执行npu-smi info 报这个错误: can't get docker_id or the docker_id is invalid 
收藏回复举报
执行npu-smi info 报这个错误: can't get docker_id or the docker_id is invalid 
t('forum.solved') 已解决
新人帖
发表于2024-09-18 16:52:44
0 查看

执行npu-smi info 报这个错误: can't get docker_id or the docker_id is invalid 

具体场景》
主机执行:不报错

docker -v+特权:dmesg就会报错: can't get docker_id or the docker_id is invalid 
docker --device:不会报错

ctr -v+特权: :dmesg就会报错: can't get docker_id or the docker_id is invalid 
ctr --device:不会报错

k8s,yaml中-v挂载设备:dmesg就会报错: can't get docker_id or the docker_id is invalid 

详细dmseg:

主机上查看demsg日志,执行上述命令会触发如下报错:

[46707.255887] [ascend] [ERROR] [devdrv] [devdrv_manager_container_get_docker_id 1283] npu-s

mi:39200 can't get docker_id or the docker_id is invalid: docker_id = 128

[46707.255904] [ascend] [ERROR] [hdcdrv] [hdcdrv_connect 3830] npu-smi:39200 container_id is

illegal. (dev_id=0; service_type="service_dmp"; container_id=129)

我要发帖子