Skip to content

Commit

Permalink
fix: correct signal for Windows
Browse files Browse the repository at this point in the history
Windows uses `signal.SIGTERM` instead of `signal.SIGKILL`.

Signed-off-by: Strobel Maximilian (IFAG PSS SIS SCE ACM) <[email protected]>
  • Loading branch information
maxstrobel committed Jan 17, 2025
1 parent 4c5f051 commit 0f7891b
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion nemo/utils/exp_manager.py
Original file line number Diff line number Diff line change
Expand Up @@ -167,7 +167,7 @@ class FaultToleranceParams:
rank_heartbeat_timeout: Optional[float] = 45.0 * 60.0
calculate_timeouts: bool = True
safety_factor: float = 5.0
rank_termination_signal: signal.Signals = signal.SIGKILL
rank_termination_signal: signal.Signals = signal.SIGKILL if os.name != 'nt' else signal.SIGTERM
log_level: str = 'INFO'
max_rank_restarts: int = 0
max_subsequent_job_failures: int = 0
Expand Down

0 comments on commit 0f7891b

Please sign in to comment.