best no need alarm

PaddlePaddle · Jul 11, 2023 · 89c8823 · 89c8823
1 parent 953282a
commit 89c8823
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 4 deletions.
diff --git a/python/paddle/distributed/launch/controllers/controller.py b/python/paddle/distributed/launch/controllers/controller.py
@@ -36,10 +36,13 @@ def __init__(self, ctx):
         signal.signal(signal.SIGTERM, self.signal_handler)
         signal.signal(signal.SIGABRT, self.signal_handler)
         signal.signal(signal.SIGINT, self.signal_handler)
-        if ctx.is_auto_tuner_mode() and not ctx.run_best:
-            # set per task timeout
-            signal.signal(signal.SIGALRM, self.not_exit_signal_handler)
-            signal.alarm(ctx.max_time_per_task)
+        if ctx.is_auto_tuner_mode():
+            if not ctx.run_best:
+                # set per task timeout
+                signal.signal(signal.SIGALRM, self.not_exit_signal_handler)
+                signal.alarm(ctx.max_time_per_task)
+            else:
+                signal.alarm(0)
 
         self.ctx = ctx
         self.master = Master.factory(self.ctx)

diff --git a/python/paddle/distributed/launch/main.py b/python/paddle/distributed/launch/main.py
@@ -487,6 +487,7 @@ def launch():
         ctx.args.job_id = "best_cfg"
         ctx.logger.info(f"Launch best cfg from auto tuner: {best_cfg}")
         ctx.args.log_dir = "best_cfg"
+        # run best cfg
         c = controllers.init(ctx)
         c.run()
         c.finalize(exit=True)