[DRBD-user] New storage: avoid split brains

Gandalf Corvotempesta gandalf.corvotempesta at gmail.com
Thu Oct 20 15:49:30 CEST 2016

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


2016-10-20 15:14 GMT+02:00 Lars Ellenberg <lars.ellenberg at linbit.com>:
> You end up with a system that will NOT experience data divergence,
> unless you force it to.

Ok.

> But you may run into (multiple failure, mind you!) situations
> where you are offline, rather than risk to go online
> with a possibly stale data set.

This could happens if the redundant cluster links both failes and the
only survived node also fails.
The other node can't go up due to the corosync constraint, and the
"master" node is failed, thus
all cluster is down but data are still consistent.

This is a situation that I can manage, I can replace a node with a
spare (by using software raid or ZFS raid,
I just need to move disks from the died server to a brand new and power on.

An inconsistent data set is much more difficult to solve (almost
impossible) as you could have some good files on both end.

Do you have any guide, tutorial, howto, samples, .... on how to
configure drbd as you told me ?
Some working configuration for Pacemaker ?



More information about the drbd-user mailing list