sagemaker-hyperpod 3.0.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (66) hide show
  1. sagemaker/__init__.py +0 -0
  2. sagemaker/hyperpod/__init__.py +2 -0
  3. sagemaker/hyperpod/cli/__init__.py +0 -0
  4. sagemaker/hyperpod/cli/clients/__init__.py +12 -0
  5. sagemaker/hyperpod/cli/clients/kubernetes_client.py +365 -0
  6. sagemaker/hyperpod/cli/commands/__init__.py +0 -0
  7. sagemaker/hyperpod/cli/commands/cluster.py +627 -0
  8. sagemaker/hyperpod/cli/commands/inference.py +659 -0
  9. sagemaker/hyperpod/cli/commands/training.py +347 -0
  10. sagemaker/hyperpod/cli/constants/__init__.py +0 -0
  11. sagemaker/hyperpod/cli/constants/command_constants.py +105 -0
  12. sagemaker/hyperpod/cli/constants/exception_constants.py +13 -0
  13. sagemaker/hyperpod/cli/constants/hyperpod_instance_types.py +100 -0
  14. sagemaker/hyperpod/cli/constants/kueue_constants.py +17 -0
  15. sagemaker/hyperpod/cli/constants/pytorch_constants.py +15 -0
  16. sagemaker/hyperpod/cli/hyp_cli.py +132 -0
  17. sagemaker/hyperpod/cli/inference_utils.py +121 -0
  18. sagemaker/hyperpod/cli/service/__init__.py +12 -0
  19. sagemaker/hyperpod/cli/service/cancel_training_job.py +70 -0
  20. sagemaker/hyperpod/cli/service/discover_namespaces.py +108 -0
  21. sagemaker/hyperpod/cli/service/exec_command.py +75 -0
  22. sagemaker/hyperpod/cli/service/get_logs.py +136 -0
  23. sagemaker/hyperpod/cli/service/get_namespaces.py +59 -0
  24. sagemaker/hyperpod/cli/service/get_training_job.py +110 -0
  25. sagemaker/hyperpod/cli/service/list_pods.py +138 -0
  26. sagemaker/hyperpod/cli/service/list_training_jobs.py +178 -0
  27. sagemaker/hyperpod/cli/service/self_subject_access_review.py +47 -0
  28. sagemaker/hyperpod/cli/telemetry/__init__.py +12 -0
  29. sagemaker/hyperpod/cli/telemetry/user_agent.py +36 -0
  30. sagemaker/hyperpod/cli/templates/__init__.py +12 -0
  31. sagemaker/hyperpod/cli/templates/k8s_pytorch_job_template.py +74 -0
  32. sagemaker/hyperpod/cli/training_utils.py +200 -0
  33. sagemaker/hyperpod/cli/utils.py +183 -0
  34. sagemaker/hyperpod/cli/validators/__init__.py +12 -0
  35. sagemaker/hyperpod/cli/validators/cluster_validator.py +52 -0
  36. sagemaker/hyperpod/cli/validators/job_validator.py +346 -0
  37. sagemaker/hyperpod/cli/validators/validator.py +75 -0
  38. sagemaker/hyperpod/common/__init__.py +0 -0
  39. sagemaker/hyperpod/common/config/__init__.py +1 -0
  40. sagemaker/hyperpod/common/config/metadata.py +18 -0
  41. sagemaker/hyperpod/common/utils.py +299 -0
  42. sagemaker/hyperpod/inference/__init__.py +11 -0
  43. sagemaker/hyperpod/inference/config/__init__.py +0 -0
  44. sagemaker/hyperpod/inference/config/constants.py +18 -0
  45. sagemaker/hyperpod/inference/config/hp_endpoint_config.py +767 -0
  46. sagemaker/hyperpod/inference/config/hp_jumpstart_endpoint_config.py +572 -0
  47. sagemaker/hyperpod/inference/hp_endpoint.py +203 -0
  48. sagemaker/hyperpod/inference/hp_endpoint_base.py +223 -0
  49. sagemaker/hyperpod/inference/hp_jumpstart_endpoint.py +232 -0
  50. sagemaker/hyperpod/observability/MonitoringConfig.py +10 -0
  51. sagemaker/hyperpod/observability/__init__.py +0 -0
  52. sagemaker/hyperpod/observability/constants.py +2 -0
  53. sagemaker/hyperpod/observability/utils.py +64 -0
  54. sagemaker/hyperpod/training/__init__.py +9 -0
  55. sagemaker/hyperpod/training/config/__init__.py +1 -0
  56. sagemaker/hyperpod/training/config/hyperpod_pytorch_job_config.py +2977 -0
  57. sagemaker/hyperpod/training/config/hyperpod_pytorch_job_status.py +3083 -0
  58. sagemaker/hyperpod/training/hyperpod_pytorch_job.py +253 -0
  59. sagemaker_hyperpod-3.0.0.dist-info/METADATA +534 -0
  60. sagemaker_hyperpod-3.0.0.dist-info/RECORD +66 -0
  61. sagemaker_hyperpod-3.0.0.dist-info/WHEEL +5 -0
  62. sagemaker_hyperpod-3.0.0.dist-info/entry_points.txt +2 -0
  63. sagemaker_hyperpod-3.0.0.dist-info/licenses/LICENSE +175 -0
  64. sagemaker_hyperpod-3.0.0.dist-info/licenses/NOTICE +1 -0
  65. sagemaker_hyperpod-3.0.0.dist-info/top_level.txt +1 -0
  66. sagemaker_hyperpod-3.0.0.dist-info/zip-safe +1 -0
@@ -0,0 +1,253 @@
1
+ from pydantic import ConfigDict, Field
2
+ from sagemaker.hyperpod.training.config.hyperpod_pytorch_job_config import (
3
+ _HyperPodPytorchJob,
4
+ )
5
+ from sagemaker.hyperpod.training.config.hyperpod_pytorch_job_status import (
6
+ HyperPodPytorchJobStatus,
7
+ )
8
+ from sagemaker.hyperpod.common.config.metadata import Metadata
9
+ from kubernetes import client, config
10
+ from typing import List, Optional, ClassVar
11
+ from sagemaker.hyperpod.common.utils import (
12
+ handle_exception,
13
+ get_default_namespace,
14
+ setup_logging,
15
+ )
16
+ import yaml
17
+ import logging
18
+
19
+ TRAINING_GROUP = "sagemaker.amazonaws.com"
20
+ API_VERSION = "v1"
21
+ PLURAL = "hyperpodpytorchjobs"
22
+ KIND = "HyperPodPyTorchJob"
23
+
24
+
25
+ class HyperPodPytorchJob(_HyperPodPytorchJob):
26
+ is_kubeconfig_loaded: ClassVar[bool] = False
27
+
28
+ model_config = ConfigDict(extra="forbid")
29
+
30
+ metadata: Metadata = Field(
31
+ description="The metadata of the HyperPodPytorchJob",
32
+ )
33
+
34
+ status: Optional[HyperPodPytorchJobStatus] = Field(
35
+ default=None, description="The status of the HyperPodPytorchJob"
36
+ )
37
+
38
+ @classmethod
39
+ def verify_kube_config(cls):
40
+ if not cls.is_kubeconfig_loaded:
41
+ config.load_kube_config()
42
+ cls.is_kubeconfig_loaded = True
43
+
44
+ @classmethod
45
+ def get_logger(cls):
46
+ return logging.getLogger(__name__)
47
+
48
+ def create(self, debug=False):
49
+ self.verify_kube_config()
50
+
51
+ logger = self.get_logger()
52
+ logger = setup_logging(logger, debug)
53
+
54
+ spec = _HyperPodPytorchJob(**self.model_dump(by_alias=True, exclude_none=True))
55
+
56
+ if not self.metadata.namespace:
57
+ self.metadata.namespace = get_default_namespace()
58
+
59
+ config = {
60
+ "apiVersion": f"{TRAINING_GROUP}/{API_VERSION}",
61
+ "kind": KIND,
62
+ "metadata": self.metadata.model_dump(exclude_none=True),
63
+ "spec": spec.model_dump(exclude_none=True),
64
+ }
65
+
66
+ custom_api = client.CustomObjectsApi()
67
+ logger.debug(
68
+ "Deploying HyperPodPytorchJob with config:\n%s",
69
+ yaml.dump(config),
70
+ )
71
+
72
+ try:
73
+ custom_api.create_namespaced_custom_object(
74
+ group=TRAINING_GROUP,
75
+ version=API_VERSION,
76
+ namespace=self.metadata.namespace,
77
+ plural=PLURAL,
78
+ body=config,
79
+ )
80
+ logger.info("Successfully submitted HyperPodPytorchJob!")
81
+ except Exception as e:
82
+ logger.error(f"Failed to create HyperPodPytorchJob {self.metadata.name}!")
83
+ handle_exception(e, self.metadata.name, self.metadata.namespace)
84
+
85
+ @classmethod
86
+ def list(cls, namespace=None) -> List["HyperPodPytorchJob"]:
87
+ cls.verify_kube_config()
88
+
89
+ if namespace is None:
90
+ namespace = get_default_namespace()
91
+
92
+ logger = cls.get_logger()
93
+ logger = setup_logging(logger)
94
+
95
+ custom_api = client.CustomObjectsApi()
96
+
97
+ try:
98
+ hp_job_list = custom_api.list_namespaced_custom_object(
99
+ group=TRAINING_GROUP,
100
+ version=API_VERSION,
101
+ namespace=namespace,
102
+ plural=PLURAL,
103
+ )
104
+ return _load_hp_job_list(hp_job_list)
105
+ except Exception as e:
106
+ logger.error(f"Failed to list HyperpodPytorchJobs!")
107
+ handle_exception(e, "", namespace)
108
+
109
+ def delete(self):
110
+ self.verify_kube_config()
111
+
112
+ logger = self.get_logger()
113
+ logger = setup_logging(logger)
114
+
115
+ custom_api = client.CustomObjectsApi()
116
+
117
+ try:
118
+ custom_api.delete_namespaced_custom_object(
119
+ group=TRAINING_GROUP,
120
+ version=API_VERSION,
121
+ namespace=self.metadata.namespace,
122
+ plural=PLURAL,
123
+ name=self.metadata.name,
124
+ )
125
+ logger.info(f"Successful deleted HyperPodPytorchJob!")
126
+ except Exception as e:
127
+ logger.error(f"Failed to delete HyperPodPytorchJob {self.metadata.name}!")
128
+ handle_exception(e, self.metadata.name, self.metadata.namespace)
129
+
130
+ @classmethod
131
+ def get(cls, name, namespace=None) -> "HyperPodPytorchJob":
132
+ cls.verify_kube_config()
133
+
134
+ if namespace is None:
135
+ namespace = get_default_namespace()
136
+
137
+ logger = cls.get_logger()
138
+ logger = setup_logging(logger)
139
+
140
+ custom_api = client.CustomObjectsApi()
141
+
142
+ try:
143
+ response = custom_api.get_namespaced_custom_object(
144
+ group=TRAINING_GROUP,
145
+ version=API_VERSION,
146
+ namespace=namespace,
147
+ plural=PLURAL,
148
+ name=name,
149
+ )
150
+ return _load_hp_job(response)
151
+ except Exception as e:
152
+ logger.error(f"Failed to describe HyperPodPytorchJob {name}: {e}")
153
+ handle_exception(e, name, namespace)
154
+
155
+ def refresh(self) -> "HyperPodPytorchJob":
156
+ self.verify_kube_config()
157
+
158
+ logger = self.get_logger()
159
+ logger = setup_logging(logger)
160
+
161
+ custom_api = client.CustomObjectsApi()
162
+
163
+ try:
164
+ response = custom_api.get_namespaced_custom_object(
165
+ group=TRAINING_GROUP,
166
+ version=API_VERSION,
167
+ namespace=self.metadata.namespace,
168
+ plural=PLURAL,
169
+ name=self.metadata.name,
170
+ )
171
+ self.status = HyperPodPytorchJobStatus.model_validate(
172
+ response["status"], by_name=True
173
+ )
174
+ except Exception as e:
175
+ logger.error(f"Failed to refresh HyperPodPytorchJob {self.metadata.name}!")
176
+ handle_exception(e, self.metadata.name, self.metadata.namespace)
177
+
178
+ def list_pods(self) -> List[str]:
179
+ self.verify_kube_config()
180
+
181
+ logger = self.get_logger()
182
+ logger = setup_logging(logger)
183
+
184
+ try:
185
+ config.load_kube_config()
186
+ v1 = client.CoreV1Api()
187
+
188
+ response = v1.list_namespaced_pod(self.metadata.namespace)
189
+ pods = []
190
+
191
+ for pod in response.items:
192
+ if pod.metadata.name.startswith(f"{self.metadata.name}-pod"):
193
+ pods.append(pod.metadata.name)
194
+ return pods
195
+ except Exception as e:
196
+ logger.error(f"Failed to list pod in namespace {self.metadata.namespace}!")
197
+ handle_exception(e, self.metadata.name, self.metadata.namespace)
198
+
199
+ def get_logs_from_pod(self, pod_name: str, container: Optional[str] = None) -> str:
200
+ self.verify_kube_config()
201
+
202
+ logger = self.get_logger()
203
+ logger = setup_logging(logger)
204
+
205
+ if container is None:
206
+ # If container name is not set, get logs from the first container in the pod
207
+ container = self.replicaSpecs[0].template.spec.containers[0].name
208
+
209
+ try:
210
+ config.load_kube_config()
211
+ v1 = client.CoreV1Api()
212
+
213
+ logs = v1.read_namespaced_pod_log(
214
+ name=pod_name,
215
+ namespace=self.metadata.namespace,
216
+ timestamps=True,
217
+ container=container,
218
+ )
219
+ return logs
220
+ except Exception as e:
221
+ logger.error(f"Failed to get logs from pod {pod_name}!")
222
+ handle_exception(e, self.metadata.name, self.metadata.namespace)
223
+
224
+
225
+ def _load_hp_job(response: dict) -> HyperPodPytorchJob:
226
+ name = response["metadata"]["name"]
227
+ namespace = response["metadata"]["namespace"]
228
+
229
+ spec = _HyperPodPytorchJob.model_validate(response["spec"], by_name=True)
230
+ metadata = Metadata(name=name, namespace=namespace)
231
+
232
+ if "status" in response:
233
+ status = HyperPodPytorchJobStatus.model_validate(
234
+ response["status"], by_name=True
235
+ )
236
+
237
+ else:
238
+ status = None
239
+
240
+ job = HyperPodPytorchJob(
241
+ metadata=metadata,
242
+ status=status,
243
+ **spec.model_dump(by_alias=True, exclude_none=True),
244
+ )
245
+ return job
246
+
247
+
248
+ def _load_hp_job_list(response: dict) -> List[HyperPodPytorchJob]:
249
+ job_list = []
250
+ for hp_job in response["items"]:
251
+ job = _load_hp_job(hp_job)
252
+ job_list.append(job)
253
+ return job_list