[DRBD-user] About mounting drbd0

Digimer lists at alteeve.ca
Tue Apr 28 16:24:12 CEST 2015

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


On 27/04/15 03:15 PM, rdiascampos at uol.com.br wrote:
> Hi.
>  
> I'm very new to drbd. In fact, I'm still testing it.
>  
> I'm using openSuse 13.2. I could configure drbd in both machines and
> seems to be working well but ...
>  
> The problem occurs when I try to reboot the 'primary' machine - the one
> that have the primary or active file system. It cannot simply boot.
>  
> It can boot if a take the device out from */etc/fstab* and mount it
> after boot:
>  
> UUID=9585d7b4-fac5-4f19-aea3-95debdf86481 swap                 swap    
>   defaults              0 0
> UUID=9c052c13-c838-4488-959f-c0d5ef16377d /                    xfs      
>  defaults              1 1
> UUID=2dc37455-631e-48da-a308-990d2b6a404c /boot                ext4    
>   acl,user_xattr        1 2
> *#*/dev/drbd0                                /dados               ext4  
>     acl,user_xattr,usrquota,grpquota 0 0
>  
>  
> I also tried to make sort of 'manual mount' after boot using
> */etc/rc.d/after.local*, using the mount command:
> * *
> *mount -o acl,user_xattr,usrquota,grpquota /dev/drbd0 /dados*
>  
> but even so I could not boot because the device drbd0 seems to be very
> lazy if I interpreted correctly this peace of log:
>  
> ...
> Apr 27 15:07:22 massa-1 after.local[751]: mount: special device
> /dev/drbd0 does not exist
> Apr 27 15:07:22 massa-1 avahi-daemon[716]: Joining mDNS multicast group
> on interface eno1.IPv6 with address fe80::a65d:36ff:fe29:d690.
> Apr 27 15:07:22 massa-1 avahi-daemon[716]: New relevant interface
> eno1.IPv6 for mDNS.
> ...
> Apr 27 15:07:23 massa-1 ntpd[1465]: Listening on routing socket on fd
> #24 for interface updates
> Apr 27 15:07:23 massa-1 start-ntpd[1438]: Starting network time protocol
> daemon (NTPD)
> Apr 27 15:07:23 massa-1 drbd[1473]: Starting DRBD resources: [
> Apr 27 15:07:23 massa-1 drbd[1473]: create res: dados
> Apr 27 15:07:23 massa-1 drbd[1473]: prepare disk: dados
> Apr 27 15:07:23 massa-1 slapd[1439]: @(#) $OpenLDAP: slapd 2.4.39 $
>                                            
>  opensuse-buildservice at opensuse.org
> Apr 27 15:07:23 massa-1 kernel: drbd dados: Starting worker thread (from
> drbdsetup [1566])
> Apr 27 15:07:23 massa-1 kernel: block drbd0: disk( Diskless -> Attaching )
> Apr 27 15:07:23 massa-1 kernel: drbd dados: Method to ensure write
> ordering: flush
> Apr 27 15:07:23 massa-1 kernel: block drbd0: max BIO size = 1048576
> Apr 27 15:07:23 massa-1 kernel: block drbd0: drbd_bm_resize called with
> capacity == 7813790768
> Apr 27 15:07:23 massa-1 kernel: block drbd0: resync bitmap:
> bits=976723846 words=15261311 pages=29808
> Apr 27 15:07:23 massa-1 kernel: block drbd0: size = 3726 GB (3906895384 KB)
> Apr 27 15:07:23 massa-1 slapd[1556]: bdb_monitor_db_open: monitoring
> disabled; configure monitor database to enable
> Apr 27 15:07:23 massa-1 slapd[1556]: slapd starting
> Apr 27 15:07:23 massa-1 start[1439]: Starting ldap-server
> ...
>  
>  
> I could circunvent this behaviour configuring systemd in order to only
> execute the script /etc/rc.d/after.local after drbd initialization but I
> think that the mount should be executed from /etc/fstab code as usual in
> linux systems.
>  
> Can anyone give me some help?
>  
> Thanks.
>  
> Ricardo

If you plan to auto-start and auto-mount DRBD, you should really
consider using a cluster resource manager like pacemaker. You will need
fencing and pacemaker's stonith can be hooked into by drbd's
crm-fence-peer.sh handler.

If you don't want or can't use pacemaker + fencing, then I would very
strongly recommend keeping DRBD promotion and mounting a manual task.
Otherwise, a split-brain is inevitable.

-- 
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?



More information about the drbd-user mailing list