We have a dual Xeon machine that we recently installed RH9 ("Shrike") on. When we try to start just about any program (including login!) it takes a very long time to do whatever it needs to do. Now, we have an identical (or almost identical) server with the same RedHat on it, and it experiences no slowdown. The only thing I could think of was incompatible or non-optimized binaries, but I don't see how that could happen, especially on an RPM-based system. It doesn't seem to be an issue with HDD thrashing. Does anyone have any ideas? How can I narrow down the cause?

that is scary for me to read, and rpm, in fact any package based distro is exactly where you would find un-optimized binaries. all the packages are compiled to a base architecture like i686 or i386, for a pentium or an older athlon this is ok. the average user does not need more. but when you get into dual xeon you are using packages optimised for a single cpu and base architecture, you are not making use of the smp beyond dividing the processes among 2 cpu's (and thats only if you have an smp kernel installed/compiled).

you can find an smp compiled distro/package set, or you can compile the stuff you use a lot and specially your kernel on your own (not recomended that you compile your own software in a package system, but compiling kernel is fine) or you can use a source based distro like gentoo or LFS (gentoo is easyer than most people make it out to be, just time consuming) this would probably be the best idea for a dual xeon system anyway, with that kind of speed/power you want a distro that isn't bloated to meet the average users every need, it will just slow the system down. with gentoo or lfs you can compile everything you need, and only what you need with full smp support and (you will have to check with gcc specs on this) full xeon instruction set optimization. if gcc does nto support xeon instruction sets you can use pentium III or pentium 4 depending on if it is a p3 xeon or a p4 xeon.

Well as far as I know, this system is booted on an SMP kernel. We even have a machine that runs 7.2, and even that came with an SMP kernel automatically. Anywy, I'm just concerned that almost identical hardware has no problems in the same situation.

I agree with you on the distro question, but it's out of my hands, as we need to keep this machine compatible with some proprietary software that is to run on it. Also, it is an upgrade from an existing machine that the entire company relies on. I made the suggestion of distro switch, but I believe the choice was made not to do that for the various reasons listed above.

All I've really done so far is run time and look at the process' distribution of attention, and that didn't really tell me anything. It didn't seem like the process was hanging anywhere, though maybe it was. I know that on my home machine, I was running mdk 10 and it would hang for a few seconds between xdm and the window manager starting up, but it only did this when I booted off the SMP kernel.

One program in particular is somewhat telling: it starts up a curses-like (maybe it is curses; I haven't tried ldd on it yet; it's doing a restore right now) GUI, and the background is redrawn way before the menu is displayed. So, I don't know if library calls are hanging, or what is going on. There must be some debugging tools I can run, but I'm not sure what. Does anyone know?

Edit: kernel version: Linux version 2.4.20-8 (bhcompile@porky.devel.redhat.com) (gcc version 3.2.2 20030222 (Red Hat Linux 3.2.2-5)) #1 Thu Mar 13 17:54:28 EST 2003.

Also, it looks like it's not the dual-Xeon machine I thought it was; it looks to have only one 1.4 GHz P3 CPU w/ 512 KB cache.