Linux most common config
plug.org at todandlorna.com
Thu Oct 6 20:38:44 MDT 2011
On 10/6/2011 1:55 PM, Aaron Toponce wrote:
> On Thu, Oct 06, 2011 at 09:51:47AM -0400, Francisco Diaz wrote:
>> I have a question for those using Linux (more in particular RHEL or CentOS)
>> in production environments: what is, based on your experience or
>> conversations with other sys admins, the most popular, favorite, reliable,
>> or what ever you want to call it, configuration/selection of FS(ext3/4,xfs,
>> reiserFS, etc) and hard drive config (partitions, raidX, LVM, JBOD, etc) for
>> a production system?
>> I know that maybe the answer will depend on what apps you have in the server
>> (DB, File Server, Mail, WebApps, etc), but in general, what is the most
>> common scenario in production regarding to FS and HD config?
> Unless I'm setting up something that has specific write/read performance
> requirements, I setup ext4 on LVM on RAID5. I never use XFS on LVM, because
> you can't shrink XFS filesystems. I would never put production on ReiserFS,
> mostly because there is no future with it.
Beware of current ext4 setups for large arrays, mind you. You can NOT
get bigger than 16TB with current tools, and they've been "heavily
working on that" for 2 years now. Just FYI.
In most of my setups I tend to use RAID6 or RAID1 (RAID1 usually on md
with whatever server, RAID6 on the storage servers). I'm no storage
nerd, but I like RAID6. btrfs is getting fun, but is still repeatedly
slow in many scenarios and I haven't looked into why, so maybe do some
research if you're going to jump into that boat this juncture.
Eventually it WILL be the FS of choice, I'm sure.
Separate boot partition, everything else in / and built on top of LVM.
Backup boot somewhere else. Done. Swap, of course, I guess.
I get the feeling it's not an exact science with me. Probably because
it's more situational than anything, but that's my starting philosophy.
More information about the PLUG