Tuning Linux for SSD operation

wirelessdreamer · 09-16-2008, 06:26 PM

I just picked up an ocz corev2 60gig ssd for my carpc. I haven't started using it yet, as I make to make sure I don't kill its write cycle limit (~10k) messing around. I've been thinking about it for a while now, and am leaning to having my ssd mounted read only.

Here are links to some sources i've been reading about it at:

http://gentoo-wiki.com/HOWTO_Read-on...Unionfs_Method
http://tombuntu.com/index.php/2008/0...-state-drives/
http://famvdploeg.com/blog/?p=40

as of now I think i'll mount the unionfs part on a flash drive, and just replace it whenever it burns out from too many writes.

wirelessdreamer · 11-02-2008, 07:34 PM

fyi putting var in a ramfs, and syncing it on shutdown is a bad idea, because when you do a partial boot, and it hasn't been fully restored and then back it up . . . you have to revert to an old image.

samr7 · 11-03-2008, 12:55 AM

Doesn't your SSD have some sort of wear leveling? I thought that with wear leveling, the write cycle lifetimes of a big device were so ridiculously large that the PC would break or get replaced with something better long before its flash started to fail.

hexxamillion · 06-28-2009, 06:07 AM

I know this is an old post...but just for any new visitors because of the links posted only one of them are actually worth looking at.

Simply put if you are using a carpc you probably aren't going to be doing consistent and major system logging/debugging..(unless you are troubleshooting). That being said this is a good start for minimizing diskwrite and save filesystem space:

Edit your /etc/fstab as follows:

Code:

[Device] [Mount Point] [Type] [Options] [Dump] [Pass]

tmpfs      /tmp            tmpfs   noatime     0         0
tmpfs      /var/tmp       tmpfs   noatime     0         0
tmpfs      /var/log        tmpfs   noatime      0         0
tmpfs      /var/spool     tmpfs   noatime      0         0

I have a eeepc and this is how I have my fstab...more or less.. If you do need the logs for troubleshooting or debugging as I have had every once and a while you could always comment out the entry (using the # symbol) and reboot which will create your logs as normal.

Code:

[Device] [Mount Point] [Type] [Options] [Dump] [Pass]

tmpfs      /tmp            tmpfs   noatime     0         0
tmpfs      /var/tmp       tmpfs   noatime     0         0
#tmpfs      /var/log        tmpfs   noatime      0         0
tmpfs      /var/spool     tmpfs   noatime      0         0

I definitely would not suggest making your entire /var temporary as this would impact packaging and other stuff...

User Name		Remember Me?
Password