I've been looking through documentation and trying to understand the current
situation with clustering Samba, and I need a little guidance. My setup is as
follows:
1. A NetApp filer, fully redundant and all that, which holds the underlying data
split among dozens of physical volumes.
2. A Samba server, which mounts all NetApp volumes using smbmount and uses a
custom VFS module to present the physical volumes as a single logical share. The
VFS module also does some app-specific permission checking and applies a whole
bunch of special rules to which directories users have access to. My users map
the share as a drive on their Windows machines, and just see what they're
supposed to see. All this works well.
Now, I have another requirement for a future version of the app I'm working
on.
I have to present a slightly different view of the underlying files to a
different group of users. Basically, I need a second Samba server, running
another VFS module, resharing the same NetApp volumes as the first Samba server.
Some users will see and use the same files on both Samba servers.
Do I need to set up a cluster using CTDB for this? I imagine I'll have all
sorts
of subtle and weird locking errors if I naively point both Samba servers to the
same NetApp shares, and people start editing the same file when connected to the
two different Samba instances. I would honestly assume that I do, except chapter
13 of the Samba by Example book suggests using SAN storage on Samba servers, and
specifically says "Many SAN-based storage systems permit more than one
server to
share a common data store." I'm wondering if the NetApp we use is good
enough.
I don't care about load balancing, high availability, or anything like that.
One
set of users will connect to samba1.site.com, and another set will connect to
samba2.site.com.
Many thanks,
CV
