I have a RAID5 software raid in Centos 5.8 that keeps going read-only. A disk dropped out of raid a few weeks back, but when I added it back in, everything worked fine... for a while. Now mdadm shows all the disks are happy and working fine. I pulled the drives (Seagates) and tested them using Seagate Diagnostics and both short and long drive tests come up clean, but the problem persists.

I have run fsck several times and it always finds errors, and fixes them. The array will then remount fine for a few hours and then problems again.

They are also in a JBOD, but the other drives in the JBOD that are not RAID are working without issue.

These are fairly new drives (less than 1 year). I'm not sure what else to test. Any better drive diagnostic software out there? Any help would be great!

You might want to investigate the drive controller(s), connections and cables. If the drives themselves are alright it sounds like your issue is at a lower level causes writes to not get committed properly.

Well I've run fsck 4-5 times over the past week and everything seems to be working for now.

Have you checked what S.M.A.R.T. (Self-Monitoring, Analysis and Reporting Technology) says about the drives? I recently had a hard disk failure in my RAID array that I suspected for a while but nothing detected till finally my S.M.A.R.T. report gave me enough confidence to determine which drive was failing and replace it.
http://pkgs.org/centos-5-rhel-5/cent....i386.rpm.html

http://sourceforge.net/projects/smartmontools/
http://sourceforge.net/projects/smartsuite/
http://sourceforge.net/projects/smartlinux/

Good luck!