Okay, so this is the deal. I downloaded the PCLinuxOS live cd and burned it as a cd image. The md5 checksum matches the one on the server from which the iso was downloaded. I've since installed the OS on my laptop (a gateway) and my roommate's computer (a dell desktop). However, within right away we both were experiencing crashes. My laptop tends to crash more repeated than his desktop though. I received no errors for either install and have tried different kernels, we have totally different hardwire, it doesn't matter what the computer is doing (i.e. surfing the web or even just running the screensaver), and the only thing that is in common is the install cd. My dad downloaded PCLinuxOS .92 (the same as I have) and has had no crashed whatsoever. I had read that maybe the cpufreq package for cpu monitoring can cause crashes so I removed it to no avail as well. So, I'm left thinking that maybe something is wrong with my live CD or the iso. So, I ripped the image off the live CD I made with KIso and the md5 of my copy doesn't match the original md5 of the iso the live cd was created from. So my question is....shouldn't it? If a cd is created as an image (an iso) than when you rip the image back off of the cd, shouldn't it be the same image and same size as what it was created from??? If so, this leads me to think that something is wrong with either the original iso that's on the server, or my cd burner is having an error while burning the iso and not telling me or something. But, why wouldn't I have problems installing if the cd is bad? Well, you can see my obvious dillema. Sorry the post is so long, but I'm really losing it here. I want to use PCLinusOS quite badly, but it's behaving just that, quite badly. Any ideas anyone?

Hi.

Looks like your CD either didn't burn properly, or was damaged after burning.
The MD5 from the download should be the same as the MD5 from the rip of the CD, which should be the same as the MD5 from the PCLinuxOS website.

Dave

Thanks for your rapid response! My next question...I've downloaded another iso of the same file from another location (different server) and the md5 checks out. Burned the newly downloaded iso to a cd, ripped it back off, and got a wrong md5 again. So, since the md5 is the same as the first cd I ripped, I can assume it's a coaster too? Is this a K3b problem? What should I try to get a good burn or get around this problem? Any ideas?

If the ripped MD5 is the same as the first ripped MD5, I'd think there's something odd going on.

Try doing 'md5sum /dev/cdromX' (replace cdromX with whatever your CD drive is) with one of the burned CDs in. See which MD5 that matches.

Dave

Well, you are wise beyond MY years... The md5 generated by the command 'md5 /dev/cdrom' is indeed the correct md5 (i.e. it matches the pclinuxos md5 from the server). So maybe for some reason k3b is not giving me the correct md5 of the cd ripped iso. However, at least now I know that the md5 on the cd is indeed correct, so a bad burn is ruled out? So I wonder why we both experience crashes from this cd install, but my dad installed the same OS with no crashes? We have all different hardware with nothing in common. Is PCLinuxOS .92 truly that unstable? Or is something else about my install fishy? I know it says it's "under heavy development and should be considered beta," but surely it can't be this unstable as I haven't heard anyone else complain about crashes. It's crashing several times a day...and that's just not like a linux OS....wish I could find a solution. But your help has been greatly appreciated!!!

I'd say a bad burn is indeed ruled out.

Do you get any messages in /var/log/messages (or any of the messages.X) logs after a crash? Anything about processor throttling or errors on CPUs?
Have a look in /var/log/xorg.log as well.

Dave

I've posted my /var/log/xorg.log here.... http://www.angelfire.com/linux/hardwire/xorglog.html As for the messages.X files, they are all blank. As is the /var/log/messages file. I very much appreciate you taking the time to help!

Hehe, maybe they are blank because it hasn't crashed yet? How can I post them after a crash....will the messages still be in there after a reboot?

Hmmm.

The messages should survive a crash.

Try using the machine, and check 'tail dmesg' every couple of minutes looking for anything interesting.
In my experience, most problems like this are due to power supply problems or overheating, not software.

Dave

P.S. Fedora's been utterly stable for me....

Well, I can do "dmesg" and read through it, but "tail dmesg" as root won't work. It says:

[root@localhost hardwire]# tail dmesg
tail: cannot open `dmesg' for reading: No such file or directory


Did I do something wrong? Yeah, Fedora is good stuff. I have Suse 10 on here as a dual boot and it's been rock solid too. But PCLinuxOS handles my ati card very nicely with greatly improved 3D support over Suse. Also, it has a ton of video programs I use, dvdshrink, kavitovcd, etc. And synaptic is a much added pleasure!!! It's really an awesome OS.

Oops. My mistake.

It should be 'dmesg | tail'

Dave

Well, I've check periodically now. Here's what I get over and over:

[root@localhost hardwire]# dmesg | tail
agpgart: Putting AGP V2 device at 0000:00:00.0 into 4x mode
agpgart: Putting AGP V2 device at 0000:01:05.0 into 4x mode
ISO 9660 Extensions: Microsoft Joliet Level 3
ISO 9660 Extensions: RRIP_1991A
NET: Registered protocol family 4
NET: Registered protocol family 5
VFS: busy inodes on changed media.
VFS: busy inodes on changed media.
ISO 9660 Extensions: Microsoft Joliet Level 3
ISO 9660 Extensions: RRIP_1991A

Nothing there seems out of whack to me...I bet you're right though. I think it has something to do with the cpu overheating or something. Again, there is an application called cpufreq that is supposed to control the cpu and some people were having problems with it causing system lockups. Someone mentioned that there is even a version of PCLos that is distributed without the cpufreq package. I've removed it in hopes it would help....which reminds me that it's been quite stable this afternoon in retrospect. Again, I very much appreciate you taking the time out to toss some ideas around.

Grab 'cpuburn-in' from my site here:
http://www.ilikejam.dsl.pipex.com/cpuburn-in

Run that for a while - it'll stress test your processor. If there's an overheating issue, you'll find out pretty quick.

Dave

I'll sure give it a shot.

Well, I didn't know how long to let it run so I did ten minutes....here's the last few lines:

425000 iterations complete.

Test Complete. No errors were found after running 429576 tests.

Should I go longer?