You are not logged in.
Hi
What could cause files (until now I've only noticed the problem on executables) on a etx4 fs on mdadm RAID 1 with 2 WD Caviar Green 1 TB HDD's to randomly become corrupt?
Usually a executable randomly segfaults or some other strange thing happens. Then I reinstall the package and it's ok again.
It's a strange issue. I've had it happen to find, ruby (2 or 3 times that I've noticed), php, rtorrent and probabily more files that I didn't notice. I have run memtest86 that comes with the latest arch installation image (2010.05) to test the RAM and it did 5 passes over the course of 8 hours with no errors.
Could it be a problem with mdadm?
Offline
It could be.. it could also be a failing drive.
Try using 'find' to get a sha1sum (or sha512sum if you prefer) of all your executables. When you find a corrupt one, check the sha(1/512)sum of the file (or just run the check of them all regularly).
I would then drop each disk in turn from the array and recheck, to find out if the corruption is happening identically to both disks or not.
Hope this helps
Offline
It happend again just now. I'm working on a website and php just didn't want to work anymore. The server just returned the standard 500 Internal Server Error message. I reinstalled php and restarted the server. That didn't work. Then I rebooted the system and now it's fine.
Major red flags going up! Xyem, thanks for the suggestions. I'll try to checkout my harddrives in the next few days, but I can't do it right now becouse of work (all of my projects are on them...great).
I'll post a update on what I found out.
Offline