dstack 0.19.27__tar.gz → 0.19.28__tar.gz

{dstack-0.19.27 → dstack-0.19.28}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dstack
-Version: 0.19.27
+Version: 0.19.28
 Summary: dstack is an open-source orchestration engine for running AI workloads on any cloud or on-premises.
 Project-URL: Homepage, https://dstack.ai
 Project-URL: Source, https://github.com/dstackai/dstack

{dstack-0.19.27 → dstack-0.19.28}/contributing/LOCKING.md

@@ -50,7 +50,7 @@ await session.commit()
 select ...
 ```
 
-> SQLite exhibits "snapshot isolation". When a read transaction starts, that reader continues to see an unchanging "snapshot" of the database file as it existed at the moment in time when the read transaction started. Any write transactions that commit while the read transaction is active are still invisible to the read transaction, because the reader is seeing a snapshot of database file from a prior moment in time. Source: https://www.sqlite.org/isolation.html
+> SQLite exhibits Snapshot Isolation. When a read transaction starts, that reader continues to see an unchanging "snapshot" of the database file as it existed at the moment in time when the read transaction started. Any write transactions that commit while the read transaction is active are still invisible to the read transaction, because the reader is seeing a snapshot of database file from a prior moment in time. Source: https://www.sqlite.org/isolation.html
 
 Thus, if a new transaction is not started, you won't see changes that concurrent transactions made before you acquired the lock.
 
@@ -82,3 +82,29 @@ In fact, using `joinedload` and `.with_for_update()` will trigger an error becau
 **Always use `.with_for_update(key_share=True)` unless you plan to delete rows or update a primary key column**
 
 If you `SELECT FOR UPDATE` from a table that is referenced in a child table via a foreign key, it can lead to deadlocks if the child table is updated because Postgres will issue a `FOR KEY SHARE` lock on the parent table rows to ensure valid foreign keys. For this reason, you should always do `SELECT FOR NO KEY UPDATE` (.`with_for_update(key_share=True)`) if primary key columns are not modified. `SELECT FOR NO KEY UPDATE` is not blocked by a `FOR KEY SHARE` lock, so no deadlock.
+
+
+**Lock unique names**
+
+The following pattern can be used to lock a unique name of some resource type:
+
+```python
+lock_namespace = f"fleet_names_{project.name}"
+if get_db().dialect_name == "sqlite":
+    # Start new transaction to see committed changes after lock
+    await session.commit()
+elif get_db().dialect_name == "postgresql":
+    await session.execute(
+        select(func.pg_advisory_xact_lock(string_to_lock_id(lock_namespace)))
+    )
+
+lock, _ = get_locker(get_db().dialect_name).get_lockset(lock_namespace)
+async with lock:
+    # ... select taken names, use a unique name
+    await session.commit()
+```
+
+Note that:
+
+* This pattern works assuming that Postgres is using default isolation level Read Committed. By the time a transaction acquires the advisory lock, all other transactions that can take the name have committed, so their changes can be seen and a unique name is taken.
+* SQLite needs a commit before selecting taken names due to Snapshot Isolation as noted above.

{dstack-0.19.27 → dstack-0.19.28}/docs/assets/stylesheets/extra.css

@@ -1318,10 +1318,6 @@ html .md-footer-meta.md-typeset a:is(:focus,:hover) {
         display: none;
     }
 
-    .md-tabs__item:nth-child(7) {
-        display: none;
-    }
-
     /* .md-tabs__item:nth-child(5):after {
         content: url('data:image/svg+xml,<svg fill="black" xmlns="http://www.w3.org/2000/svg" width="20px" height="20px" viewBox="0 0 16 16"><polygon points="5 4.31 5 5.69 9.33 5.69 2.51 12.51 3.49 13.49 10.31 6.67 10.31 11 11.69 11 11.69 4.31 5 4.31" data-v-e1bdab2c=""></polygon></svg>');
         line-height: 14px;

dstack-0.19.28/docs/blog/posts/digitalocean-and-amd-dev-cloud.md

@@ -0,0 +1,151 @@
+---
+title: Orchestrating GPUs on DigitalOcean and AMD Developer Cloud
+date: 2025-09-04
+description: "TBA"  
+slug: digitalocean-and-amd-dev-cloud
+image: https://dstack.ai/static-assets/static-assets/images/digitalocean-and-amd-dev-cloud.png
+categories:
+  - Changelog
+---
+
+# Orchestrating GPUs on DigitalOcean and AMD Developer Cloud
+
+Orchestration automates provisioning, running jobs, and tearing them down. While Kubernetes and Slurm are powerful in their domains, they lack the lightweight, GPU-native focus modern teams need to move faster.
+
+`dstack` is built entirely around GPUs. Our latest update introduces native integration with [DigitalOcean :material-arrow-top-right-thin:{ .external }](https://www.digitalocean.com/products/gradient/gpu-droplets){:target="_blank"} and 
+[AMD Developer Cloud :material-arrow-top-right-thin:{ .external }](https://www.amd.com/en/developer/resources/cloud-access/amd-developer-cloud.html){:target="_blank"}, enabling teams to provision cloud GPUs and run workloads more cost-efficiently.
+
+<img src="https://dstack.ai/static-assets/static-assets/images/digitalocean-and-amd-dev-cloud.png" width="630"/>
+
+<!-- more -->
+
+## About Digital Ocean
+
+DigitalOcean is one of the leading cloud platforms offering GPUs both as VMs and as bare-metal clusters equipped with NVIDIA and AMD GPUs.
+
+## About AMD Developer Cloud
+
+AMD Developer Cloud is a new cloud platform designed to make AMD GPUs easily accessible to developers, academics, open-source contributors, and AI innovators worldwide.
+
+## Why dstack
+
+`dstack` provides a high-level, AI-engineer-friendly interface where GPUs work out of the box—no K8S custom operators or low-level setup required. It’s use-case agnostic, equally suited for training, inference, benchmarking, and dev environments.
+
+With the new DigitalOcean and AMD Developer Cloud backends, you can now provision NVIDIA or AMD GPU VMs and run workloads with a single CLI command.
+
+## Getting started
+
+Best part about `dstack` is that it's very easy to get started.
+
+1. Create a project in Digital Ocean or AMD Developer Cloud
+2. Get credits or approve a payment method
+3. Create an API key
+
+Then, configure the backend in `~/.dstack/server/config.yml`:
+
+<div editor-title="~/.dstack/server/config.yml">
+
+```yaml
+projects:
+- name: main
+  backends:
+    - type: amddevcloud
+      project_name: my-amd-project
+      creds:
+        type: api_key
+        api_key: ...
+```
+
+</div>
+
+For DigitalOcean, set `type` to `digitalocean`.
+
+Install and start the `dstack` server:
+
+<div class="termy">
+
+```shell
+$ pip install "dstack[server]"
+$ dstack server
+```
+
+</div>
+
+For more details, see [Installation](../../docs/installation/index.md).
+
+Use the `dstack` CLI to
+manage [dev environments](../../docs/concepts/dev-environments.md), [tasks](../../docs/concepts/tasks.md),
+and [services](../../docs/concepts/services.md).
+
+<img src="https://dstack.ai/static-assets/static-assets/images/amd-dev-cloud.png" width="750"/>
+
+The `digitalocean` and `amddevcloud` backends support NVIDIA and AMD GPU VMs, respectively, and allow you to run 
+[dev environments](../../docs/concepts/dev-environments.md) (interactive development), [tasks](../../docs/concepts/tasks.md) 
+(training, fine-tuning, or other batch jobs), and [services](../../docs/concepts/services.md) (inference).
+
+Here’s an example of a service configuration:
+
+<div editor-title="examples/modes/gpt-oss/120b.dstack.yml">
+
+```yaml
+type: service
+name: gpt-oss-120b
+
+model: openai/gpt-oss-120b
+
+env:
+  - HF_TOKEN
+  - MODEL=openai/gpt-oss-120b
+  # To enable AITER, set below to 1. Otherwise, set it to 0.
+  - VLLM_ROCM_USE_AITER=1
+  # To enable AITER Triton unified attention
+  - VLLM_USE_AITER_UNIFIED_ATTENTION=1
+  # below is required in order to enable AITER unified attention by disabling AITER MHA
+  - VLLM_ROCM_USE_AITER_MHA=0
+image: rocm/vllm-dev:open-mi300-08052025
+commands:
+  - |
+    vllm serve $MODEL \
+      --tensor-parallel $DSTACK_GPUS_NUM \
+      --no-enable-prefix-caching \
+      --disable-log-requests \
+      --compilation-config '{"full_cuda_graph": true}'
+port: 8000
+
+volumes:
+  # Cache downloaded models
+  - /root/.cache/huggingface:/root/.cache/huggingface
+
+resources:
+  gpu: MI300X:8
+  shm_size: 32GB
+```
+
+</div>
+
+As with any configuration, you can apply it via `dstack apply`. If needed, `dstack` will automatically provision new VMs and run the inference endpoint.
+
+<div class="termy">
+
+```shell
+$ dstack apply -f examples/models/gpt-oss/120b.dstack.yml
+
+ #  BACKEND             RESOURCES                                   PRICE   
+ 1  amddevcloud (alt1)  cpu=20 mem=240GB disk=720GB MI300X:192GB:8  $15.92
+ 
+ Submit the run? [y/n]:
+```
+
+</div>
+
+> If you prefer to use bare-metal clusters with `dstack`, you can create an [SSH fleet](../../docs/concepts/fleets.md#ssh-fleets).
+> This way, you’ll be able to run [distributed tasks](../../docs/concepts/tasks.md#distributed-tasks) efficiently across the cluster.
+
+!!! info "What's next?"
+    1. Check [Quickstart](../../docs/quickstart.md)
+    2. Learn more about [DigitalOcean :material-arrow-top-right-thin:{ .external }](https://www.digitalocean.com/products/gradient/gpu-droplets){:target="_blank"} and
+         [AMD Developer Cloud :material-arrow-top-right-thin:{ .external }](https://www.amd.com/en/developer/resources/cloud-access/amd-developer-cloud.html){:target="_blank"}
+    3. Explore [dev environments](../../docs/concepts/dev-environments.md), 
+        [tasks](../../docs/concepts/tasks.md), [services](../../docs/concepts/services.md), 
+        and [fleets](../../docs/concepts/fleets.md)
+    4. Join [Discord :material-arrow-top-right-thin:{ .external }](https://discord.gg/u8SmfwPpMd){:target="_blank"}

{dstack-0.19.27 → dstack-0.19.28}/docs/docs/concepts/backends.md

@@ -925,24 +925,23 @@ projects:
       project_name: my-amd-project
       creds:
         type: api_key
-        api_key: dop_v1_71ea79a0c4bf2ffa70ac9d2a7b2689d2b41768567b22ebabe58a80066dcc5e92
+        api_key: ...
 ```
 
 </div>
 
-??? info "Project Name"
-    **project_name** configuration is optional. If it is not provided, the default project is used.
+??? info "Project"
+    If `project_name` is not set, the default project will be used.
 
 ??? info "Required permissions"
     The API key must have the following scopes assigned:
 
-    * **account** - read
-    * **droplet** - create,read,update,delete,admin
-    * **project** - create,read,update,delete
-    * **regions** - read
-    * **sizes**   - read
-    * **ssh_key**  - create,read,update,delete
-
+    * `account` - read
+    * `droplet` - create, read, update, delete, admin
+    * `project` - create, read, update, delete
+    * `regions` - read
+    * `sizes` - read
+    * `ssh_key` - create, read, update, delete
 
 
 ## Digital Ocean
@@ -960,24 +959,23 @@ projects:
       project_name: my-digital-ocean-project
       creds:
         type: api_key
-        api_key: dop_v1_71ea79a0c4bf2ffa70ac9d2a7b2689d2b41768567b22ebabe58a80066dcc5e92
+        api_key: ...
 ```
 
 </div>
 
-??? info "Project Name"
-    **project_name** configuration is optional. If it is not provided, the default project is used.
+??? info "Project"
+    If `project_name` is not set, the default project will be used.
 
 ??? info "Required permissions"
     The API key must have the following scopes assigned:
 
-    * **account** - read
-    * **droplet** - create,read,update,delete,admin
-    * **project** - create,read,update,delete
-    * **regions** - read
-    * **sizes**   - read
-    * **ssh_key**  - create,read,update,delete
-
+    * `account` - read
+    * `droplet` - create, read, update, delete, admin
+    * `project` - create, read, update, delete
+    * `regions` - read
+    * `sizes` - read
+    * `ssh_key` - create, read, update,delete
 
 ## Hot Aisle
 

{dstack-0.19.27 → dstack-0.19.28}/frontend/public/index.html

@@ -34,6 +34,7 @@
       data-modal-image="/logo-notext.svg"
       data-modal-z-index="1100"
       data-modal-title="Ask me anything"
+      data-font-family='metro-web, Metro, -apple-system, "system-ui", "Segoe UI", Roboto'
       data-project-logo="/assets/images/kapa.svg"
       data-modal-disclaimer="This is a custom LLM for dstack with access to Documentation, API references and GitHub issues. This feature is experimental - Give it a try!"
       data-user-analytics-fingerprint-enabled="true"