To my amusement... or horror today, I found that the RaQ3 and RaQ4 are slightly different machines! And although the RaQ3 looks almost identical on the outside and inside, it does not preform upto what a RaQ4 can deliever...
This was what happened :
I got 5 units to run the test :
2 units were a RaQ4 with 450Mhz
3 units were RaQ3 with 500Mhz
I used the same HDDs and also the same RAM chips so that the test is fair.
Everything was fine to the bit where I needed to set up DirectAdmin.
The RaQ3s WILL and always fail when you get to the bit when trying to configure some modules for Apace ( I think it was CURL ) The installer will say that there is a hardware problem and exit the setup but still allow you access to a glitched up DirectAdmin panel.
The RaQ4s however will complete the installation like any other normal CentOS installation with no errors or glitches in the panel.
I have used 20GB HDDs ( U5, U6 or WD ) as the IDE0 and a 160GB Barracude as the /home directory.
Has anyone with more experience with the internals of a RaQ3 and RaQ4 have a explaination of hy this is happening? Because If I can't get around this problem... I will have 50% of my cobalts unable to run DirectAdmin
To my amusement... or horror today, I found that the RaQ3 and RaQ4 are slightly different machines! ... 3 units were RaQ3 with 500Mhz
I think you need to supply more details. I run cURL on both machines, and (so far) have found them to run the same software, interchangeably. However, I'm not saying you're wrong. I recompile all software for my systems on each machine -- I hardly ever install binaries. I've never run DirectAdmin.
The most likely problem is your CPU "upgrade" -- you say your RaQ 3's are 500Mhz -- which means you've upgraded the CPU with the speed hack, and this is known to occasionally cause some instability vs. the original 300Mhz Raq 3 systems.
Also, you hard disk setup may be stressing the power supply, and on top of that, there are known LBA24/LBA48 problems with the systems, and those might be the source of differences, but I can't see how that would reflect itself at the application level.
If you really want to experiment, back your systems down to 300Mhz, pull the second disk, and use the original CPU, and see what happens.