Re: Debian Wheezy & SSD+HDD configuration
On 06/07/13 18:27, Fran wrote:
I just installed Debian 7.0 "Wheezy" on what I call my new workstation.
It's a pretty standard setup: right now I installed Debian on the first
SATA drive (a 120Gb SSD drive) and mounted my 1Tb HDD as my home
partition with no problems at all.
I have a SOHO network with Debian Linux and Windows XP business desktop
caliber machines. I built a new Intel Core i7-2600S, 8 GB RAM, 60 GB
SSD, and 3 TB HDD machine for myself last year, and spent over 8 months
trying various Linux and BSD distributions looking for good support for
Intel Sandy Bridge processors/ graphics, crypto, virtualization, ZFS,
and a desktop I liked. I also STFW/ RTFM for SSD's, and found a
plethora of conflicting and outdated advice.
I finally gave up on ZFS and settled on Debian Linux amd64 Testing/
Wheezy, dm-crypt, ext4, VirtualBox, and XFCE desktop last December, and
did an OOTB install onto the SSD with a 1 GB ext4 boot partition
(currently 38 MB used), a 19 GB dm-crypt/ ext4 root partition (currently
4.9 GB used), and 40 GB of unpartitioned space. RTFM and checking the
system files led me to believe that the key adjustments for the SSD were
already included. I then configured the 3 TB HDD with a 3 TB dm-crypt/
ext4/ Samba partition for large/ shared data (photos, music, video,
etc.). I later added a 180 GB SSD with a 180 GB dm-crypt/ ext4
partition for VirtualBox disk images. Everything has been working fine.
:-)
I've been reading here and there though, and it's fair to say that the
SSD stuff is kind of new to me. I found nice documentation in the Wiki
(http://wiki.debian.org/SSDOptimization)
Interesting, and seems reasonably up to date. RTFM and checking my
system files again, it looks like what I need is already there. The
only (basic) improvement I'm missing is to make /tmp a RAM disk. (I did
so just now; we'll see if any of my apps break.) Thankfully, our needs
are simple enough, the hardware is powerful enough, and Debian Wheezy is
good enough, that I don't need to dig deeper.
but I hope you don't mind if I
ask for your personal opinion about a few things:
- Do I really need to set up a complex configuration for partitioning
like seen here (http://wiki.debian.org/Multi%20HDD/SSD%20Partition%
20Scheme)?
That depends upon your use-case, service level requirements, knowledge,
skills, time, money, etc.. For my needs, the strategy that works is to
have one drive per machine for the O/S, applications, and home
directories (system drives), and one drive shared over the network for
large files (data drive). I'd love to have SSD system drives in all my
machines, but that's not in the budget...
Should I be worried for unexpected errors if not?
You should always be prepared for failures. The above drive strategy is
integrated with my disaster planning/ recovery strategy, tools, and
processes. Note that I have put mobile docks in all the machines, and
that I have dedicated my second newest machine to disaster recovery and
workbench purposes. (Unfortunately, I haven't found a mobile dock that
works correctly with the new Intel DQ67SW motherboard and Intel 520
Series SSD's; they are installed internally.)
"Backup & Recovery Inexpensive Backup Solutions for Open Systems" by W.
Curtis Preston was a good read on the overall subject and on specific
open-source tools:
http://shop.oreilly.com/product/9780596102463.do
My SSD's are new, and neither has shown any signs of error. But, I've
dealt with at a dozen or more failed HDD's over the past 20+ years
(personally and professionally). Many were exhibiting symptoms, and I
was able to pull all or part of the data off them before they failed
completely. Others were toast by the time I got to them.
In any case, there is no substitute for good back-up and/or archive images.
Automation is key to good disaster recovery habits. There are various
backup/ archive/ recovery software packages available, FOSS and
commercial, but for my systems, I prefer lowest-common demoninator,
portable, FOSS, and KISS. I use rsync, tar, gzip, and ccrypt, wrapped
in Bash and Perl scripts. Most recently, I burned my home videos to Blu
Ray using xorriso, wrapped in Bash.
- What are (if any) the *must take into account* aspects when working
with an SSD on Debian? Are there any "new routines" I should be aware of
right now?
For myself, Wheezy has everything I need OOTB. RTFM:
1. fstab(5)
2. mount(8) -- relatime and ext4 trim options
3. tmpfs(5)
FWIW Here are the relevant /etc/fstab entries:
2013-06-08 01:12:45 root@i72600s ~
# egrep "(/ |/mnt/s |/tmp )" /etc/fstab
/dev/mapper/sda2_crypt / ext4
errors=remount-ro 0 1
/dev/mapper/ssd180 /mnt/s ext4
defaults 0 2
tmpfs /tmp tmpfs
nodev,nosuid,size=20%,mode=1777 0 0
HTH,
David
Reply to: