Short version --
random errors during install of Fedora Core 5, that would seem to indicate, bad media, CD drive, or HD
Have tested media twice on two separate CD drives.
Tried at least 6 CD drives, for install.
Tested HD with utility on EBCD (emergency boot cd) - and quick and advanced tested came back good.
The HD IS OLD though. Big enough (13.6GB), but old -- only has 428KB cache. (could this cause problem? - if not what else could?)

System info -
Trying to install Fedora Core 5
ASUS K8V-MX mobo (built on VGA and LAN)
AMD 64 Sempron 2800+
PQI PC3200 256MB (though only 192MB is registering)
Long version --
I am trying to install Fedora Core 5 on a system that consists of brand new MB, CPU, and RAM. (built on video). And really old hard drive, case etc. I know once I had problems with installing fedora, because something wonky with particular CD drive. However, I have tried, like 6 different CD drives - and there is no correlation, in exactly where it freeze's and which CD drive is in there. Mostly it freezes - just after it tells me where the log, and kickstart file will be. Then I click next – it goes to screen where files normally start copying and nothing at all happens. I’ve tried text and gui mode. Sometimes it doesn’t freeze there, sometimes it will format, etc. then freeze when trying to copy install image (says: Probably not enough space on hard drive (plenty space on HD for this)) -- or sometimes get past that start installing packages etc. -- and then all of a sudden say it cant read/copy file -- I select retry and it gives error about not being able to unmount CD drive. I have checked the media twice with 2 separate CD drives just to make sure. I have tested HD, and it is fine, albeit old. (13.6GB 428KB cache) (The only thing I see for HD requirements in fedora documentation is harddrive space available requirements. Never anything about cache. Could the small cache size be causing my problems? If not – any other ideas what might be causing my problem?

I'm guessing it's the ram. 192MB is listed as the minimum system requirement for a graphical desktop. The installer may not be working in graphical mode because of too little ram.

Also, sharing ram with the onboard graphics can cause problems. It may be fixed now but grub used to have problems with this type of configuration because the graphics system would grab a certain block of ram that grub needed to use. Something in the installer might be having a similar problem. The installer basically runs in ram and if there is not enough you could get erratic behaviour.

Try running memtest on your ram for starters to make sure what you have is OK. Check and make sure the ram is well seated as well. If you can lower the amount of ram that the graphics adapter can glom onto in your bios setup, try doing that as well.

I assume from your post that you did this build yourself. Another issue you can run into is a dodgy power supply. That can cause all kinds of erratic problems, particularly when the voltage to the ram goes out of spec. This is most likely to occur when your optical and hard drives are drawing a lot of current which is typically the case for an install from a cd. The only way to test that is to swap out the psu unless you have some specialized test equipment.

Finally, it could be an overheating problem typically caused by the heatsink/fan not being properly installed. Try running with the sides of the case off. If the symptoms improve, you probably have an overheating problem.

I have two of these sticks of RAM (I had just been using one - because the other stick is for the other mobmo, and CPU i bought). I've run memetest on one of them for 17 hours, came back 154 passed - 0 failed.

I tried install with both sticks of ram in there in case running out of RAM was the problem -- it did get further than ever (asked for disc 2), but still had an "unhandled excpetion error".

I am now running memtest on both sticks of ram together.

I also lowered ammount of shared onboard emery for VGA to 8MB. Is that enough for web browsing? (if I ever get these things working!!) - Also in doing this I realized why the mobo was only showing 192MB for the sticks of RAM I ordered. It was subtracting out the ammount it was ussing for the onboard VGA. Because one I lowered that, all of a sudden the RAM chip registered as a larger ammount. The mobo appears to be hiding from the other hardware that this memory even exists. memtest86+ read the chip like it was a 192MB chip, and now says it's a 248MB chip.


As far as overheating -- I've checked the CPU temp in the bios acouple times - and the highest I ever saw it was 39.7 C. During some of the attempts, as well, the case has been open -- just because I was switching CD drive,s out so much ect. So Im pretty positive, it's not related to overheating.

As far as the power supply -- it is an old power supply. In the bios -- the voltages appear normal -- though as you said - it's most liekly to have problems when CD and HD are drawing a lot of power, and that is ofcourse not the case when i am in the bios. What is normal as far as voltages go? How far off from the spec does it have to be before it really starts causing problems? (i.e. i assume 3.29 is OK for the 3.3 voltage. How low would it have to be to cause problems.)

3.29v is certainly OK. How far you can go out depends on a lot of factors. Post your psu manufacturer and the wattage. Also, not to belabor the obvious but make sure you have plug 'n play disabled in your bios setup.

I haven't installed RH since back in the 9.0 days before they went fedora. IIRC there was an option to go to a command line where you could type in the kernel name, usually "linux" and pass some boot parameters before proceeding with the install. If you can do that, try running:

linux noapic nolapic acpi=off

acpi can still cause a lot of problems with certain kernel/motherboard combinations and the above will eliminate those types of problems during the install. If that doesn't work, try:

linux noapic nolapic acpi=off nodma

I'm trying to eliminate some of the typical sofware linux compatibility issues. However, I'm still suspicious of the psu.

Another thing I would try is running a livecd like knoppix or kannotix and see if you run into problems. If these run OK, it's most likely a hardware compatibility issue with the fedora kernel.

I treid installing FC4 -- and it installed without a hitch and seems to be working fine.

Do you think I still need to worry about the PSU? Or should I just forget about it, unless the computer starts acting up again.




BTW PSU is APEX Allied AL-A350ATX 350W 20/24-pin Power Supply
All I have in the case is one IDE HD, one IDE CD reader, and mobo.

Well Allied is on the "Bad" list here for what it's worth:

http://forum.pcmech.com/showthread.php?t=131195

And it's on the "Crap" list here:

http://forums.nvidia.com/index.php?showtopic=4146


All this should tell you that you do not have a quality psu. That does not mean it is defective now, just that you're more likely to have problems with that psu. I learned the hard way that it is penny wise and pound foolish to skimp on your psu when doing a build. It makes no sense to spend hundreds if not thousands of dollars on a new build and power it with a cheap psu capable of taking every component out if it blows. I'd look to replace that psu with a quality one from antec, sparkle or enermax.

However, since FC-4 installed OK, you probably don't have a problem with that psu just yet. If you start having erratic issues with your current install, that psu is the first thing I'd take a look at.

Thanks for the reply. Since this is just a computer for my work that will have minimal web browsing use (and thats it) -- and we actually only spent about $135 total for new parts -- I think I'll go with your suggestion to let it be for now -- but look into new PSU first thing if I have an erratic problems.

Off the main subject of the thread:
I'm just curious -- because I also just built a new PC for myself at home, that I did spend nearly a thousand building. So i have considerably more concern for it's well being. I noticed thermaltake -- was on the good list at the first link you referenced -- but was not bolded, and one of the BEST. Do you think thermaltake is a good enough powersupply for my most prized PC? Or do you think I would be considerably better off with one of the best of the best?

I wouldn't worry about since you seemed to get it working all well. It's true that "crap brand" power supplies can cause issue, but so can premiere ones if they are defective. It is of course always better to buy a good quality PSU, but it's not the end all be all of computing nirvana either. Really, the truth is that all PC PSU are crap in the big picture of power supplies. They really aren't what you can consider a "true power supply" considering how dodgy and misleading all the manufacturers are with their specs. They are what they are -- a cheap impmentation of a power supply. If you want to know why I say that, just look at power supplies for any device that non-PC related. It's two different worlds in terms of quality, specification, and tolerances.

So what I am saying is that more expensive power supplies tend to be better than the el-cheapo ones. Like the ones you might buy from Fortran, PC Power and Cooling, OCZ, and the list goes on and on. If you want the absolute best PC PSU then it's PC Power and Cooling, but they also cost an arm and a leg. Bottom-line is you get what you pay for. However, realize that alot of the "good brands" are just as deceptive and the "crap brands". This whole dual 12V rail paradigm is BS for instance. If anything it's nothing more than a marketting gimick because they truely aren't seperate 12V rails.

So in your case, if your system is working, don't worry about it. If it ain't broke, don't fix it. If you want the extra piece of mind of having a higher quality PSU then go for it, but there is still the possibility that it may not solve your issue since it may not be what is causing them. Your component are more likely to get cleaner power for a high end PSU which doesn't hurt and is probably good, but on the other hand, you may go years without any problems from your el-cheapo existing CPU. What I'm saying, is all too often, alot of users see going out and buying a cadillac quality PSU as the end all be all fix for everything. The fix is dependant upon what the problem is. Also, anything can be defective no matter how great the quality. A person could go out and buy a high end PSU and still possibly have it fail or die within a week due to luck of the draw. The probability is that the higher end ones are less likely to fail and are more likely to be reliable though.

Anyway, I've found anomolous and intermittent problems to usually be caused by low quality memory than low quality PSUs most of the time. Especially when there is the slow creep of data corruption, random lockups, and crashes into the system. Like my dad's system that he built, he had a really nice PC Power and Cooling PSU in it, but it was extremely unreliable and would crash alot, show up bad clusters in Windows, freeze-up, display wierd error message during installs, you name it. He had me come over to work on it and try to fix it. It ended up being bad memory. And guess what. The memory passed every kind of memory test I could get my hands on. Memory testing programs are pretty much worthless IMHO, as are memory testing hardware since they to pass bad memory (false positive) a great deal of the time. The way I tested it in his system, was running with 1 stick of memory over a period of time. He originally had 2 sticks, but I would run with 1 over a period of time and see how the system behaved then try to the other one. I found out exactly that it was particularly 1 stick of memory that was bad while the other was fairly good. Neither one was what I would consider reliable though. So he ended up ordering some Kingston memory modules and his system rock-solid from that day on. Even in Windows. We ended up throwing out the original sticks that were giving the problems. They were some OEM bargain of the month. Who know what brand they were. That's another thing, I tend to stay away from OEM crap.