[DRBD-user] Best way to use LVM with drbd

Pierguido pierg75 at yahoo.it
Fri Feb 15 18:35:04 CET 2008

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


rois wrote:
> I'm running XFS on top of DRBD on top of LVM on top of a hardware
> RAID-5.  My reasoning was I didn't want to leave a ton of room for my
> system and log files but wanted to be able to expand them later if I
> needed to.  I figured the easiest way was to put my whole HWRaid-5 Array
> into the LVM and then parse it out as needed.  I'm also running several
> servers/apps on this cluster whose data I wanted isolated so one service
> couldn't be crashed by another one using all the space up.  This also
> seemed the most flexible.
> 
> /dev/sda1 (100%) -> vg0 (lvm2)  [sda1 is my HWRAID-5 array]
> 
> /dev/vg0 >	vg0/root “/” (1G) xfs filesystem
> 		vg0/var “/var” (1G) xfs filesystem
> 		vg0/swap “swap” (1g) swap
> 
> drbd0 = /dev/vg0/home  - HA mounts drbd0 on /home (XFS)
> drbd1 = /dev/vg0/mysql - HA mounts drbd1 on /mnt/mysql (XFS)
> drbd2 = /dev/vg0/basic - HA mounts drbd2 on /usr/lib/basic (XFS)
> drbd3 = /dev/vg0/smb   - HA mounts drbd3 on /mnt/smb (XFS)
I have a similar configuration....raid5 -> lvm -> drbd
For me it was actually a instictive choise...normally i use almost
everywhere lvm on top of hw raid, so i even didn't think too much about
the opposite situation (drbd -> lvm).

> Expanding the drbd's is a little bit of a hassle but since I should have
> to do it that often I don't care that much. 
These i din't try


> I wonder if there is a
> linux quota service somewhere that could monitor directory usage instead
> of using partitions to do that.  Hadn't thought of that before but may
> need to look into it.  Expanding a quota (like on a webserver) would be
> a lot easier.  Oh well.
Xfs has a directory quota service...I used it on a file server once and
was workinig pretty well.

http://linux.die.net/man/8/xfs_quota

Pier




More information about the drbd-user mailing list