Set backends requests timeouts (#1793)

* Add runpod requests timeout

* Add azure requests timeout

* Add lambda requests timeout

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

@@ -41,7 +41,7 @@
     get_user_data,
 )
 from dstack._internal.core.backends.base.offers import get_catalog_offers
-from dstack._internal.core.errors import NoCapacityError
+from dstack._internal.core.errors import ComputeError, NoCapacityError
 from dstack._internal.core.models.backends.base import BackendType
 from dstack._internal.core.models.gateways import (
     GatewayComputeConfiguration,
@@ -460,7 +460,18 @@ def _launch_instance(
             message = e.error.message if e.error.message is not None else ""
             raise NoCapacityError(message)
         raise e
-    vm = poller.result()
+    vm = poller.result(timeout=600)
+    if not poller.done():
+        logger.error(
+            "Timed out waiting for instance {instance_name} launch. "
+            "The instance will be terminated."
+        )
+        _terminate_instance(
+            compute_client=compute_client,
+            resource_group=resource_group,
+            instance_name=instance_name,
+        )
+        raise ComputeError(f"Timed out waiting for instance {instance_name} launch")
     return vm
 
 

src/dstack/_internal/core/backends/lambdalabs/api_client.py

@@ -47,40 +47,40 @@ def launch_instances(
             "name": name,
         }
         resp = self._make_request("POST", "/instance-operations/launch", data)
-        if resp.ok:
-            return resp.json()["data"]["instance_ids"]
-        resp.raise_for_status()
+        if not resp.ok:
+            resp.raise_for_status()
+        return resp.json()["data"]["instance_ids"]
 
     def terminate_instances(self, instance_ids: List[str]) -> List[str]:
         data = {"instance_ids": instance_ids}
         resp = self._make_request("POST", "/instance-operations/terminate", data)
-        if resp.ok:
-            return resp.json()["data"]
-        resp.raise_for_status()
+        if not resp.ok:
+            resp.raise_for_status()
+        return resp.json()["data"]
 
     def list_ssh_keys(self) -> List[Dict]:
         resp = self._make_request("GET", "/ssh-keys")
-        if resp.ok:
-            return resp.json()["data"]
-        resp.raise_for_status()
+        if not resp.ok:
+            resp.raise_for_status()
+        return resp.json()["data"]
 
     def add_ssh_key(self, name: str, public_key: str) -> List[Dict]:
         data = {
             "name": name,
             "public_key": public_key,
         }
         resp = self._make_request("POST", "/ssh-keys", data)
-        if resp.ok:
-            return resp.json()["data"]
-        resp.raise_for_status()
+        if not resp.ok:
+            resp.raise_for_status()
+        return resp.json()["data"]
 
     def _make_request(self, method: str, path: str, data: Any = None):
-        # TODO: fix S113 by setting an adequate timeout here or in every method
-        return requests.request(  # noqa: S113
+        return requests.request(
             method=method,
             url=API_URL + path,
             json=data,
             headers={"Authorization": f"Bearer {self.api_key}"},
+            timeout=120,
         )
 
     def _url(self, path: str) -> str:

src/dstack/_internal/core/backends/runpod/api_client.py

@@ -230,11 +230,11 @@ def delete_network_volume(self, volume_id: str):
 
     def _make_request(self, data: Any = None) -> Response:
         try:
-            # TODO: fix S113 by setting an adequate timeout here or in every method
-            response = requests.request(  # noqa: S113
+            response = requests.request(
                 method="POST",
                 url=f"{API_URL}?api_key={self.api_key}",
                 json=data,
+                timeout=120,
             )
             response.raise_for_status()
             if "errors" in response.json():