I am running the below command to format the nvme drive to lustre with backfstype as zfs
mkfs.lustre --mdt --reformat --mgsnode=mgsmaster@tcp --backfstype=zfs --fsname=lustre mdtpool/mdt --index=0 /dev/nvme1n1
After formatting the drive
I am trying to setup drbd using the below resource config
resource "r0" {
protocol C; //Updated
device /dev/drbd0; //Updated
disk /dev/nvme1n1;
meta-disk internal;
options {
auto-promote no;
}
on "hostname1" {
node-id 0;
}
on "hostname2" {
node-id 1;
}
connection {
host "hostname1" address 10.40.40.1:7789;
host "hostname2" address 10.40.40.2:7789;
}
}
after this I run the below command to initialize drbd
drbdadm create-md r0
open(/dev/nvme1n1) failed: Device or resource busy
...
I found this post on the drbd website where it's mentioned that ZFS doesn't hold the device open in the kernel the same way other filesystems or processes in Linux do
[DRBD Reference website for zfs](https://kb.linbit.com/using-zfs-over-drbd-with-pacemaker)
Need some help here.
I tried first setting up drbd on the drive and then formatting with mkfs.lustre but then it's the same mkfs.lustre wants drbd service to be down, and only after that it can format
update: The issue has been resolved
The steps for replication are as follows:
drbdadm create-md r0
drbdadm up r0
drbdadm primary r0 --force
mkfs.lustre --mdt --reformat --mgsnode=mgsmaster@tcp --backfstype=zfs --fsname=lustre mdtpool/mdt --index=0 /dev/drbd0
mount -t mdtpool/mdt /mnt/MDT
If we want to remount after reboot
drbdadm up r0
drbdadm primary r0 --force
zpool import -o cachefile=none mdtpool
mount -t mdtpool/mdt /mnt/MDT
Asked by Neil Karania
(3 rep)
Jun 11, 2024, 09:09 AM
Last activity: Jun 13, 2024, 08:56 AM
Last activity: Jun 13, 2024, 08:56 AM