Random X restarts on new install
 

Dear All

I have just installed SuSE 8.2 (ftp install) and I am getting frequent, seemingly random X restarts. I have a feeling it may be hardware related. It happens most often when I am actually using the computer, I left it on for about 18hrs without using it with no problems, and within an hour or so of starting to use it again, X crashed. Also I ran memtest on the suse boot CD for about 20 hrs and got 4 memory errors - is this significant? Do I need to change my RAM?

Thanks

Tom

You would definitely suspect something is wrong with the RAM. But before you rush out and buy new RAM, make sure something else is not causing your RAM to misbehave. For example, is your BIOS set up to overclock the CPU or Front Side Bus? Either or both could cause timing errors which could upset proper functioning of the RAM. To investigate this, go into the BIOS and select 'defaults' (most BIOSes have this setting). This will set the BIOS CPU, Front Side Bus and other motherboard settings to conservative values which will not tax a serviceable RAM. Then run memtest again; if you still get RAM errors, you can be sure it is the RAM, rather than motherboard settings. If all seems well, try operating your computer for a while to see if re-setting the BIOS values has fixed your problem.

You could also try re-seating the RAM in its slot. Sometimes a bit of dust, dirt or corrosion on the slot's connector pins or the RAM's connector surfaces can cause problems. Take the RAM out and make sure there's no dust in the connector slot and clean the RAM's connector tracks with a paper tissue. Re-inserting the RAM stick can then provide a good connection. Try this one before trying the BIOS method I mentioned before.

Thanks for your advice

I have reseated the RAM and I'll probably run memtest again over night.

I forgot to mention in my orginal post that I get quite a few Segmentation Fault errors or other internal compiler errors when trying to make source files... If I re-run make a few times it usually gets the job done. Am I right in thinking this is another indiactor of faulty RAM? I have an all-in-one micro-ATX board, the only card installed is a LAN card, and I am using the on-board VGA and sound. I assume that it is a problem with the RAM, therefore, rather than any expansion cards, as I do not have any...

Thanks again...

Tom

Yes Tom, Seg Faults are normally an indication of faulty RAM, so you are on the right track. Your RAM may not be faulty, per se, but just incompatible with your motherboard. Check with your motherboard manufacturer's website; they usually have info on what RAM specs are compatible with their boards. Your symptoms seem to me to be of the 'incompatible' kind, rather than a physical fault. The only RAM I had trouble with failed in a big way - no boot, and lots of beeping from the motherboard that pointed to a RAM fault. If you do need to buy replacement RAM, go for a good quality brand that's compatible with your motherboard specs. Generic no-name RAM can also suffer the same problem that you're having.

You are saying that X is crashing? Only X? What about Linux?

Are you able to open a console and restart X? Or do you have to reboot?

If just X, then I wouldn't be looking at hardware.

If you mean that Linux is crashing and you have to reboot, then RAM is a possibility, but so is the power supply or the motherboard. In fact, I would pick motherboard, then RAM, then PS more or less in that order when troubleshooting a problem.

The motherboard has onboard voltage regulators and one of the more common failures is for one of these regulators, or the components associated with them, to fail at least partially. This leads to intermittent behavior; lockups and crashes mostly - usually under heavy load.

jiml8 is right about other possible causes for your problems. More info here on power supplies:

http://www.linuxquestions.org/questi...lys#post323816

Your power supply would be my first port of call for checking after you run memtest again. You should be able to get the power specs (current for each voltage rail) from a sticker on the side of the power supply case.

On motherboard failures, the most common of recent times has been a fault in capacitors supplied to lots of motherboard manufacturers that has just started to make its presence felt. If you see capacitors with rounded tops (capacitors should appear as small cylinders with perfectly flat tops) then you have this problem. More info here on that:

http://www.timcousins.com.au/

(From this page, Click on Faulty Capacitors in the sidebar, then on Motherboard Failures, when it appears.)

I have come across a motherboard affected by this (Gateway, made by ASUS in mid-2001) and fixed it for a friend by replacing the capacitors. An addition to jiml8's last few words: 'usually under heavy load'; compiling your kernel produces a fairly heavy CPU and memory load.