Hello all:

I am using an old Athlon-600MHz that I built in 1999 and which has survived with a few RAM and HD upgrades throughout the years. I've been running Ubuntu Dapper Drake on it for the last 1.5 years to great success, however, a few months ago I started encountering hard drive issues. The hard drive started needing fsck to be run manually every few boots, and sometimes, the hard drive would inexplicably change to being read-only a few hours into a session. When I would reboot, things would be ok again, but this happened every few days.

I figured the HD was old and I just needed to replace it (Maxtor IDE 133 120Gb), so I just bought a cheap Seagate Barracuda IDE 100 160Gb and installed Ubuntu Feisty Fawn on it. Much to my dismay, I'm experiencing the *same* problem on the new drive--the system goes read-only after a while. The only thing that I can think might be wrong is bad RAM. Any other suggestions or tests I can do to confirm this, aside from buying new RAM or a new system altogether?

Thanks in advance!

Cameron

Run memtest for a while - should be on the Ubuntu CD I'd imagine.

What does 'dmesg' show after the drive goes read-only?

Thanks for the suggestion. I'm going to run it for 24 hours to see what it finds. I'll keep you guys posted!

Well, now it hasn't gone read-only all week. But it has given me strange errors like SegFaults when I try to apt-get upgrade, of course, 2 hours later, apt-get works fine. It is very strange indeed. We'll see how this memtest comes out--my RAM is the only thing I can think of that would cause intermittent problems of this nature.

So I ran memtest86+ on it (right off of the ubuntu CD), and it turns out that my RAM is rife with errors. I reported something like 50,000 errors in the battery of tests, in two particular spots. Fortunately, this machine is old and only has 384Mb of RAM, so I'm just going to invest and grab some cheap, new RAM rather than playing with the badRAM kernel that can just ignore the bad spots in bad ram.

Thanks for the help, everyone! I'll definitely be testing my RAM dimms in the future to look for errors. RAM errors are so sneaky because they look like other kinds of problems.

I also figured out why the hard drive went readonly after a few hours, in case there is someone who reads this in the future.

By default, ubuntu sets up the main hard drive to be mounted with the following settings (found in /etc/fstab, or just type 'mount' at the prompt to show all mounted systems):

/dev/sda2 on / type ext3 (rw,errors=remount-ro)

Thus, when my drive encountered errors from the bad RAM, it would remount it as read-only. Hence things going awry after a few hours. Anyway, I hope this helps the perusing masses.

---------------
Why not compile and use badRAM in the kernel and be absolutely sure that it is a RAM issue before you buy new? Besides, if it works and performance isn't an issue afterwards, you'ved saved enough money to take the little woman out on a cheap date.