Added telemetry system to get information about issues.

ptlinux/command.py

@@ -15,6 +15,8 @@ import threading
 import time
 from typing import Callable, Mapping, Sequence, TextIO
 
+from .telemetry import current_telemetry
+
 
 OutputLineFilter = Callable[[str], str | None]
 MAX_CAPTURE_CHARACTERS = 1024 * 1024
@@ -85,9 +87,16 @@ class CommandError(RuntimeError):
 class CommandRunner:
 	"""Run argument arrays with timeouts and append-only JSONL logging."""
 
-	def __init__(self, log_path: Path, *, dry_run: bool = False) -> None:
+	def __init__(
+		self,
+		log_path: Path,
+		*,
+		dry_run: bool = False,
+		operation: str | None = None,
+	) -> None:
 		self.log_path = log_path
 		self.dry_run = dry_run
+		self.operation = operation
 
 	def run(
 		self,
@@ -100,6 +109,7 @@ class CommandRunner:
 		stream_output: bool = False,
 		output_line_filter: OutputLineFilter | None = None,
 		pass_fds: Sequence[int] = (),
+		telemetry_stage: str | None = None,
 	) -> CommandOutcome:
 		arguments = [str(value) for value in argv]
 		if not arguments:
@@ -169,6 +179,18 @@ class CommandRunner:
 					dry_run=False,
 				)
 		self._record(outcome, cwd, overrides)
+		if outcome.timed_out or outcome.returncode != 0:
+			telemetry = current_telemetry()
+			if telemetry is not None:
+				telemetry.report_subprocess_failure(
+					"subprocess_failed",
+					outcome,
+					command=outcome.argv[:1],
+					metadata={
+						"operation": self.operation or "subprocess",
+						**({"stage": telemetry_stage} if telemetry_stage else {}),
+					},
+				)
 		if check and (outcome.timed_out or outcome.returncode != 0):
 			raise CommandError(outcome)
 		return outcome