[DRBD-user] drbd newbie

Tim Hibbard hibbard at ohio.edu
Fri May 6 22:20:18 CEST 2005

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

I am in the beginning stage of planing a high availability environment. I have 
some general questions about deployment. Any suggestions or comments
are welcomed.

I have two servers connected to a shared storage device (server1 and server2).
We also have a 'slave' machine on the network which I would like to have all 
data duplicated to in case of an 'emergency'.

My main concern is the the 'slave' remains a secondary at all times until I 
change it manually as the primary in the event of an 'emergency'.  Do I need
to change anything in my drdb.conf to make this happen.  What about reboots?  
Or should I make a startup file which executes "drbdadm secondary all" 
on the 'slave' machine?  

Also I question on how heartbeat/drbd handles the exchange of drbd primary 
on the heartbeat cluster.  The scenario is 'server1' had /dev/drdb0 mounted and 
is the primary.  'Server1' has problems and heartbeat transfers resources to 
'server2' via the haresources. (we have stonith network powerswitch to take care 
of dual mounts).

server1 datadisk::drbd0 Filesystem::/dev/drbd0::/home Filesystem::/dev/sda1::/usr/local/apache apache
server2 Filesystem::/dev/sda3::/usr/local/mysql mysql

Does 'server2' automatically become primary? Will the 'slave' still remain secondary the 
whole time?  Is the same true with the heartbeat event of auto_failback?  Or should I 
put a "drbdadm primary all" in my apache init.d file?

The mysql database has built-in replication happening with the slave machine.  
Also the data is saved once an hour with mysql_dump.  It it possible to replicate 
the data in /usr/local/mysql database with drbd?  Could this cause corruption/unreliability 
in the database files located on the slave node?

Here are my config files.  Any feedback is greatly appreciated.

resource drbd0 {
  protocol B;
  startup {
    degr-wfc-timeout 120;    # 2 minutes.

  disk {
    on-io-error   detach;
  net {
    on-disconnect reconnect;
  syncer {
    rate 100M;

  on cluster {
    device     /dev/drbd0;
    disk       /dev/sda4;
    meta-disk internal;

  on slave {
    device    /dev/drbd0;
    disk      /dev/hdc1;
    meta-disk internal; 

logfacility daemon         
node server1 server2 
keepalive 1
deadtime 10    
bcast eth0
auto_failback yes
respawn hacluster /usr/lib/heartbeat/ipfail
stonith apcmaster /etc/ha.d/apcmaster.conf
stonith_host *  apcmaster xxx.xxx.xxx.xxx username password

Tim Hibbard
Software Engineer For 
Vice President Of Research
Ohio University
Athens, Ohio 45701
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20050506/1d3577ab/attachment.pgp>

More information about the drbd-user mailing list