I inherited a custom-built with an ABIT-IL8 motherboard (I had the machine checked out by a computer technician and the installation appears to have been done correctly), I just installed RedHat Ver. 5 and everything seems to be working fine with the exception that every time I run md5sum for the same, static file, I get a different string. Has anybody run into this problem? Red Hat tech support doesn't have an answer for this, I accept the fact that all bets are off on custom-builts, I'm putting this out there to see if anyone has any thoughts/experience with this. Of course, md5sum is important because I cannot reliably download packages, etc. for my system.

Could you paste an example of your md5sum command on the same file?

You are talking RHEL5 and not RH 5 right?

You should probably be using sha1sum instead of md5sum.

I would run memtest86+ overnight. Flaky memory can take a few minutes to show up, come and go seemingly randomly, and can cause all kinds of strange problems (bad dvd burns is one that drove me up a wall) that are nearly impossible to diagnose.

Good Luck
Lazlow

An example follows. Also, something that may be note worthy: if I do a reboot and execute md5sum multiple times, the first time takes about 6-7 times longer; ~13 sec. first time, ~2 sec. afterwards. I did this little experment 5 times. T

[root@localhost Desktop]# ls -l rhel-5-client-i386-disc1.iso
-rw-r--r-- 1 root root 662095872 Jul 19 10:36 rhel-5-client-i386-disc1.iso

[root@localhost Desktop]# md5sum rhel-5-client-i386-disc1.iso
bcf9c6c5cc7c1f9ae8b845a12c6ed401 rhel-5-client-i386-disc1.iso

[root@localhost Desktop]# md5sum rhel-5-client-i386-disc1.iso
868c0b74c06d10c3c9f2ca446ef3243d rhel-5-client-i386-disc1.iso

[root@localhost Desktop]# sha1sum rhel-5-client-i386-disc1.iso
07ee5f98f1acd7c0ab5de20e77a613f279b6975f rhel-5-client-i386-disc1.iso

[root@localhost Desktop]# sha1sum rhel-5-client-i386-disc1.iso
943d2d583bf688b47967953da3708edb35d6e0f4 rhel-5-client-i386-disc1.iso

[root@localhost Desktop]# ls -l rhel-5-client-i386-disc1.iso
-rw-r--r-- 1 root root 662095872 Jul 19 10:36 rhel-5-client-i386-disc1.iso
[root@localhost Desktop]#

It's normal that the first time takes long (at least the way you do it). First time, file is on HD, second time file is still in cache (memory).

Can not explain while reseults are different each time; suspect memory as well.

You're right about the longer first time response: I think I was just grabbing for something. I/O always takes longer on this machine on the first access after a (re)boot, then it's consistent for any file(s) after that... I'm sure it's some initialization logicistics.

Anyway, I don't think the problem is memory related. I ran memstat86+ for 6 full passes of their test suite and no errors. Considering md5sum never runs correctly, I don't think it's due to some subtle memory, or memory access, error. Also, I happen to have a duel boot on this machine and my other OS is Windows XP. I took the ISO file I used in my example and copied it into that domain along with md5sum that I'd downloaded from the public domain and ran it in DOS. It worked perfectly: it matched continuously and matched the string published by Red Hat for that ISO. Previously, I ran the same md5sum.exe copy on the other 5 downloaded ISOs Red Hat has on their RHN site and it matched perfectly. FYI, the reason I was using MS to download Linux was due to logistics which I won't bore you with.

I think there is some "collision" between Linux I/O and the IL8 I/O, or God-knows-what. I come to the conclusion that Red Hat tech support is right: If you don't have a full system that is supported, then all bets are off. My take on this is if you don't buy a full system with the OS installed, and a copy on media, then you're going to be spending a lot of time at your keyboard when you could be out dancing.

Any thoughts, comments... since I'm a newbie here could someone let me know if this gets out to everyone who got involved? Thanks to everyone for you help, you've got some very good ideas!

If anyone is still interested, I found/fixed the problem: It was in the BIOS settings. Thanks again for everybody's help; your suggestions helped me narrow down the possibilities very quickly.

Yes, I am curious about that. Can you post what BIOS settings trigger such a behaviour? Just for completeness of the thread. Thank you.

I'm using Phoenix AwardBIOS and with reference to their set-up utility's menu: Integrated Peripherals -> On-Chip IDE Device -> On-Chip SATA Controller -> Enhanced Mode. Originally it had been set to Automatic. I had noticed in another LQ forum that someone else had migrated to Enhanced Mode, so I took my que from them. My CD-ROM/DVD has a PATA interface and my hard drive a SATA interface.