skypilot-nightly 1.0.0.dev20241125__tar.gz → 1.0.0.dev20241127__tar.gz

{skypilot_nightly-1.0.0.dev20241125/skypilot_nightly.egg-info → skypilot_nightly-1.0.0.dev20241127}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: skypilot-nightly
-Version: 1.0.0.dev20241125
+Version: 1.0.0.dev20241127
 Summary: SkyPilot: An intercloud broker for the clouds
 Author: SkyPilot Team
 License: Apache 2.0

{skypilot_nightly-1.0.0.dev20241125 → skypilot_nightly-1.0.0.dev20241127}/sky/__init__.py

@@ -5,7 +5,7 @@ from typing import Optional
 import urllib.request
 
 # Replaced with the current commit when building the wheels.
-_SKYPILOT_COMMIT_SHA = '894330bd5455e8b6739a5c22abdee2529f5fe282'
+_SKYPILOT_COMMIT_SHA = '23f9821bd165c8d154a44c3505a18b1e28cc0f48'
 
 
 def _get_git_commit():
@@ -35,7 +35,7 @@ def _get_git_commit():
 
 
 __commit__ = _get_git_commit()
-__version__ = '1.0.0.dev20241125'
+__version__ = '1.0.0.dev20241127'
 __root_dir__ = os.path.dirname(os.path.abspath(__file__))
 
 

{skypilot_nightly-1.0.0.dev20241125 → skypilot_nightly-1.0.0.dev20241127}/sky/cli.py

@@ -486,7 +486,7 @@ def _parse_override_params(
         image_id: Optional[str] = None,
         disk_size: Optional[int] = None,
         disk_tier: Optional[str] = None,
-        ports: Optional[Tuple[str]] = None) -> Dict[str, Any]:
+        ports: Optional[Tuple[str, ...]] = None) -> Dict[str, Any]:
     """Parses the override parameters into a dictionary."""
     override_params: Dict[str, Any] = {}
     if cloud is not None:
@@ -539,7 +539,14 @@ def _parse_override_params(
         else:
             override_params['disk_tier'] = disk_tier
     if ports:
-        override_params['ports'] = ports
+        if any(p.lower() == 'none' for p in ports):
+            if len(ports) > 1:
+                with ux_utils.print_exception_no_traceback():
+                    raise ValueError('Cannot specify both "none" and other '
+                                     'ports.')
+            override_params['ports'] = None
+        else:
+            override_params['ports'] = ports
     return override_params
 
 
@@ -730,7 +737,7 @@ def _make_task_or_dag_from_entrypoint_with_overrides(
     image_id: Optional[str] = None,
     disk_size: Optional[int] = None,
     disk_tier: Optional[str] = None,
-    ports: Optional[Tuple[str]] = None,
+    ports: Optional[Tuple[str, ...]] = None,
     env: Optional[List[Tuple[str, str]]] = None,
     field_to_ignore: Optional[List[str]] = None,
     # job launch specific
@@ -1084,7 +1091,7 @@ def launch(
     env: List[Tuple[str, str]],
     disk_size: Optional[int],
     disk_tier: Optional[str],
-    ports: Tuple[str],
+    ports: Tuple[str, ...],
     idle_minutes_to_autostop: Optional[int],
     down: bool,  # pylint: disable=redefined-outer-name
     retry_until_up: bool,

{skypilot_nightly-1.0.0.dev20241125 → skypilot_nightly-1.0.0.dev20241127}/sky/clouds/service_catalog/__init__.py

@@ -324,9 +324,8 @@ def get_common_gpus() -> List[str]:
         'A100',
         'A100-80GB',
         'H100',
-        'K80',
         'L4',
-        'M60',
+        'L40S',
         'P100',
         'T4',
         'V100',
@@ -337,13 +336,13 @@ def get_common_gpus() -> List[str]:
 def get_tpus() -> List[str]:
     """Returns a list of TPU names."""
     # TODO(wei-lin): refactor below hard-coded list.
-    # There are many TPU configurations available, we show the three smallest
-    # and the largest configuration for the latest gen TPUs.
+    # There are many TPU configurations available, we show the some smallest
+    # ones for each generation, and people should find larger ones with
+    # sky show-gpus tpu.
     return [
-        'tpu-v2-512', 'tpu-v3-2048', 'tpu-v4-8', 'tpu-v4-16', 'tpu-v4-32',
-        'tpu-v4-3968', 'tpu-v5litepod-1', 'tpu-v5litepod-4', 'tpu-v5litepod-8',
-        'tpu-v5litepod-256', 'tpu-v5p-8', 'tpu-v5p-32', 'tpu-v5p-128',
-        'tpu-v5p-12288'
+        'tpu-v2-8', 'tpu-v3-8', 'tpu-v4-8', 'tpu-v4-16', 'tpu-v4-32',
+        'tpu-v5litepod-1', 'tpu-v5litepod-4', 'tpu-v5litepod-8', 'tpu-v5p-8',
+        'tpu-v5p-16', 'tpu-v5p-32', 'tpu-v6e-1', 'tpu-v6e-4', 'tpu-v6e-8'
     ]
 
 

{skypilot_nightly-1.0.0.dev20241125 → skypilot_nightly-1.0.0.dev20241127}/sky/clouds/service_catalog/data_fetchers/fetch_azure.py

@@ -64,7 +64,7 @@ FAMILY_NAME_TO_SKYPILOT_GPU_NAME = {
     'standardNVSv2Family': 'M60',
     'standardNVSv3Family': 'M60',
     'standardNVPromoFamily': 'M60',
-    'standardNVSv4Family': 'Radeon MI25',
+    'standardNVSv4Family': 'MI25',
     'standardNDSFamily': 'P40',
     'StandardNVADSA10v5Family': 'A10',
     'StandardNCadsH100v5Family': 'H100',

{skypilot_nightly-1.0.0.dev20241125 → skypilot_nightly-1.0.0.dev20241127}/sky/provision/azure/instance.py

@@ -306,7 +306,7 @@ def _create_vm(
         identity=compute.VirtualMachineIdentity(
             type='UserAssigned',
             user_assigned_identities={provider_config['msi']: {}}),
-        priority=node_config['azure_arm_parameters']['priority'])
+        priority=node_config['azure_arm_parameters'].get('priority', None))
     vm_poller = compute_client.virtual_machines.begin_create_or_update(
         resource_group_name=provider_config['resource_group'],
         vm_name=vm_name,

{skypilot_nightly-1.0.0.dev20241125 → skypilot_nightly-1.0.0.dev20241127}/sky/provision/kubernetes/utils.py

@@ -1045,11 +1045,15 @@ def get_kube_config_context_namespace(
             the default namespace.
     """
     k8s = kubernetes.kubernetes
-    # Get namespace if using in-cluster config
     ns_path = '/var/run/secrets/kubernetes.io/serviceaccount/namespace'
-    if os.path.exists(ns_path):
-        with open(ns_path, encoding='utf-8') as f:
-            return f.read().strip()
+    # If using in-cluster context, get the namespace from the service account
+    # namespace file. Uses the same logic as adaptors.kubernetes._load_config()
+    # to stay consistent with in-cluster config loading.
+    if (context_name == kubernetes.in_cluster_context_name() or
+            context_name is None):
+        if os.path.exists(ns_path):
+            with open(ns_path, encoding='utf-8') as f:
+                return f.read().strip()
     # If not in-cluster, get the namespace from kubeconfig
     try:
         contexts, current_context = k8s.config.list_kube_config_contexts()
@@ -1136,7 +1140,11 @@ class KubernetesInstanceType:
         name = (f'{common_utils.format_float(self.cpus)}CPU--'
                 f'{common_utils.format_float(self.memory)}GB')
         if self.accelerator_count:
-            name += f'--{self.accelerator_count}{self.accelerator_type}'
+            # Replace spaces with underscores in accelerator type to make it a
+            # valid logical instance type name.
+            assert self.accelerator_type is not None, self.accelerator_count
+            acc_name = self.accelerator_type.replace(' ', '_')
+            name += f'--{self.accelerator_count}{acc_name}'
         return name
 
     @staticmethod
@@ -1167,7 +1175,9 @@ class KubernetesInstanceType:
             accelerator_type = match.group('accelerator_type')
             if accelerator_count:
                 accelerator_count = int(accelerator_count)
-                accelerator_type = str(accelerator_type)
+                # This is to revert the accelerator types with spaces back to
+                # the original format.
+                accelerator_type = str(accelerator_type).replace('_', ' ')
             else:
                 accelerator_count = None
                 accelerator_type = None

{skypilot_nightly-1.0.0.dev20241125 → skypilot_nightly-1.0.0.dev20241127}/sky/templates/aws-ray.yml.j2

@@ -173,6 +173,7 @@ setup_commands:
   # Line 'sudo grep ..': set the number of threads per process to unlimited to avoid ray job submit stucking issue when the number of running ray jobs increase.
   # Line 'mkdir -p ..': disable host key check
   # Line 'python3 -c ..': patch the buggy ray files and enable `-o allow_other` option for `goofys`
+  # Line 'rm ~/.aws/credentials': explicitly remove the credentials file to be safe. This is to guard against the case where the credential files was uploaded once as `remote_identity` was not set in a previous launch.
   - mkdir -p ~/.ssh; touch ~/.ssh/config;
     {%- for initial_setup_command in initial_setup_commands %}
     {{ initial_setup_command }}
@@ -185,7 +186,12 @@ setup_commands:
     sudo grep -e '^DefaultTasksMax' /etc/systemd/system.conf || (sudo bash -c 'echo "DefaultTasksMax=infinity" >> /etc/systemd/system.conf'); sudo systemctl set-property user-$(id -u $(whoami)).slice TasksMax=infinity; sudo systemctl daemon-reload;
     {%- endif %}
     mkdir -p ~/.ssh; (grep -Pzo -q "Host \*\n  StrictHostKeyChecking no" ~/.ssh/config) || printf "Host *\n  StrictHostKeyChecking no\n" >> ~/.ssh/config;
-    [ -f /etc/fuse.conf ] && sudo sed -i 's/#user_allow_other/user_allow_other/g' /etc/fuse.conf || (sudo sh -c 'echo "user_allow_other" > /etc/fuse.conf'); # This is needed for `-o allow_other` option for `goofys`;
+    [ -f /etc/fuse.conf ] && sudo sed -i 's/#user_allow_other/user_allow_other/g' /etc/fuse.conf || (sudo sh -c 'echo "user_allow_other" > /etc/fuse.conf');
+    {%- if remote_identity != 'LOCAL_CREDENTIALS' %}
+    rm ~/.aws/credentials || true;
+    {%- endif %}
+
+
 
 # Command to start ray clusters are now placed in `sky.provision.instance_setup`.
 # We do not need to list it here anymore.

{skypilot_nightly-1.0.0.dev20241125 → skypilot_nightly-1.0.0.dev20241127/skypilot_nightly.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: skypilot-nightly
-Version: 1.0.0.dev20241125
+Version: 1.0.0.dev20241127
 Summary: SkyPilot: An intercloud broker for the clouds
 Author: SkyPilot Team
 License: Apache 2.0