Hello everyone!
I have to implement a high availability solution for Mozilla's internal
SVN repository. It doesn't store the firefox code base but it's used
extensively for puppet configs, dns zones, etc. I've looked through the
archives and googled plenty. It's a critical service for us, so I would
really appreciate any feedback fellow SVN administrators can offer.
I've got two svn servers behind a round-robin load balancer connected to
a Netapp NFS volume. All of the repo data is stored on the Netapp with
the FSFS backend. Despite my best efforts I've been unable to break
anything or otherwise cause corruption. It's entirely possible the lack
of corruption is with my test methodology, but I don't know any better.
Is this a sane architecture?
Does anyone have operational experience with this type of setup?
Is there a better way?
- Brian
