[zfs-discuss] Re: Production use ?

Gordan Bobic gordan.bobic at gmail.com
Fri May 25 10:54:32 EDT 2012


On 05/25/2012 03:36 PM, josh at burstyns.com wrote:
> We do not automate failover -- that wasn't our mandate with this project.
> (The project was designed to protect the data via DRBD at all costs.) We
> are aware that manual intervention will be required to "failover" the ZFS
> ZVOLs and the targets.

Why manual? Surely you could automate this using RHCS (or heartbeat). I 
fairly regularly deploy redundant MySQL servers that use a DRBD backed 
FS for the databases/binlogs, and the failover works quite well 
(obviously the client has to reconnect).

> I spent some time trying to conceive of a way to automate a failover but
> in the end decided that it was better to have a person attend to a failure
> in lieu of trying to come up with every possible scenario. Again, this is
> mainly due to the non-integration of DRBD, ZFS and the iSCSI daemon.

You could have it automated with a cascade of dependencies in RHCS. You 
have the iscsi service dependant on zfs service, which is in turn 
dependant on DRBD being primary.

Gordan



More information about the zfs-discuss mailing list