[DRBD-user] Data Rollback on DRBD 8.3.4

Lars Ellenberg lars.ellenberg at linbit.com
Thu Feb 24 17:51:19 CET 2011

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


On Thu, Feb 24, 2011 at 02:24:16AM +0800, jayson.cena at teamglac.com wrote:
> but only vmcluster02 is writing to r0
> because the startup configuration of
> the VM that is using r0 will only reside on 1 host.

Then the other should not have been made Primary.

> is there a possibility that when the hosts writing to the resource
> crashed, a split brain will occur?

"split brain" in this context is defined as situation where both nodes
have been in Primary role without communicating with the peer.
Not necessarily at the same time, but during the same time period
without communication.

The "split brain" is detected on the next handshake
by means of the data generation uuids.

If they diverge, DRBD has to assume that data _could_ have been modified
independently.

There is only one "safe" (in some sense) auto-recovery strategy, and
that is "discard-zero-changes", where zero changes means DRBD is _sure_
that there have not been any changes. As soon as DRBD has some doubt
about whether or not there could have been changes, it needs to throw
away potentially changed data. And you should be pretty sure that you
mean it if you configure it to throw away data _automatically_.

If you tell DRBD to throw away your data,
don't blame DRBD if it does just that.

-- 
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com

DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
__
please don't Cc me, but send to list   --   I'm subscribed



More information about the drbd-user mailing list