skypilot-nightly 1.0.0.dev20250509__py3-none-any.whl → 1.0.0.dev20251107__py3-none-any.whl

sky/templates/aws-ray.yml.j2

@@ -19,7 +19,7 @@ docker:
     username: |-
       {{docker_login_config.username}}
     password: |-
-      {{docker_login_config.password}}
+      {{docker_login_config.password | indent(6) }}
     server: |-
       {{docker_login_config.server}}
   {%- endif %}
@@ -48,9 +48,10 @@ provider:
   # The upper-level SkyPilot code has make sure there will not be resource
   # leakage.
   disable_launch_config_check: true
+  max_efa_interfaces: {{max_efa_interfaces}}
 
 auth:
-  ssh_user: ubuntu
+  ssh_user: {{ssh_user}}
   ssh_private_key: {{ssh_private_key}}
 {% if ssh_proxy_command is not none %}
   ssh_proxy_command: {{ssh_proxy_command}}
@@ -68,7 +69,7 @@ available_node_types:
       ImageId: {{image_id}}  # Deep Learning AMI (Ubuntu 18.04); see aws.py.
       # https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/ec2.html#EC2.ServiceResource.create_instances
       BlockDeviceMappings:
-        - DeviceName: /dev/sda1
+        - DeviceName: {{root_device_name}}
           Ebs:
             VolumeSize: {{disk_size}}
             VolumeType: {{disk_tier}}
@@ -131,6 +132,12 @@ available_node_types:
           - systemctl disable apt-daily.timer apt-daily-upgrade.timer unattended-upgrades.service
           - systemctl mask apt-daily.service apt-daily-upgrade.service unattended-upgrades.service
           - systemctl daemon-reload
+        {%- if runcmd %}
+        runcmd:
+        {%- for cmd in runcmd %}
+          - {{cmd}}
+        {%- endfor %}
+        {%- endif %}
       TagSpecifications:
         - ResourceType: instance
           Tags:

sky/templates/azure-ray.yml.j2

@@ -19,7 +19,7 @@ docker:
     username: |-
       {{docker_login_config.username}}
     password: |-
-      {{docker_login_config.password}}
+      {{docker_login_config.password | indent(6) }}
     server: |-
       {{docker_login_config.server}}
   {%- endif %}

sky/templates/do-ray.yml.j2

@@ -19,7 +19,7 @@ docker:
     username: |-
       {{docker_login_config.username}}
     password: |-
-      {{docker_login_config.password}}
+      {{docker_login_config.password | indent(6) }}
     server: |-
       {{docker_login_config.server}}
   {%- endif %}

sky/templates/gcp-ray.yml.j2

@@ -69,6 +69,15 @@ provider:
 {%- if enable_gvnic %}
   enable_gvnic: {{ enable_gvnic }}
 {%- endif %}
+{%- if enable_gpu_direct %}
+  enable_gpu_direct: {{ enable_gpu_direct }}
+{%- endif %}
+{%- if placement_policy %}
+  placement_policy: {{ placement_policy }}
+{%- endif %}
+{%- if network_tier %}
+  network_tier: {{ network_tier }}
+{%- endif %}
 
 auth:
   ssh_user: gcpuser
@@ -103,12 +112,27 @@ available_node_types:
 {%- if tpu_vm %}
       acceleratorType: {{tpu_type}}
       runtimeVersion: {{runtime_version}}
+    {%- if volumes %}
+      dataDisks:
+      {%- for volume in volumes %}
+        {%- if volume.source %}
+        - sourceDisk: {{volume.source}}
+        {%- endif %}
+        {%- if volume.attach_mode %}
+          mode: {{volume.attach_mode}}
+        {%- endif %}
+      {%- endfor %}
+    {%- endif %}
       metadata:
         # TPU VM's metadata has different format than normal VMs.
         # After replacing the variables, this will become username:ssh_public_key_content.
         # This is a specific syntax required by GCP https://cloud.google.com/compute/docs/connect/add-ssh-keys
         ssh-keys: |-
               skypilot:ssh_user:skypilot:ssh_public_key_content
+      {%- if user_data is not none %}
+        startup-script: |-
+          {{ user_data | indent(10) }}
+      {%- endif %}
   {%- if use_spot %}
       schedulingConfig:
         preemptible: true
@@ -132,6 +156,34 @@ available_node_types:
           {%- if disk_iops %}
             provisionedIops: {{disk_iops}}
           {%- endif %}
+      {%- for volume in volumes %}
+        - boot: false
+          autoDelete: {{volume.auto_delete}}
+          type: {{volume.storage_type}}
+          deviceName: {{volume.device_name}}
+        {%- if volume.source %}
+          source: {{volume.source}}
+        {%- endif %}
+        {%- if volume.attach_mode %}
+          mode: {{volume.attach_mode}}
+        {%- endif %}
+        {%- if volume.interface_type %}
+          interface: {{volume.interface_type}}
+        {%- endif %}
+        {%- if volume.disk_tier %}
+          initializeParams:
+            diskType: zones/{{zones}}/diskTypes/{{volume.disk_tier}}
+        {%- endif %}
+          {%- if volume.disk_name %}
+            diskName: {{volume.disk_name}}
+          {%- endif %}
+          {%- if volume.disk_size %}
+            diskSizeGb: {{volume.disk_size}}
+          {%- endif %}
+          {%- if volume.iops %}
+            provisionedIops: {{volume.iops}}
+          {%- endif %}
+      {%- endfor %}
   {%- if gpu is not none %}
       guestAccelerators:
         - acceleratorType: projects/{{gcp_project_id}}/zones/{{zones}}/acceleratorTypes/{{gpu}}
@@ -148,6 +200,11 @@ available_node_types:
           - key: install-nvidia-driver
             value: "True"
   {%- endif %}
+  {%- if user_data is not none %}
+          - key: user-data
+            value: |-
+              {{ user_data | indent(14) }}
+  {%- endif %}
   {%- if use_spot or gpu is not none %}
       scheduling:
   {%- if use_spot %}

sky/templates/hyperbolic-ray.yml.j2

@@ -0,0 +1,67 @@
+# Ray cluster config template for Hyperbolic Cloud
+
+cluster_name: {{cluster_name_on_cloud}}
+
+# Hyperbolic only supports a single node (the head node).
+max_workers: 0
+upscaling_speed: 0
+idle_timeout_minutes: 60
+
+provider:
+  type: external
+  module: sky.provision.hyperbolic
+  region: "default"
+
+auth:
+  ssh_user: ubuntu
+  ssh_private_key: {{ssh_private_key}}
+
+available_node_types:
+  ray_head_default:
+    resources: {}
+    node_config:
+      InstanceType: {{instance_type}}
+
+head_node_type: ray_head_default
+
+# Format: `REMOTE_PATH : LOCAL_PATH`
+file_mounts: {
+  "{{sky_ray_yaml_remote_path}}": "{{sky_ray_yaml_local_path}}",
+  "{{sky_remote_path}}/{{sky_wheel_hash}}": "{{sky_local_path}}",
+{%- for remote_path, local_path in credentials.items() %}
+  "{{remote_path}}": "{{local_path}}",
+{%- endfor %}
+}
+
+rsync_exclude: []
+
+initialization_commands: []
+
+# List of shell commands to run to set up nodes.
+# NOTE: these are very performance-sensitive. Each new item opens/closes an SSH
+# connection, which is expensive. Try your best to co-locate commands into fewer
+# items!
+#
+# Increment the following for catching performance bugs easier:
+#   current num items (num SSH connections): 1
+setup_commands:
+  # Disable unattended-upgrades and handle apt-get locks
+  # Install patch utility for Ray
+  # Install conda and Ray
+  # Set system limits for Ray performance (nofile and TasksMax)
+  - {%- for initial_setup_command in initial_setup_commands %}
+    {{ initial_setup_command }}
+    {%- endfor %}
+    sudo systemctl stop unattended-upgrades || true;
+    sudo systemctl disable unattended-upgrades || true;
+    sudo sed -i 's/Unattended-Upgrade "1"/Unattended-Upgrade "0"/g' /etc/apt/apt.conf.d/20auto-upgrades || true;
+    sudo kill -9 `sudo lsof /var/lib/dpkg/lock-frontend | awk '{print $2}' | tail -n 1` || true;
+    sudo pkill -9 apt-get;
+    sudo pkill -9 dpkg;
+    sudo dpkg --configure -a;
+    which patch > /dev/null || sudo apt install -y patch;
+    {{ conda_installation_commands }}
+    {{ ray_skypilot_installation_commands }}
+    sudo bash -c 'rm -rf /etc/security/limits.d; echo "* soft nofile 1048576" >> /etc/security/limits.conf; echo "* hard nofile 1048576" >> /etc/security/limits.conf';
+    sudo grep -e '^DefaultTasksMax' /etc/systemd/system.conf || (sudo bash -c 'echo "DefaultTasksMax=infinity" >> /etc/systemd/system.conf'); sudo systemctl set-property user-$(id -u $(whoami)).slice TasksMax=infinity; sudo systemctl daemon-reload;
+    {{ ssh_max_sessions_config }}

sky/templates/jobs-controller.yaml.j2

@@ -3,6 +3,7 @@
 name: {{dag_name}}
 
 file_mounts:
+  {{remote_original_user_yaml_path}}: {{original_user_dag_path}}
   {{remote_user_yaml_path}}: {{user_yaml_path}}
   {%- if local_user_config_path is not none %}
   {{remote_user_config_path}}: {{local_user_config_path}}
@@ -14,6 +15,12 @@ file_mounts:
   {{controller_file_mount_path}}: {{local_file_mount_path}}
   {%- endfor %}
 
+# NOTE(dev): This needs to be a subset of sky/templates/sky-serve-controller.yaml.j2.
+# It is because we use the --fast flag to submit jobs and no --fast flag to launch pools.
+# So when we launch a new pool, it will install the required dependencies.
+# TODO(tian): Add --fast to launch pools as well, and figure out the dependency installation.
+# Maybe in the --fast implementation, we can store the hash of setup commands that used to be
+# run and don't skip setup phase if the hash is different.
 setup: |
   {{ sky_activate_python_env }}
   # Disable the pip version check to avoid the warning message, which makes the
@@ -29,31 +36,13 @@ setup: |
   grep -q 'alias sky-env=' ~/.bashrc || echo 'alias sky-env="{{ sky_activate_python_env }}"' >> ~/.bashrc
   {% endif %}
 
-  # Create systemd service file
-  mkdir -p ~/.config/systemd/user/
-
-  # Create systemd user service file
-  cat << EOF > ~/.config/systemd/user/skypilot-dashboard.service
-  [Unit]
-  Description=SkyPilot Jobs Dashboard
-  After=network.target
-
-  [Service]
-  Environment="SKYPILOT_USER_ID={{ dashboard_user_id }}"
-  Restart=always
-  StandardOutput=append:/home/$USER/.sky/job-dashboard.log
-  StandardError=append:/home/$USER/.sky/job-dashboard.log
-  ExecStart={{ sky_python_cmd }} -m sky.jobs.dashboard.dashboard
-
-  [Install]
-  WantedBy=default.target
-  EOF
-
-  export SKYPILOT_USER_ID="{{ dashboard_user_id }}"
-  {{ dashboard_setup_cmd }}
+  # This is used by the skylet events to check if we are a jobs controller.
+  touch {{job_controller_indicator_file}}
 
 run: |
+  {%- if consolidation_mode_job_id is none %}
   {{ sky_activate_python_env }}
+  {%- endif %}
 
   # Write env vars to a file
   {%- for env_name, env_value in controller_envs.items() %}
@@ -64,9 +53,23 @@ run: |
   # Note: The job is already in the `spot` table, marked as PENDING.
   # CloudVmRayBackend._exec_code_on_head() calls
   # managed_job_codegen.set_pending() before we get here.
-  python -u -m sky.jobs.scheduler {{remote_user_yaml_path}} \
+  {%- if consolidation_mode_job_id is not none %}
+  {{sky_python_cmd}} \
+  {%- else %}
+  python \
+  {%- endif %}
+    -u -m sky.jobs.scheduler {{remote_user_yaml_path}} \
+    --user-yaml-path {{remote_original_user_yaml_path}} \
+    {%- if consolidation_mode_job_id is not none %}
+    --job-id {{consolidation_mode_job_id}} \
+    {%- else %}
     --job-id $SKYPILOT_INTERNAL_JOB_ID \
-    --env-file {{remote_env_file_path}}
+    {%- endif %}
+    --env-file {{remote_env_file_path}} \
+    {%- if pool is not none %}
+    --pool {{pool}} \
+    {%- endif %}
+    --priority {{priority}}
 
 
 envs:

sky/templates/kubernetes-loadbalancer.yml.j2

@@ -12,6 +12,8 @@ service_spec:
       {%- for key, value in annotations.items() %}
       {{ key }}: {{ value|tojson }}
       {%- endfor %}
+      {# Note: It's ok to add cloud-specific annotations here since they will be ignored by other clouds #}
+      service.beta.kubernetes.io/coreweave-load-balancer-type: public
   spec:
     type: LoadBalancer
     selector: