2 Intel Xserves, both running fast and both unable to connect to Server Mon

Question

Hey guys -
Since this is happening to two identical machines, I'm wondering if I'm just not doing something right.
I have two Xserves, both essentially functioning as metadata controllers for an Xsan volume. They don't do anything else but they both run fullblast at all times.
When I initially set them up, I disabled LOM on them. However, I can't connect to either of them using the Server Monitor. This is being done locally through a belkin KVM switch, so it's nothing to do with firewall issues. They're both on two networks, a private one (for the Xsan metadata) and a public one that operates through proxy servers. Currently, neither of them can physically connect to the internet (owing to an error in the proxy settings I'm guessing), both other machines in the local network and see them and connect to them.
They're both the same configuration, with 2GB of RAM, and operating at 10.4.8. They have dual power supplies and are running into two separate UPSs for power redundancy.
I was wondering if updating to 10.4.10 would have an effect on them, or if there was a procedure to reset the PMU (or SMU or SMC or whatever they call it these days) on them.
I've been searching these forums for similar issues, but most of them seem to be related to intel xserves which actually HAD LOM turned on, and these guys do not.
If it isn't any of these issues, how would I go about thermally recalibrating both of them?

Answer 1

well, I done fixed it.
After I got LOM up and running (127.0.0.1 was the way to go), I was able to see that the CPU intake fans were all blowing at a strong 16K rpm. Not a good sign, even though there were no warning lights going off and everything was reported as green and normal. Since this SAN lived in a hushrack (two servers and two RAIDs all told), I first tried to isolate airflow out of the equation by removing the back enclosure and opening the front door. Even after the heat dissipated, it ran at top speed. I ran all software updates possible, installed the OS onto an empty hard drive, and still got the same thing. I yanked one of the systems out of the rack it was in and swapped the fibre card from slot 1 to slot 2. I started it back up before returning it to the rack and, goodness me, it topped off between 5K-6K. I was so happy with myself that I tossed it back into the rack, patched it all back in, and fired it up. Boom. After about a minute of being on, it rocketed upwards straight to 16K rpm.
So now it was a matter of playing the isolation game. I pulled fibre cables, I tested the power cables in both the UPS and straight into the wall, I pulled the KVM switch out and just stood by listening for any changes.
It turned out to be two bad ethernet cables, of all things. I'd plug the cable in and the fans would start their best jet engine impression. The primary metadata controller had a bad metadata cable, and the open directory master had a bad public ethernet cable.