refactor: move all datetime.now() calls to UTC (#1589)

# What does this PR do?

Updated all instances of datetime.now() to use timezone.utc for
consistency in handling time across different systems. This ensures that
timestamps are always in Coordinated Universal Time (UTC), avoiding
issues with time zone discrepancies and promoting uniformity in
time-related data.

Signed-off-by: Sébastien Han <seb@redhat.com>

llama_stack/cli/download.py

@@ -10,7 +10,7 @@ import json
 import os
 import shutil
 from dataclasses import dataclass
-from datetime import datetime
+from datetime import datetime, timezone
 from functools import partial
 from pathlib import Path
 from typing import Dict, List, Optional
@@ -404,7 +404,7 @@ def _download_from_manifest(manifest_file: str, max_concurrent_downloads: int):
         d = json.load(f)
         manifest = Manifest(**d)
 
-    if datetime.now() > manifest.expires_on:
+    if datetime.now(timezone.utc) > manifest.expires_on:
         raise ValueError(f"Manifest URLs have expired on {manifest.expires_on}")
 
     console = Console()

llama_stack/models/llama/llama3/prompt_templates/system_prompts.py

@@ -34,7 +34,9 @@ class SystemDefaultGenerator(PromptTemplateGeneratorBase):
         )
         return PromptTemplate(
             template_str.lstrip("\n"),
-            {"today": datetime.now().strftime("%d %B %Y")},
+            {
+                "today": datetime.now().strftime("%d %B %Y")  # noqa: DTZ005 - we don't care about timezones here since we are displaying the date
+            },
         )
 
     def data_examples(self) -> List[Any]:

llama_stack/providers/inline/agents/meta_reference/agent_instance.py

@@ -11,7 +11,7 @@ import re
 import secrets
 import string
 import uuid
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import AsyncGenerator, List, Optional, Union
 from urllib.parse import urlparse
 
@@ -239,7 +239,7 @@ class ChatAgent(ShieldRunnerMixin):
             in_progress_tool_call_step = await self.storage.get_in_progress_tool_call_step(
                 request.session_id, request.turn_id
             )
-            now = datetime.now().astimezone().isoformat()
+            now = datetime.now(timezone.utc).isoformat()
             tool_execution_step = ToolExecutionStep(
                 step_id=(in_progress_tool_call_step.step_id if in_progress_tool_call_step else str(uuid.uuid4())),
                 turn_id=request.turn_id,
@@ -264,7 +264,7 @@ class ChatAgent(ShieldRunnerMixin):
             start_time = last_turn.started_at
         else:
             messages.extend(request.messages)
-            start_time = datetime.now().astimezone().isoformat()
+            start_time = datetime.now(timezone.utc).isoformat()
             input_messages = request.messages
 
         output_message = None
@@ -295,7 +295,7 @@ class ChatAgent(ShieldRunnerMixin):
             input_messages=input_messages,
             output_message=output_message,
             started_at=start_time,
-            completed_at=datetime.now().astimezone().isoformat(),
+            completed_at=datetime.now(timezone.utc).isoformat(),
             steps=steps,
         )
         await self.storage.add_turn_to_session(request.session_id, turn)
@@ -386,7 +386,7 @@ class ChatAgent(ShieldRunnerMixin):
                 return
 
             step_id = str(uuid.uuid4())
-            shield_call_start_time = datetime.now().astimezone().isoformat()
+            shield_call_start_time = datetime.now(timezone.utc).isoformat()
             try:
                 yield AgentTurnResponseStreamChunk(
                     event=AgentTurnResponseEvent(
@@ -410,7 +410,7 @@ class ChatAgent(ShieldRunnerMixin):
                                 turn_id=turn_id,
                                 violation=e.violation,
                                 started_at=shield_call_start_time,
-                                completed_at=datetime.now().astimezone().isoformat(),
+                                completed_at=datetime.now(timezone.utc).isoformat(),
                             ),
                         )
                     )
@@ -433,7 +433,7 @@ class ChatAgent(ShieldRunnerMixin):
                             turn_id=turn_id,
                             violation=None,
                             started_at=shield_call_start_time,
-                            completed_at=datetime.now().astimezone().isoformat(),
+                            completed_at=datetime.now(timezone.utc).isoformat(),
                         ),
                     )
                 )
@@ -472,7 +472,7 @@ class ChatAgent(ShieldRunnerMixin):
             client_tools[tool.name] = tool
         while True:
             step_id = str(uuid.uuid4())
-            inference_start_time = datetime.now().astimezone().isoformat()
+            inference_start_time = datetime.now(timezone.utc).isoformat()
             yield AgentTurnResponseStreamChunk(
                 event=AgentTurnResponseEvent(
                     payload=AgentTurnResponseStepStartPayload(
@@ -582,7 +582,7 @@ class ChatAgent(ShieldRunnerMixin):
                             turn_id=turn_id,
                             model_response=copy.deepcopy(message),
                             started_at=inference_start_time,
-                            completed_at=datetime.now().astimezone().isoformat(),
+                            completed_at=datetime.now(timezone.utc).isoformat(),
                         ),
                     )
                 )
@@ -653,7 +653,7 @@ class ChatAgent(ShieldRunnerMixin):
                             turn_id=turn_id,
                             tool_calls=[tool_call],
                             tool_responses=[],
-                            started_at=datetime.now().astimezone().isoformat(),
+                            started_at=datetime.now(timezone.utc).isoformat(),
                         ),
                     )
                     yield message
@@ -670,7 +670,7 @@ class ChatAgent(ShieldRunnerMixin):
                         "input": message.model_dump_json(),
                     },
                 ) as span:
-                    tool_execution_start_time = datetime.now().astimezone().isoformat()
+                    tool_execution_start_time = datetime.now(timezone.utc).isoformat()
                     tool_call = message.tool_calls[0]
                     tool_result = await self.execute_tool_call_maybe(
                         session_id,
@@ -708,7 +708,7 @@ class ChatAgent(ShieldRunnerMixin):
                                     )
                                 ],
                                 started_at=tool_execution_start_time,
-                                completed_at=datetime.now().astimezone().isoformat(),
+                                completed_at=datetime.now(timezone.utc).isoformat(),
                             ),
                         )
                     )

llama_stack/providers/inline/agents/meta_reference/persistence.py

@@ -7,7 +7,7 @@
 import json
 import logging
 import uuid
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import List, Optional
 
 from pydantic import BaseModel
@@ -36,7 +36,7 @@ class AgentPersistence:
         session_info = AgentSessionInfo(
             session_id=session_id,
             session_name=name,
-            started_at=datetime.now(),
+            started_at=datetime.now(timezone.utc),
         )
         await self.kvstore.set(
             key=f"session:{self.agent_id}:{session_id}",

llama_stack/providers/inline/post_training/torchtune/post_training.py

@@ -3,7 +3,7 @@
 #
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import Any, Dict, Optional
 
 from llama_stack.apis.datasetio import DatasetIO
@@ -64,7 +64,7 @@ class TorchtunePostTrainingImpl:
         job_status_response = PostTrainingJobStatusResponse(
             job_uuid=job_uuid,
             status=JobStatus.scheduled,
-            scheduled_at=datetime.now(),
+            scheduled_at=datetime.now(timezone.utc),
         )
         self.jobs[job_uuid] = job_status_response
 
@@ -84,7 +84,7 @@ class TorchtunePostTrainingImpl:
                 )
 
                 job_status_response.status = JobStatus.in_progress
-                job_status_response.started_at = datetime.now()
+                job_status_response.started_at = datetime.now(timezone.utc)
 
                 await recipe.setup()
                 resources_allocated, checkpoints = await recipe.train()
@@ -93,7 +93,7 @@ class TorchtunePostTrainingImpl:
                 job_status_response.resources_allocated = resources_allocated
                 job_status_response.checkpoints = checkpoints
                 job_status_response.status = JobStatus.completed
-                job_status_response.completed_at = datetime.now()
+                job_status_response.completed_at = datetime.now(timezone.utc)
 
             except Exception:
                 job_status_response.status = JobStatus.failed

llama_stack/providers/inline/post_training/torchtune/recipes/lora_finetuning_single_device.py

@@ -8,7 +8,7 @@ import gc
 import logging
 import os
 import time
-from datetime import datetime
+from datetime import datetime, timezone
 from functools import partial
 from pathlib import Path
 from typing import Any, Dict, List, Optional, Tuple
@@ -532,7 +532,7 @@ class LoraFinetuningSingleDevice:
             checkpoint_path = await self.save_checkpoint(epoch=curr_epoch)
             checkpoint = Checkpoint(
                 identifier=f"{self.model_id}-sft-{curr_epoch}",
-                created_at=datetime.now(),
+                created_at=datetime.now(timezone.utc),
                 epoch=curr_epoch,
                 post_training_job_id=self.job_uuid,
                 path=checkpoint_path,

llama_stack/providers/inline/telemetry/meta_reference/console_span_processor.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 import json
-from datetime import datetime
+from datetime import datetime, timezone
 
 from opentelemetry.sdk.trace import ReadableSpan
 from opentelemetry.sdk.trace.export import SpanProcessor
@@ -34,7 +34,7 @@ class ConsoleSpanProcessor(SpanProcessor):
         if span.attributes and span.attributes.get("__autotraced__"):
             return
 
-        timestamp = datetime.utcfromtimestamp(span.start_time / 1e9).strftime("%H:%M:%S.%f")[:-3]
+        timestamp = datetime.fromtimestamp(span.start_time / 1e9, tz=timezone.utc).strftime("%H:%M:%S.%f")[:-3]
 
         print(
             f"{COLORS['dim']}{timestamp}{COLORS['reset']} "
@@ -46,7 +46,7 @@ class ConsoleSpanProcessor(SpanProcessor):
         if span.attributes and span.attributes.get("__autotraced__"):
             return
 
-        timestamp = datetime.utcfromtimestamp(span.end_time / 1e9).strftime("%H:%M:%S.%f")[:-3]
+        timestamp = datetime.fromtimestamp(span.end_time / 1e9, tz=timezone.utc).strftime("%H:%M:%S.%f")[:-3]
 
         span_context = (
             f"{COLORS['dim']}{timestamp}{COLORS['reset']} "
@@ -74,7 +74,7 @@ class ConsoleSpanProcessor(SpanProcessor):
                 print(f"    {COLORS['dim']}{key}: {str_value}{COLORS['reset']}")
 
         for event in span.events:
-            event_time = datetime.utcfromtimestamp(event.timestamp / 1e9).strftime("%H:%M:%S.%f")[:-3]
+            event_time = datetime.fromtimestamp(event.timestamp / 1e9, tz=timezone.utc).strftime("%H:%M:%S.%f")[:-3]
 
             severity = event.attributes.get("severity", "info")
             message = event.attributes.get("message", event.name)

llama_stack/providers/inline/telemetry/meta_reference/sqlite_span_processor.py

@@ -8,7 +8,7 @@ import json
 import os
 import sqlite3
 import threading
-from datetime import datetime
+from datetime import datetime, timezone
 
 from opentelemetry.sdk.trace import SpanProcessor
 from opentelemetry.trace import Span
@@ -124,8 +124,8 @@ class SQLiteSpanProcessor(SpanProcessor):
                     trace_id,
                     service_name,
                     (span_id if not parent_span_id else None),
-                    datetime.fromtimestamp(span.start_time / 1e9).isoformat(),
+                    datetime.fromtimestamp(span.start_time / 1e9, timezone.utc).isoformat(),
-                    datetime.fromtimestamp(span.end_time / 1e9).isoformat(),
+                    datetime.fromtimestamp(span.end_time / 1e9, timezone.utc).isoformat(),
                 ),
             )
 
@@ -143,8 +143,8 @@ class SQLiteSpanProcessor(SpanProcessor):
                     trace_id,
                     parent_span_id,
                     span.name,
-                    datetime.fromtimestamp(span.start_time / 1e9).isoformat(),
+                    datetime.fromtimestamp(span.start_time / 1e9, timezone.utc).isoformat(),
-                    datetime.fromtimestamp(span.end_time / 1e9).isoformat(),
+                    datetime.fromtimestamp(span.end_time / 1e9, timezone.utc).isoformat(),
                     json.dumps(dict(span.attributes)),
                     span.status.status_code.name,
                     span.kind.name,
@@ -161,7 +161,7 @@ class SQLiteSpanProcessor(SpanProcessor):
                     (
                         span_id,
                         event.name,
-                        datetime.fromtimestamp(event.timestamp / 1e9).isoformat(),
+                        datetime.fromtimestamp(event.timestamp / 1e9, timezone.utc).isoformat(),
                         json.dumps(dict(event.attributes)),
                     ),
                 )

llama_stack/providers/inline/tool_runtime/code_interpreter/code_execution.py

@@ -168,7 +168,7 @@ def process_matplotlib_response(response, matplotlib_dump_dir: str):
     image_paths = []
     for i, img in enumerate(images):
         # create new directory for each day to better organize data:
-        dump_dname = datetime.today().strftime("%Y-%m-%d")
+        dump_dname = datetime.today().strftime("%Y-%m-%d")  # noqa: DTZ002 - we don't care about timezones here since we are displaying the date
         dump_dpath = Path(matplotlib_dump_dir, dump_dname)
         dump_dpath.mkdir(parents=True, exist_ok=True)
         # save image into a file

llama_stack/providers/utils/telemetry/tracing.py

@@ -11,7 +11,7 @@ import logging
 import queue
 import threading
 import uuid
-from datetime import datetime
+from datetime import datetime, timezone
 from functools import wraps
 from typing import Any, Callable, Dict, List, Optional
 
@@ -86,7 +86,7 @@ class TraceContext:
             span_id=generate_short_uuid(),
             trace_id=self.trace_id,
             name=name,
-            start_time=datetime.now(),
+            start_time=datetime.now(timezone.utc),
             parent_span_id=current_span.span_id if current_span else None,
             attributes=attributes,
         )
@@ -203,7 +203,7 @@ class TelemetryHandler(logging.Handler):
             UnstructuredLogEvent(
                 trace_id=span.trace_id,
                 span_id=span.span_id,
-                timestamp=datetime.now(),
+                timestamp=datetime.now(timezone.utc),
                 message=self.format(record),
                 severity=severity(record.levelname),
             )

pyproject.toml

@@ -132,6 +132,7 @@ select = [
     "N",   # Naming
     "W",   # Warnings
     "I",   # isort
+    "DTZ", # datetime rules
 ]
 ignore = [
     # The following ignores are desired by the project maintainers.
@@ -145,6 +146,10 @@ ignore = [
     "C901", # Complexity of the function is too high
 ]
 
+# Ignore the following errors for the following files
+[tool.ruff.lint.per-file-ignores]
+"tests/**/*.py" = ["DTZ"] # Ignore datetime rules for tests
+
 [tool.mypy]
 mypy_path = ["llama_stack"]
 packages = ["llama_stack"]