[Leaplist] Remote paths/storage, automounter, layout/considerations -- WAS: Scanner Setup Help?

Bryan J. Smith b.j.smith at ieee.org
Thu Jul 23 14:40:59 EDT 2009


Jason Boxman wrote:  
> As do I running stock everywhere.

"Stock" still means there are user-specific configuration files.

Jason Boxman wrote:  
> NFS wouldn't be my first choice for replicating my configuration 
> everywhere due to performance and reliability issues.

Specifics?  I see you talking NFS quite a bit, but for a remote
filesystem, NFS' performance is unmatched (several times over
SMB, especially with 32KiB blocks and jumbo frames).  Regarding
reliability, NFS is fairly stateful, and NFSv4 addresses a lot of
details.  Again, compared to ... ?

The only one I could think of is GFS, but that requires a SAN
target to be of any, good performance and reliability.  Although
the iSCSI target is officially supported in Red Hat Enterprise
Linux Release 5 Update 3 (EL 5.3)

Kevin Korb wrote:  
> I do use rsync to keep certain configurations consistent but
> it is a pain compared to the simplicity of using NFS on
> desktop systems.

There are a few LVM options for disconnected/resync'd filesystems
replicated between systems.  I have not used them heavily though.

GFS is also an option, especially since, again, the iSCSI target is
now officially supported in EL 5.3.  It's no longer a tech preview
(despite CentOS and CentOS off-shoots offering iSCSI target support
well before even Red Hat thought it was ready).

GFS can be a much better option than NFS in many circumstances.
GFSv2 addresses a lot of performance issues with lots of small
files, and is also officially supported as of EL 5.3 as well.
GFS will have overhead, but compared to the RPC calls required
for some NFS options (dentry, lstat, etc...), GFS will kick NFS'
butt.  GFSv2 further addresses locking performance.

GFS isn't for all circumstances.  But small numbers of nodes, and
where your storage is reliable, definitely.  The iSCSI target can
be your server.  I do _not_ recommend running GFS atop of GNDB
though (or running GNDB in general).  But that's another story.
You can run GNDB, but don't expect to get reliability in
production environments.

The one nice thing I love about GFS is that is solves on, critical
problem ... people being chronically stupid with NFS (let alone
SMB resharing).  It kills those "clusters" (you know what I mean),
with a real, "cluster" filesystem.  But even at home, with your
server as the iSCSI target, GFS can be more ideal than NFS in many
cases.


-- 
Bryan J Smith          Professional, Technical Annoyance
b.j.smith at ieee.org    http://www.linkedin.com/in/bjsmith
--------------------------------------------------------
I don't have a "favorite Linux distro."  I use, develop
and support community efforts, often built around Linux.
Technology and solutions are my focus, not dragging in
assumptions, marketing and other concepts which dominate
non-community developed software, which I left long ago.

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.



More information about the Leaplist mailing list