[DRBD-user] Some question

Lars Ellenberg Lars.Ellenberg at linbit.com
Sat Mar 27 12:03:10 CET 2004

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

/ 2004-03-27 11:23:25 +0100
\ gianlu]{a:
> Hi, 
> i have built a two node HA-cluster with suse 9 and heartbet.
> This cluster replies our mail system based on postfix, apache( for squirrel)
> and courier-imap.
> I use the maildir so I have to sync the /home mountpoint.
> For budget reason we haven’t a shared storage so I find this great software
> and decide to try it.
> On suse it’s simple to install because it’s include in the distro, so I
> create my drbd.conf and all seems to works fine.

Note however that suse includes 0.6.6 (with a minimal patch),
and we by now are at 0.6.12 for the stable series.

> I sync the /dev/hda3  50GB partition on the two node.
> This is my conf file
>   resource drbd0 {
>   protocol=C
>   fsckcmd=/bin/true
>   inittimeout=-60
> disk {
>   do-panic  
>   disk-size=52434552
>   }
> net {
>   sync-nice=-10   
>   sync-min=80000

default unit  is KB/sec, but you can append M or G to change the unit
so this means 80M, maybe this would be easier to read.

btw, you are sure you can get 80MB/sec throughput
from your disks and network?

> A thing i haven't understand, it is better to have a row like this
> /dev/nb0 /home auto noauto 1 1 
> In /etc/fstab or remove this line and use something like this
> Filesystem::/dev/nb0::/home::auto
> In /etc/ha.d/haresources.

This is just different.
Heartbeat likes it better if you use its Filesystem resource.
You still have to use the "datadisk" resource, too.

	Lars Ellenberg

More information about the drbd-user mailing list