Add Nebius backend (#2463)

Author: jvstme

docs/docs/concepts/backends.md

@@ -596,6 +596,56 @@ projects:
 
 </div>
 
+### Nebius
+
+Log into your [Nebius AI Cloud :material-arrow-top-right-thin:{ .external }](https://console.eu.nebius.com/) account, navigate to Access, and select Service Accounts. Create a service account, add it to the editors group, and upload its authorized key.
+
+Then configure the backend:
+
+<div editor-title="~/.dstack/server/config.yml">
+
+```yaml
+projects:
+- name: main
+  backends:
+  - type: nebius
+    creds:
+      type: service_account
+      service_account_id: serviceaccount-e00dhnv9ftgb3cqmej
+      public_key_id: publickey-e00ngaex668htswqy4
+      private_key_file: ~/path/to/key.pem
+```
+
+</div>
+
+??? info "Configuring in the UI"
+    If you are configuring the backend in the `dstack` UI, specify the contents of the private key file in `private_key_content`.
+
+    <div editor-title="~/.dstack/server/config.yml">
+
+    ```yaml
+    type: nebius
+    creds:
+      type: service_account
+      service_account_id: serviceaccount-e00dhnv9ftgb3cqmej
+      public_key_id: publickey-e00ngaex668htswqy4
+      private_key_content: |
+        -----BEGIN PRIVATE KEY-----
+        MIIJQQIBADANBgkqhkiG9w0BAQEFAASCCSswggknAgEAAoICAQChwQ5OOhy60N7m
+        cPx/9M0oRUyJdRRv2nCALbdU/wSDOo8o5N7sP63zCaxXPeKwLNEzneMd/U0gWSv2
+        [...]
+        8y1qYDPKQ8LR+DPCUmyhM2I8t6673Vz3GrtEjkLhgQo/KqOVb3yiBFVfkA5Jov5s
+        kO7y4T0ynsI8b6wlhCukQTLpIYJ5
+        -----END PRIVATE KEY-----
+    ```
+
+    </div>
+
+!!! info "Python version"
+    Nebius is only supported if `dstack server` is running on Python 3.10 or higher.
+
+
+
 ### RunPod
 
 Log into your [RunPod :material-arrow-top-right-thin:{ .external }](https://www.runpod.io/console/) console, click Settings in the sidebar, expand the `API Keys` section, and click

docs/docs/reference/server/config.yml.md

@@ -15,7 +15,7 @@ to configure [backends](../../concepts/backends.md) and other [sever-level setti
     overrides:
         show_root_heading: false
         backends:
-            type: 'Union[AWSBackendConfigWithCreds, AzureBackendConfigWithCreds, GCPBackendConfigWithCreds, LambdaBackendConfigWithCreds, RunpodBackendConfigWithCreds, TensorDockBackendConfigWithCreds, VastAIBackendConfigWithCreds, KubernetesConfig]'
+            type: 'Union[AWSBackendConfigWithCreds, AzureBackendConfigWithCreds, GCPBackendConfigWithCreds, LambdaBackendConfigWithCreds, NebiusBackendConfigWithCreds, RunpodBackendConfigWithCreds, TensorDockBackendConfigWithCreds, VastAIBackendConfigWithCreds, KubernetesConfig]'
 
 #### `projects[n].backends` { #backends data-toc-label="backends" }
 
@@ -143,6 +143,23 @@ to configure [backends](../../concepts/backends.md) and other [sever-level setti
         type:
             required: true
 
+##### `projects[n].backends[type=nebius]` { #nebius data-toc-label="nebius" }
+
+#SCHEMA# dstack._internal.core.backends.nebius.models.NebiusBackendFileConfigWithCreds
+    overrides:
+        show_root_heading: false
+        type:
+            required: true
+        item_id_prefix: nebius-
+
+###### `projects[n].backends[type=nebius].creds` { #nebius-creds data-toc-label="creds" }
+
+#SCHEMA# dstack._internal.core.backends.nebius.models.NebiusServiceAccountFileCreds
+    overrides:
+        show_root_heading: false
+        type:
+            required: true
+
 ##### `projects[n].backends[type=runpod]` { #runpod data-toc-label="runpod" }
 
 #SCHEMA# dstack._internal.core.backends.runpod.models.RunpodBackendConfigWithCreds

frontend/src/types/backend.d.ts

@@ -7,6 +7,7 @@ declare type TBackendType = | "aws"
     |"kubernetes"
     |"lambda"
     |"local"
+    |"nebius"
     |"remote"
     |"oci"
     |"runpod"

setup.py

@@ -49,7 +49,7 @@ def get_long_description():
     "python-multipart>=0.0.16",
     "filelock",
     "psutil",
-    "gpuhunt>=0.1.0,<0.2.0",
+    "gpuhunt>=0.1.1,<0.2.0",
     "argcomplete>=3.5.0",
 ]
 
@@ -110,10 +110,23 @@ def get_long_description():
 
 LAMBDA_DEPS = AWS_DEPS
 
+# Nebius requires Python 3.10. On 3.9:
+# `pip install dstack[nebius]` is expected to fail
+# `pip install dstack[all]` is expected to ignore Nebius
+NEBIUS_DEPS = ["nebius>=0.2.19,<0.3"]
+MAYBE_NEBIUS_DEPS = ['nebius>=0.2.19,<0.3; python_version>="3.10"']
+
 OCI_DEPS = ["oci"]
 
 ALL_DEPS = (
-    SERVER_DEPS + AWS_DEPS + AZURE_DEPS + GCP_DEPS + DATACRUNCH_DEPS + KUBERNETES_DEPS + OCI_DEPS
+    SERVER_DEPS
+    + AWS_DEPS
+    + AZURE_DEPS
+    + GCP_DEPS
+    + DATACRUNCH_DEPS
+    + KUBERNETES_DEPS
+    + MAYBE_NEBIUS_DEPS
+    + OCI_DEPS
 )
 
 setup(
@@ -155,6 +168,7 @@ def get_long_description():
         "gcp": SERVER_DEPS + GCP_DEPS,
         "kubernetes": SERVER_DEPS + KUBERNETES_DEPS,
         "lambda": SERVER_DEPS + LAMBDA_DEPS,
+        "nebius": SERVER_DEPS + NEBIUS_DEPS,
         "oci": SERVER_DEPS + OCI_DEPS,
     },
     classifiers=[

src/dstack/_internal/core/backends/base/compute.py

@@ -94,6 +94,9 @@ def terminate_instance(
         """
         Terminates an instance by `instance_id`.
         If the instance does not exist, it should not raise errors but return silently.
+
+        Should return ASAP. If required to wait for some operation, raise `NotYetTerminated`.
+        In this case, the method will be called again after a few seconds.
         """
         pass
 

src/dstack/_internal/core/backends/base/models.py

@@ -1,14 +1,14 @@
 from pathlib import Path
 
 
-def fill_data(values: dict):
-    if values.get("data") is not None:
+def fill_data(values: dict, filename_field: str = "filename", data_field: str = "data") -> dict:
+    if values.get(data_field) is not None:
         return values
-    if "filename" not in values:
-        raise ValueError()
+    if (filename := values.get(filename_field)) is None:
+        raise ValueError(f"Either `{filename_field}` or `{data_field}` must be specified")
     try:
-        with open(Path(values["filename"]).expanduser()) as f:
-            values["data"] = f.read()
+        with open(Path(filename).expanduser()) as f:
+            values[data_field] = f.read()
     except OSError:
-        raise ValueError(f"No such file {values['filename']}")
+        raise ValueError(f"No such file {filename}")
     return values

src/dstack/_internal/core/backends/configurators.py

@@ -63,6 +63,15 @@
 except ImportError:
     pass
 
+try:
+    from dstack._internal.core.backends.nebius.configurator import (
+        NebiusConfigurator,
+    )
+
+    _CONFIGURATOR_CLASSES.append(NebiusConfigurator)
+except ImportError:
+    pass
+
 try:
     from dstack._internal.core.backends.oci.configurator import OCIConfigurator
 

src/dstack/_internal/core/backends/models.py

@@ -34,6 +34,11 @@
     LambdaBackendConfig,
     LambdaBackendConfigWithCreds,
 )
+from dstack._internal.core.backends.nebius.models import (
+    NebiusBackendConfig,
+    NebiusBackendConfigWithCreds,
+    NebiusBackendFileConfigWithCreds,
+)
 from dstack._internal.core.backends.oci.models import (
     OCIBackendConfig,
     OCIBackendConfigWithCreds,
@@ -65,6 +70,7 @@
     GCPBackendConfig,
     KubernetesBackendConfig,
     LambdaBackendConfig,
+    NebiusBackendConfig,
     OCIBackendConfig,
     RunpodBackendConfig,
     TensorDockBackendConfig,
@@ -86,6 +92,7 @@
     KubernetesBackendConfigWithCreds,
     LambdaBackendConfigWithCreds,
     OCIBackendConfigWithCreds,
+    NebiusBackendConfigWithCreds,
     RunpodBackendConfigWithCreds,
     TensorDockBackendConfigWithCreds,
     VastAIBackendConfigWithCreds,
@@ -105,6 +112,7 @@
     KubernetesBackendFileConfigWithCreds,
     LambdaBackendConfigWithCreds,
     OCIBackendConfigWithCreds,
+    NebiusBackendFileConfigWithCreds,
     RunpodBackendConfigWithCreds,
     TensorDockBackendConfigWithCreds,
     VastAIBackendConfigWithCreds,

src/dstack/_internal/core/backends/nebius/__init__.py

@@ -0,0 +1,16 @@
+from dstack._internal.core.backends.base.backend import Backend
+from dstack._internal.core.backends.nebius.compute import NebiusCompute
+from dstack._internal.core.backends.nebius.models import NebiusConfig
+from dstack._internal.core.models.backends.base import BackendType
+
+
+class NebiusBackend(Backend):
+    TYPE = BackendType.NEBIUS
+    COMPUTE_CLASS = NebiusCompute
+
+    def __init__(self, config: NebiusConfig):
+        self.config = config
+        self._compute = NebiusCompute(self.config)
+
+    def compute(self) -> NebiusCompute:
+        return self._compute