UNIX for Advanced & Expert Users

I've got a E6500 with 6 cpu's over 3 cpu boards.

we lost one cpu should the box keel over ??

I wouldn't have seen this as a Single Point Of Failure

Kie

I know for sure that a box can panic and reboot for failed CPUs. In your case, it depends a great deal on what the system was doing when the crash occured. If that CPU was executing system calls for the OS, then it could have crashed the box.

CPU crashes can have very unusual affect on a system. I have seen everything from no effect to a crash. However, on my HPUX systems the host usually will PANIC ( a system call) and reboot then disable the bad CPU. However, I had one the other day that didn't disable it. It came back as though it was a good one. I had to replace it soon after so it wouldn't bring down the box again.

Having multiple CPUs does not necessarily translation to fault tolerant processing (as you are finding out). The following BYTE article might be useful:

http://www.byte.com/art/9702/sec11/art7.htm


Here is another good article:

http://sunsite.nstu.nsk.su/sunworldo...-clusters.html

Hope these help.