[DRBD-user] Upgrading Standalone DRDB Server to a Cluster

[J.J.]

hi List,

i have a production system that acts as NFS Server and (soon) uses DRBD to replicate about 1,4TB of Data to his neighbour. I have setup the whole environment before, just the storagebox of the second node is missing.

The Servers are both up and running, but due a project time issue i didn´t get the Storagebox for the second Node, btw. we use HP DL380G4 with a MSA20 attached over SCSI.

My question / concern now is, how drbd recognize which node acts as primary, and where should it be synced from. I´m a little worried that if i start drbd on the second node it syncs garbage over my data. How can i make sure that, node1 will be synced to node2 and not vice versa?

Also we use heartbeat to manage the other resources on this NFS Cluster, and hopefully heartbeat does not try to enable drbd in primary mode on node2 when i start it for the very first time on node2.

On the first node it looks like this:

/dev/drbd0            1,4T  408G  899G  32% /data

>cat /proc/drbd 
version: 8.2.6 (api:88/proto:86-88)
GIT-hash: 3e69822d3bb4920a8c1bfdf7d647169eba7d2eb4 build by root at server1, 2008-10-06 14:19:31
 0: cs:WFConnection st:Primary/Unknown ds:UpToDate/DUnknown C r---
    ns:0 nr:0 dw:43565540 dr:850845673 al:105978 bm:105851 lo:0 pe:0 ua:0 ap:0 oos:1465010928

drbd.conf looks like this:

global { 
  usage-count no; 
}
common {
  protocol C;
}
resource r0 {
  syncer {
     rate 200M;
  }

  disk {
    on-io-error pass_on;
  }


  on server1{ 
    device    /dev/drbd0;
    disk       /dev/cciss/c0d1p1;
    address   87.783.284.12:7789;
    meta-disk internal;
  }

  on server2{
    device    /dev/drbd0;
    disk       /dev/cciss/c0d1p1;
    address   87.783.284.13:7789;
    meta-disk internal;
  }

}


and we use heartbeat, /etc/ha.d/haresources looks like this:

server1 IPaddr::87.783.284.10/25/bond0 drbddisk::r0 Filesystem::/dev/drbd0::/data::ext3 nfs


- 


I know it sounds a little complicated, but maybe some of you gurus can help me out, or give me a clue.. 

Also i´ll test this scenario in a test-env...

thanks&greets

jakob







-- 
Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger01