[DRBD-user] drbd resource problem on reboot

[Isaac González]

Hi all,

I've a production system with this setup.

2 Xen servers in HA sharing DRBD resources.

DRBD resources are over LVM's.

DomU's are firewalls with differents VLAN's for each one. (There are not
I/O problems because I've not much read/write proccesses).

The DRBD setup is for each domU (we've 16 vm):

1 DomU
1 - DRBD for base system (5GB)
1 - DRBD resource for swap (512M)

We use a different port for each resource, and physical server are
connected through crossover 10gigabit wire. (not bandwidth problem)

For example:

resource VM1 {
    on M1  {
        device /dev/drbd38;
        disk /dev/vg00/VM1;
        meta-disk internal;
        address 10.0.0.1:7827;
    }

    on M2 {
        device /dev/drbd38;
        disk /dev/vg00/VM1;
        meta-disk internal;
        address 10.0.0.2:7827;
    }
}

resource M1-swap {
    on M1  {
        device /dev/drbd39;
        disk /dev/vg00/VM1-swap;
        meta-disk internal;
        address 10.0.0.1:7828;
    }

    on M2 {
        device /dev/drbd39;
        disk /dev/vg00/VM1-swap;
        meta-disk internal;
        address 10.0.0.2:7828;
    }
}

THE PROBLEM.

If I do a "reboot" on the domU, the DRDB resource freezes and domU
doesn't boot.
Then I've to kill the Xen proccess that are using drbd's resources of
this domU, set the drbd resource as secondary and start the domU again.
We'use the drbd helper script to manage drbd resources on xen.

Some idea?

Thanks in advance