[ULTRAMONKEY-USERS] High availability and drbd

[pierre-damien at gateway.gatouillat.net]

Hello,

I've a cluster with 2 machines under debian sarge, with three NIC : eth0
and eth2 for a bonding interface bond0 and eth1 for drbd replication
(back-to-back)

the cluster is OK when I stop one machine and vice-versa. But when I
unplugg the cables of eth0 and eth2 of one machine, the resources of drbd
are not "transferred". OK for IPAddr.

I read somewhere a problem with deadtime (ha.cf) and timeout (drbd.conf) :
deadtime must be > timeout (unit is 0.1 for drbd.conf). It is OK for my
config.

I monitor the log on the machine where I shutdown the 2 interfaces
eth0/eth2, I don't see something like drbddisk::xxxx stop. I think, with
ipfail, when the machine can't ping the host, she must release her
resources   ???

Thanks in advance,
Pierre-Damien

My configs :
drbd.conf :
resource backup {
  protocol C;
  incon-degr-cmd "echo '!DRBD! pri on incon-degr' | wall ; sleep 60 ; halt
-f";
  startup {
    # wfc-timeout  0;
    degr-wfc-timeout 120;    # 2 minutes.
  }
  disk {
    on-io-error   detach;
  }
  net {
        timeout       40;
        connect-int   10;
        ping-int      5;
  }
  syncer {
    rate 80M;
    group 1;
    al-extents 257;
  }
  on m-6 {
    device     /dev/drbd0;
    disk       /dev/mapper/vg00-backup;
    address    192.168.254.1:7788;
    meta-disk  internal;
  }
  on m-3 {
    device    /dev/drbd0;
    disk      /dev/mapper/vg00-backup;
    address   192.168.254.2:7788;
    meta-disk internal;
  }
}

ha.cf
debugfile /var/log/ha.debug
logfile /var/log/ha.log
keepalive 1
warntime 2
deadtime 8
apiauth ipfail gid=haclient uid=hacluster
node    m-6
node    m-3
auto_failback off
ucast bond0 172.21.1.39
udpport 694
respawn hacluster /usr/lib/heartbeat/ipfail
ping 172.21.1.33

haresources
m-6 \
        IPaddr2::172.21.1.40/28/bond0/172.21.1.47 \
        drbddisk::backup \
        Filesystem::/dev/drbd0::/backup::reiserfs \
        mysql \
        freeradius \
        backuppc \
        apache2