Creating a Log Directory

Procedure

1. Run the following commands to create parent directories of component logs on each node based on Log paths of cluster scheduling components.

   mkdir -m 755 /var/log/mindx-dl
   chown root:root /var/log/mindx-dl

2. Create log directories according to the actual situation of the components.

   Table 1 Log paths of cluster scheduling components

   Component	Log Directory Creation Command	Target Node	Description
   Ascend Device Plugin	mkdir -m 750 /var/log/mindx-dl/devicePlugin chown root:root /var/log/mindx-dl/devicePlugin	Compute node	-
   NPU Exporter	mkdir -m 750 /var/log/mindx-dl/npu-exporter chown root:root /var/log/mindx-dl/npu-exporter
   NodeD	mkdir -m 750 /var/log/mindx-dl/noded chown root:root /var/log/mindx-dl/noded
   Elastic Agent	mkdir -m 750 /var/log/mindx-dl/elastic chown User-defined owner /var/log/mindx-dl/elastic NOTE: Mount the Elastic Agent log directory to the container. For details, see Configuring YAML.		The directory owner is user-defined. Note: The owner group of the Elastic Agent installation user, the owner group of the running user for invoking Elastic Agent, and the owner of the directory to which the host is mounted must be the same. You can customize the flush path of run logs of Elastic Agent. In this path, you can view logs of all Elastic Agent nodes without logging in to each node.
   TaskD	mkdir -m 750 Training script directory/taskd_log chown User-defined owner Training script directory/taskd_log		The directory owner is user-defined. TaskD can automatically create a log directory during running. The log directory prefix is usually the directory where the bash command is executed or the training is started in the job YAML file.
   Ascend Operator	mkdir -m 750 /var/log/mindx-dl/ascend-operator chown hwMindX:hwMindX /var/log/mindx-dl/ascend-operator	Management node	-
   Resilience Controller	mkdir -m 750 /var/log/mindx-dl/resilience-controller chown hwMindX:hwMindX /var/log/mindx-dl/resilience-controller
   ClusterD	mkdir -m 750 /var/log/mindx-dl/clusterd chown hwMindX:hwMindX /var/log/mindx-dl/clusterd
   Volcano	mkdir -m 750 /var/log/mindx-dl/volcano-controller chown hwMindX:hwMindX /var/log/mindx-dl/volcano-controller
   	mkdir -m 750 /var/log/mindx-dl/volcano-scheduler chown hwMindX:hwMindX /var/log/mindx-dl/volcano-scheduler
   Container Manager	mkdir -m 750 /var/log/mindx-dl/container-manager chown root:root /var/log/mindx-dl/container-manager	Node where the container recovery feature is required	-