LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: drbd

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: drbd
From: Christophe Mailhebuau <admins@xxxxxxxxx>
Date: Wed, 22 Feb 2006 09:36:48 +0100
sorry,

My drbd with nfs + heartbeat run successfully.

Le 20/02/06 at  19:41, Roberto Nibali a ecrit:
>Hello,
>
>There is a drdb user mailinglist, which is more appropriate to ask such 
>questions.
>
>><drbd.conf>
>>resource drbd-resource-0 {
>>protocol C;
>>incon-degr-cmd "halt -f"; # killall heartbeat would be a good alternative 
>>:->
>>
>> disk {
>>  on-io-error detach;
>>}
>>   syncer {
>>   rate 10M; # Note: 'M' is MegaBytes, not MegaBits
>>  }
>>      on RESURA-SDB1 {
>>      device    /dev/drbd0;
>>      disk      /dev/sda2;
>>      address   192.168.1.40:7789;
>>      meta-disk  internal;
>>  }
>>      on RESURA-SDB2 {
>>      device    /dev/drbd0;
>>      disk      /dev/sda2;
>>      address 192.168.1.50:7789;
>>      meta-disk internal;
>>  }
>>}
>></drbd.conf>
>>
>>I've start drbd on master an slave, run sucessfully.
>>on master i do mke2fs -j /dev/drbd0 an is ok
>
>Ok.
>
>>i mount on master /dev/drdb0 /opt, is ok
>>
>>i do modification in opt on master, example i made a directory
>>I don't understand why the new directory on master not replicate on
>>slave when partition is mount (/dev/sda2 /opt)
>
>You're not allowed to mount either /dev/drbd0 (won't let you anyway) or 
>any underlying partition on a slave while you have a master server 
>actively mounted the drbd block device. DRBD is not designed that way. 
>If you need a simultaneous read-write (cluster-wide) you better use a 
>cluster filesystem.
>
>If you really need to have RO access on the slave you need to export 
>your share via NFS or something to the slave. And when the slave takes 
>over, you swap the game. It's a bit tricky in the linux-ha setup but works.
>
>>On slave, I don't understand, when i umount and remount partition the
>>replication is ok.
>>
>>I must umount and remount everytimes ?
>
>No, the data is synchronised on write, however you're not allowed to 
>mount the partition on the slave so long as the master has I/O write 
>access to it. You can check /proc/drbd for information of your shared 
>disk status.
>
>You're lucky that you didn't lose your data.
>
>Regards,
>Roberto Nibali, ratz
>-- 
>echo 
>'[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq' | dc
fin du message de Roberto Nibali

-- 
Cordialement,       
Christophe Mailhebuau  - http://www.justlinux.org
Administrateur systèmes et réseaux GNU/Linux
GPG: 1024D/75B07F8D: BD18 2D01 D954 A339 E35B 9EA1 4240 0410 75B0 7F8D

<Prev in Thread] Current Thread [Next in Thread>
  • drbd, Christophe Mailhebuau
    • Re: drbd, Roberto Nibali
      • Re: drbd, Christophe Mailhebuau <=
    • DRBD, Mailhebuau Christophe