I use hdp and use the script from :
But my server user raid5 ,so how to set the number of the disk?
Option 1: Search for dfs.datanode.data.dir in HDFS service in Ambari.
The values are separated by ,
Enter the number of directories specified over there.
You can also go to your terminal and issue:
sudo fdisk -l
and see the discs. Option 1 is easier though.
I use the script to calc how to set the mem setting:
so a param is -d disk number.
My server had raided disks,so I don't how to determine how to set the disk number.
Set to 1 means the number of raid number,set 12 ,means the hard disk number.
Is this some sandbox cluster that you are setting up to play with Hadoop for learning purpose? If not, you should not be using RAID 5 for Hadoop disks. That's okay for OS and even okay for namenode if there is no HA/journal nodes. But you don't use RAID 5 for Hadoop data disks.