude but reset to the default values, `None` in this case) fields
    containing SSH key content when calculating the fleet configurations
    diff.
    
    Fixes: #2222
    un-def authored Jul 3, 2025
    Configuration menu
    Copy the full SHA
    a6dbb3f View commit details
    Browse the repository at this point in the history
  2. [Blog] Refactoring (#2873)

    * [Docs] Refactoring
    
    - [x] Replaced `Blog` in the top menu with `Benchmarks` and `Case studies`
    - [x] Refactored blog categories
    - [x] Minor edits
    peterschmidt85 authored Jul 3, 2025
    Configuration menu
    Copy the full SHA
    66d8a52 View commit details
    Browse the repository at this point in the history
  3. Configuration menu
    Copy the full SHA
    17a8e22 View commit details
    Browse the repository at this point in the history

Commits on Jul 4, 2025

  1. Support processing more resources per replica (#2871)

    * Process fleets in batches
    
    * Increase offers_cache
    
    * Cache redundant AWS API calls
    
    * Set MIN_PROCESSING_INTERVAL
    
    * Prevent concurrent quotas requests
    
    * Increase MIN_PROCESSING_INTERVAL
    
    * Make runs stopping async
    
    * Introduce SERVER_BACKGROUND_PROCESSING_RATE
    
    * Use dummy locking on Postgres
    
    * Process submitted jobs in one steps if no instance lock
    
    * Fix missing cache lock
    
    * Configure custom executor
    
    * Fix typo
    
    * Increase process_submitted_jobs max_instances
    
    * Implement offers cache warm up trick
    
    * Replace DSTACK_SERVER_BACKGROUND_PROCESSING_RATE with DSTACK_SERVER_BACKGROUND_PROCESSING_FACTOR
    
    * Document DSTACK_SERVER_BACKGROUND_PROCESSING_FACTOR
    
    * Do not create clusters from burstable AWS instances
    
    Fixes #2872
    
    * Set MIN_PROCESSING_INTERVAL for process_fleets
    
    * Add TODO
    r4victor authored Jul 4, 2025
    Configuration menu
    Copy the full SHA
    2a89dff View commit details
    Browse the repository at this point in the history
  2. Use uvloop by default (#2874)

    * Use uvloop by default
    
    * Remove print
    r4victor authored Jul 4, 2025
    Configuration menu
    Copy the full SHA
    0ddaa13 View commit details
    Browse the repository at this point in the history
  3. Configuration menu
    Copy the full SHA
    f5c8fcf View commit details
    Browse the repository at this point in the history

Commits on Jul 7, 2025

  1. Fix NVIDIA container toolkit bug in all backends (#2877)

    Use cgroupfs as a Docker cgroup driver on all
    backends by default to work around an NVIDIA
    container toolkit bug where the container looses
    access to the GPU.
    
    The patch to `/etc/docker/daemon.json` is
    automatically applied in all VM-based backends if
    `/etc/docker/daemon.json` exists, has the NVIDIA
    runtime, does not explicitly set another cgroup
    driver, and `jq` is installed. This is not the
    case in Nebius. Lambda, and CUDO, so they still
    need custom code to apply the workaround - either
    installing `jq` or just writing a hardcoded
    `/etc/docker/daemon.json` that is known to work on
    this backend.
    jvstme authored Jul 7, 2025
    Configuration menu
    Copy the full SHA
    66ffdd1 View commit details
    Browse the repository at this point in the history

Commits on Jul 8, 2025

  1. Private GCP gateways (#2881)

    Allow configuring private GCP gateways without TLS
    certificates.
    
    ```yaml
    type: gateway
    name: example
    domain: gateway.example.com
    backend: gcp
    region: europe-west9
    public_ip: false
    certificate: null
    ```
    jvstme authored Jul 8, 2025
    Configuration menu
    Copy the full SHA
    faab7aa View commit details
    Browse the repository at this point in the history

Commits on Jul 9, 2025

  1. Fix log level

    r4victor committed Jul 9, 2025
    Configuration menu
    Copy the full SHA
    2cf10cc View commit details
    Browse the repository at this point in the history
  2. Switch to e2-medium for GCP gateways (#2886)

    Switching from `e2-small` to `e2-medium` decreases
    provisioning time from about 5 minutes to about 2
    minutes. The bottleneck is the snap, apt, and pip
    package installation, which takes too long on
    `e2-small`, likely because of the burstable CPU -
    `e2-small` comes with 0.5-2 vCPU, while
    `e2-medium` comes with 1-2 vCPU, depending on the
    burst.
    jvstme authored Jul 9, 2025
    Configuration menu
    Copy the full SHA
    7435ecf View commit details
    Browse the repository at this point in the history
Loading