[XCSSA] file server questions

[xcssa@xcssa.org]

> will do is use mdadm to create a RAID-5 array out of the disks, then
> run LVM on top of that so I can allocate them into more than one file
> system and/or resize them, etc.  So now my globally shared data is
> protected against single-drive failure.

Cool use of KVM.


> Then, what if the root disk dies?   So I think I want RAID-1 on /.  I
> can do that, but do I want LVM on top of that?

Ask yourself the simple question of why.. And what the pros and cons 
are.  I myself would probably run a mirror on /, but NOT run LVM atop 
that specific sw RAID-1
just because of the added filesystem complexity and my lack of 
experience with recovering data from failed LVM based filesystems 
(recovering data and config files with just plain extX is much easier). 
  And for what?  The ability to grow something that doesn,t really 
doesn't change in size except for maybe log files....

> Also, what about /boot?  Is there any advantage to mirroring that?

mirror?  sure... LVM? Prob. Not...


>   If
> not, what if the disk containing /boot gets trashed?  Do I dd or rsync
> to the other disk's identical /boot partition every night or
> something?

Huh?  No..it's a mirror.  The raid subsystem keeps things in sync.

I'm typing this on my treo from a tent in north CA, so forgive any typos 
and the like.

> I'm using whatever kernel the FC3 apt respositories give me when I
> "apt-get install kernel#whatever".  Do I need to worry about this
> initrd stuff?

The stock kernel handles both sw raid and lvm just fine.

If you want to see LVM  running atop sw RAID-1, come to the September 
XCSSA meeting.  I'll probably be demoing my MythTV box which is running 
with this config.

Well... Gotta get out of my sleeping bag now and break camp...

L8ER..

Tweeks