Lots of good advice above.
Thanks.
It's a tradeoff of time vs. money. If a new 80 GB hard drive costs $40, as they did at CompUSA last week, decide how much troubleshooting time would pay for that. Of course, on principle, it is much more interesting to solve the problem without replacing the hardware.
carl
Again, that's assuming the problem is with the hard drive. If it's not, one could spend literally hundreds of dollars and associated hours replacing hardware that is perfectly good.
Bad RAM or a failing motherboard don't always show themselves as what they really are. Often the first sign is data, or worse, structure corruption on the hard drive. Like I said, I recently had such an experience:
The first sign of trouble was bad inodes, then corrupted data, then misbehaving apps. A first guess would say bad drive. I backed up the drive immediately to another drive.
Next I ran Memtest86. It showed bad RAM on both of the two installed modules - way too coincidental that both would go out at the same time.
So, I tested both RAM modules in another known good box [1]. They tested fine. I ran the manufacturer's diagnostic software on the suspect drive in another known good box. It tested fine [2].
The PSU tested fine (at least to the point I could check it). The last possible point of failure was the motherboard itself, which I replaced. All the original components are still working fine with the new mobo.
Had I just tossed the complaining hard drive, I would possibly have gone through at least another round of the same problems, at the risk of damaging more hardware in a cascading effect, not to mention possibly losing data, all because I had jumped to a faulty conclusion, mis-diagnosed the problem, and replaced the wrong part [3].
While there is little that can be repaired on modern computers, there is still a fair amount of troubleshooting which can be done to narrow down the problem to the few components that can at least be swapped out.
Keep in mind that data only gets to the hard drive at the pleasure of a properly working RAM, which itself is at the mercy of the motherboard.
All Ralph really knows at this point is that Linux doesn't like /what/ it sees on the hard drive. It doesn't know /how/ that "what" got corrupted in the first place.
[1] In such a situation where I am suspicious of RAM, I run a /complete/ test with Memtest86. That means testing *all* memory with *all* tests. Just running the basic tests is not good enough. With a moderately fast CPU it can take better than 24 hours to test a gigabyte of RAM.
[2] I run as thorough a test on the hard drive as the diagnostic software will allow. As with RAM, running just the bare minimum tests is not good enough for real peace of mind (and pocketbook).
[3] Not to mention pissing off my wife when I tell her that I have to spend even /more/ money to fix something she thinks I spend way too much time and money on already [4].
[4] Unless, of course we're talking about *her* computer. That's different [5].
[5] Although usually accompanied with "Why can't you just fix it?". And "Why do I always get your hand-me-downs?".
--
Best Regards,
~DJA.
-- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
