[DRBD-user] question about start drbd on single node after a power outage

Kaloyan Kovachev kkovachev at varna.net
Tue Jan 31 10:50:47 CET 2012


Hi,

On Mon, 30 Jan 2012 09:20:24 -0500, "Xing, Steven"
<SXing at BroadViewNet.com>
wrote:
> Hi, please help me on following question:
> 
>  
> 
> I have 2 drbd node running master/slave under pacemaker,
> 
> It is running fine, the problem is:
> 
>  
> 
> After a 2 node power outage, if I just boot up one node, no matter the
> node is the previous drbd primary or secondary, the drbd disk status
> always "consistent" and not "UpToDate", thus, it can not be promote to
> primary, I have to force it to be primary manually, or boot up the
> second node, when the 2 nodes connected, the status changed to
> "UpToDate".
> 
>  
> 
> Is this behavior normal? Is it a design to avoid some problem? Can
> somebody help on explain this?, I did not see detail info in the
> documentation.
> 

Consider your primary (say node1) dies first, but the secondary (node2)
lives long enough to became primary and start writing to disk before it
dies too. Now when power is back, do you still want your previous primary
(node1) to start before it can check the status of the other?

>  
> 
> Is there a safe way to allow the previous primary node can be boot up
> only and promoted to primary automatically without the other node?
> 

There is a way, but not a safe way

>  
> 
> Thanks in advance.
> 
>  
> 
>  
> 
> Steven Xing.


More information about the drbd-user mailing list