[DRBD-user] DRBD Question

Brad Barnett lists at l8r.net
Tue Jan 9 22:15:00 CET 2007

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


On Tue, 09 Jan 2007 08:33:55 -0500
Diego Julian Remolina <diego.remolina at ibb.gatech.edu> wrote:

> > 
> > works fine. in principle.
> > 
> > still I recommend to NOT use such a setup, because:
> > if one box cannot take the full load on its own,
> > the failover cluster is useless:
> 
> However, I can argue that if one machine can take the load, then it
> would be a waste of money to  have two machines and one of them there
> sitting all the time doing nothing, just waiting.

No, it is not a waste of money.  If you believe it is a waste of money,
then frankly you aren't looking at what drbd is for.

> 
> I do not currently have an active/active setup, but have DRBD 0.7.x
> running on two machines with two  drbd partitions. One machine is mail
> master (sendmail+mimedefang+spamassassin+clamav+cyrus-imap) and  the
> other machine is web master (apache+php+mysql). Since we do not get that
> much traffic in our web  server and mail server, I am sure one machine
> would be able to take both loads simultaneously and  have lots of power
> to spare. They are very decent machines hardware wise (Dual Opteron 270
> with 4GB  Ram, Areca SATA raid controllers in RAID 10 and Gbit links to
> our campus network).

If you are sure one machine would be able to take both loads, your
argument is not logical.  It does not matter, economically, if you use one
machine and have another acting as purely a drbd slave.  You have not
"saved" anything.  You still have two machines.  

Again, since you claim one can easily take the load of both, then you in
the precise same position as both machines sharing the load.

The only way you can "save", is by having both machines loaded, so that
services slow down when one machine dies.  In such a case, you are not
truly redundant.  Worse, your language above admits that you just think
that one machine can handle the load, you have not actually empirically
tested it.

Right now, you're not really redundant.  When was the last time you pulled
the plug from your drbd master, and let the slave run for a few hours?



More information about the drbd-user mailing list