No subject

1.) chain-replication A <-> B, as you do it.
2.) MySQL cluster  or some other kind of db clustering solution (e.g.
continuent.org)
3.) parallel writes of all changes into two independent mysql database
servers (worst solution IMO)

I'm still undecided between 1 and 2. On the one side I've got ample
experience with replication, so I'm somewhat inclined to lean to this
solution. On the other side however I'd really like the RAID-like
approach of a cluster (as well as it's extendability), although I've
got no practical experience with this sort of thing.

Cheers,
Joe

On 1/21/07, Martin Burton <mvb at sanger.ac.uk> wrote:
> Hi Joe,
>
> >
> > I've searched and read a bit about high availability on this list, but
> > most of the posts are quite out-dated and only a few mention actual
> > working setups beyond pure theory. So I was wondering if someone had
> > such a setup running or could give a few thoughts about the available
> > options.
>
> The simplest solution which works nicely for us is simply to use two boxes.
>
> I have Radiator installed on both and a Makefile in the configuration
> directory that uses a bit of perl munging (to change parts of the config
> that are machine specific) and rsync to keep the two machines in sync so
> that configuration is as simple as making a change on one server and
> then using:
>
> make sync
>
> and optinally
>
> make restart  (which restarts both radiator instances).
>
> For authentication schemes which use a mySQL DB, I just use mySQL
> replication to keep the DBs in sync on both machines (the DB and
> Radiator aren't hit hard enough to require having the database on
> another machine).  Yes, it's a little bit of work to get the databases
> back in sync after a failure, but it's not going to happen too often if
> you are using robust hardware and RAID.  I've only had to do it once in
> the last 4 years.  Having said that I've probably just invited another
> failure ;-)
>
> Failover is performed at the client side (by setting primary and
> secondary radius servers) which helps keep things simple.
>
> If you need something more robust than this then the Linux-HA project is
> probably the best place to start researching.
>
> HTH.
>
> Regards,
>
> Martin.
> --
> Martin Burton
> Senior Systems Administrator               \\\|||///
> Special Projects Team                     \\  ^ ^  //
> Wellcome Trust Sanger Institute            (  6 6  )
> -----------------------------------------oOOo-(_)-oOOo---
> t: +44 (0)1223 496945             http://www.sanger.ac.uk
>
> --
> Archive at http://www.open.com.au/archives/radiator/
> Announcements on radiator-announce at open.com.au
> To unsubscribe, email 'majordomo at open.com.au' with
> 'unsubscribe radiator' in the body of the message.
>

--
Archive at http://www.open.com.au/archives/radiator/
Announcements on radiator-announce at open.com.au
To unsubscribe, email 'majordomo at open.com.au' with
'unsubscribe radiator' in the body of the message.