[DRBD-user] primary/secondary problem

paddy paddy at panici.net
Tue Jun 13 01:27:30 CEST 2006

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

On Mon, Jun 12, 2006 at 09:12:41PM +0200, Gernot W. Schmied wrote:
> paddy wrote:
> > On Mon, Jun 12, 2006 at 03:28:49PM +0200, Gernot W. Schmied wrote:
> >> This is no cluster setup, I run them isolated and once a day I power up
> >> the secondary for synchronization
> > 
> > If you don't mind me asking: Why ???
> Because I am excentric an like it that way, I am well aware of the
> disadvantages doing it that awkward way, I have my reasons and I do not
> wish to discuss it.

No problem, I've no wish to press you on details you do not wish to share.

FWIW, for a once a day sync at the block device level, you might wish
to consider lvm snapshots.  Of course, if running drbd synchronised in
the event of some contingency is part of your game plan, then that
wouldn't interest, but that's one reason I asked.

> I'd rather and humbly appreciate some enlightening input on the question
> raised ;-).

the drbd nodes communicate by way of a network connection, and it appears
that they are not connecting.  The usual debugging techniques for
network connections that are not happening apply.  The other correspondent
was correct to suggest a firewall problem as a likely candidate.

can you ping ?

> By the way, when both cluster participants are in secondary mode, can
> one absolutely rely on the fact that they will never attempt
> synchronization in any direction?

It is my understanding that is what 'both in secondary mode' entails. You 
would need to satisfy yourself as to how far you can trust any particular
piece of software for your purposes.  You should also be aware of systems
like heartbeat that might be configured to tell drbd to change state
(thereby possibly triggering a synchronisation).

Perl 6 will give you the big knob. -- Larry Wall

More information about the drbd-user mailing list