[DRBD-user] meta-disk

[Chip Burke]

***Ugh, replied directly to Corey and not the list. Here is it is for the
list.

Hrm, back to the drawing board then. What is a typical situation then? For
instance, if I make a change to httpd.conf on one machine in the cluster, is
there a way for it to automatically update on the other machine so if it has
to take over the configs are correct? I am certain I am missing something
being a newbie at this. Is Heartbeat the answer for the above? 

So is a more typical situation:

Use drbd to replicate /usr /home /var

Then heartbeat does something to keep the /etc/ type things in sync? Sorry
to be so dense here. I know the tools involved, just not exactly where each
one comes into play. Again, the end goal is essentially to have a fail over
box that automatically kicks in and operates exactly as its primary in case
of the primary failure, Apache in this case and if that works  suppose I can
apply this to most anything.

Sorry to ask so many questions. I would be all over the linux-ha examples,
but all of their docs are blank... ie:
http://www.linux-ha.com/GettingStarted/ApacheWithSharedDisk is empty.

________________________________________
Chip Burke
________________________________________


-----Original Message-----
From: drbd-user-bounces at lists.linbit.com
[mailto:drbd-user-bounces at lists.linbit.com] On Behalf Of Corey Edwards
Sent: Thursday, January 05, 2006 11:46 AM
To: drbd-user at lists.linbit.com
Subject: RE: [DRBD-user] meta-disk

On Thu, 2006-01-05 at 09:45 -0500, Chip Burke wrote:
> Exactly. I was trying to replicate /boot with 'meta-data internal' 
> which would cause it to die being that the space wasn't available. I 
> am slowly getting my head around the whole process.
> 
> However, I can't seem to mount the drbd device. I edited my fstab from
> 
> /dev/hda2 / ext3 defaults 1 1
> 
> to
> 
> /dev/drbd0 / ext3 defaults 1 1
> 
> and it does not like it. When I try to start drbd manually using 
> 'service brbd start' or 'drbdadm up all', I get an error that 
> /dev/hda2 is already mounted. So I am a bit confused as to how to 
> actually mount the /dev/drbdx device.

The error means exactly what it says. /dev/hda2 is already mounted and DRBD
will refuse to start in that situation. DRBD must ensure that it alone can
write to the underlying block device, otherwise you *will* get data
corruption. That's a Bad Thing(tm).

>  Essentially I want the root file system to be identical between two 
> boxes for a failover situation. Am I headed in the right direction?

I recommend you reconsider your goals. Do you really need the entire system
replicated? Doing so gets complicated quickly because to mount /dev/drbd0
you have to umount /dev/hda2. Well, where is /etc? It's on /dev/hda2. How
can DRBD read /etc/drbd.conf if the disk isn't mounted? That doesn't even
get into problems of IP address configurations.

I have heard rumors of it being possible. I believe Linbit offers a premium
service to that effect. If that's really what you want, I recommend shelling
out the cash because otherwise you're going to end up hurting yourself.

Corey