From d9fc24a7aa1aa1938ea04693ee1836cae8f20972 Mon Sep 17 00:00:00 2001 From: Richard Liaw Date: Tue, 15 Dec 2020 13:49:45 -0800 Subject: [PATCH] [core] recover startup logs (#12876) --- python/ray/node.py | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/python/ray/node.py b/python/ray/node.py index 13d73956c..f5d7afbe6 100644 --- a/python/ray/node.py +++ b/python/ray/node.py @@ -739,12 +739,19 @@ class Node: raise NotImplementedError def start_monitor(self): - """Start the monitor.""" + """Start the monitor. + + Autoscaling output goes to these monitor.err/out files, and + any modification to these files may break existing + cluster launching commands. + """ + stdout_file, stderr_file = self.get_log_file_handles( + "monitor", unique=True) process_info = ray._private.services.start_monitor( self._redis_address, self._logs_dir, - stdout_file=subprocess.DEVNULL, - stderr_file=subprocess.DEVNULL, + stdout_file=stdout_file, + stderr_file=stderr_file, autoscaling_config=self._ray_params.autoscaling_config, redis_password=self._ray_params.redis_password, fate_share=self.kernel_fate_share)