[CentOS] Monitor disc write in preparation for move to Solid State Disc

Dennis Jacobfeuerborn dennisml at conversis.de
Mon Apr 9 11:57:35 UTC 2012


On 04/09/2012 12:56 PM, Brian Jameson wrote:
> I am planning on changing/upgrading one of my systems to Centos 6.x with a
> solid state system drive and conventional 'data' drives. Any suggestions for
> a good simple tool, or way to give ongoing monitoring of writes to files on
> what will become the system drive. The I can then move as much as possible
> of the changeable files to the data drives.

Why bother using a ssd drive then? By moving i.e. the ~/.mozilla directory
off the ssd drive because the files change frequently you also loose a lot
of the performance.

I have a 128gb ssd system drive and 2 1TB raid-1 disks for data and I've
setup my ~/Music and ~/Videos directories as symlinks to that data drive
because of volume and not because they change frequently.

Modern drives are quite good at wear leveling so that with regular desktop
use you don't have to worry much about the drive going bad. Some vendors
are so confident in their current generation controllers that they now
offer a 5 year warranty on their drives.

Since it's a single system drive you should also setup a backup from your
system disk to your data disk just to be safe.

Also you should keep your firmware updated. I had a interesting experience
on Saturday. My ssd drive suddenly "disappeared" and my desktop froze and
in a console I could see lots of i/o errors. After a reboot the drive no
longer appeared in the bios screen. After a complete off/on cycle of the
system the drive was there again and I could boot normaly. Then after about
an hour the exact same thing happened again. At that point I feared that
the drive had developed a defect and was gone for good.

After a bit of research this turned out to be a firmware problem. This
problem hits all Crucial M4 SSD drives after the SMART counter for hours
online hits 5184 hours. The drive will then stop responding and only come
back after a full power cycle and the same thing will happen once per hour.
Luckily you can get a firmware update on Crucials firmware page that fixes
this and the data on the drive is not affected.

So as long as you use a current generation drive and keep up with the
firmware updates you should be good without any special monitoring of the
filesystem/drive.

Regards,
  Dennis



More information about the CentOS mailing list