"Johnathan Gifford" <jgifford@wernervas.com> writes:
> It's been explained to me that the with FSFS format, the file
> locking with the Subversion binaries the transaction files in the
> FSFS directories keep things in check. But this means putting a set
> of Subversion binaries on each server and not the NFS mount.
> However, the binaries don't require much configuration. As you
> said, hooks scripts and other stuff involving the repositories
> themselves is where all the configuration occurs.
I'd love confirmation on that myself.
If not, could you set it up so that both servers have READ access to
the NFS FSFS repo but only one at a time has write access? I'd expect
any load balancer could discriminate between GET and POST if you're
using HTTP access. That way you wouldn't be wasting one of your CPUs
most of the time.
BTW, if you don't have a HW load balancer, I've found "pen" to be
rather nice: trivial to set up and get going. I expect you could run
it on each of the servers themselves and eliminate the need for a
separate (pair of) balancers; the hard part would be getting both pens
to talk to each other and decide who the master is but see the docs on
The Ultimate Cheapskate Cluster.
http://siag.nu/pen/
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Tue Jan 30 02:01:30 2007