So the only way is to reinstall?

Hi -

I'm surprised that RedHat didn't auto-detect that you had multiple CPU's (unless, of course, your VM was only configured for one CPU when you originally installed)...

... but no, there isn't any magic configuration setting you can change.

You *can* install a new kernel.

But honestly - it might be easiest for you to save a copy of your current VMimage, and just build a new image.

IMHO .. PSM

It's been a while since I did it last, but I believe that all I had to do was a "yum install kernel-smp" and edit grub.conf so "default" pointed to the added paragraph. It was not particularly complicated.

Is it possible to use the old disks attached to this virtual machine on a new install?

I will continue with this tomorrow tuesday.

Hi -

1. I believe jonesr is correct

2. I've never used ESX ... but I've used VMWare/Workstation a *lot*.

In Workstation, I'd just:
a) clone my existing image (i.e. make a backup)
b) Try installing the new kernel (e.g. "yum install kernel-smp" and editing "grub.conf")

3. I certainly wouldn't suggest that adding SMP support will necessarily "fix the problem" (I suspect you'll probably need to investigate further to find the actual problem)...
... but taking advantage of both CPUs is definitely the first thing you need to do.

'Hope that helps .. PSM

Installing a new kernel is probably a complete cure. However, I would suggest to also check the applications installed. Sometimes, they also discover during installation if there is a single CPU or a dual CPU. Some may use different code for each of these cases. This practice has hopefully been discontinued by now, but I know of at least some cases in the past (and, yes, this should properly be taken care of by the kernel, not the apps - but still...).

Just my 2 ct..

I will try to use VCB (Vmware Consolidated Backup) to create an image of my virtual machine, and then try to do as suggested!

I will update this post when i've done this...

I made an Imagebackup of the virtual machine and converted it into Vmware Workstation 6.
Started the server just to verify that it came online OK
Shutdown the virtual machine
Added an extra CPU and tried to start the server again...... AND IT WORKS!
when using top-command i can see that it uses 2 CPUs now.

So, i didn't have to do anything but to add an extra CPU! How is that?

I have to try this in the ESX environment before i do any changes to the "real virtual machine"

I've added an extra CPU to the server and it works.
I will await performance result after this change in config.

The extra CPU did much in performance.
server is still kind of busy sometimes but not nearly as much as before...


Is it true that an 32 bit OS should not have more than 4GB of RAM?

I've heard that if we need more than that we should consider an 64bit OS instead.

This is right, because a 32 bit CPU (without using any tricks) can access up to 2^32 byte of memory and that is 4GB. A 64 bit CPU can address much more (but not necessarily 2^64, as the number of bits must not necessarily be propagated to the address bus size).

Rainer

Could this be an performance issue as well, because we have 5.5GB RAM in this server.

32-bit systems are not limited to 4 Gig of RAM. PAE hardware has been around for years - as have supporting kernels. Using more than that has a small impact - as the memory gets larger, the impact increases. You will not be able to measure this normally - certainly not by looking at process CPU usage.

Merely changing to a 64-bit OS will not necessarily be a solution. If the application code is still linked as 32-bit, the major benefit will not be to that application. If the distro is not multilib, the application may not work at all.

Here's a link that has collected some useful information on PAE.

http://www.spack.org/wiki/LinuxRamLimits