On 2.2.16, our main web server which is a Sun SS-10 with quad Ross RTK-625 Hypersparc CPU's made it four days and a few hours before crashing. This longer than it has made it on a 2.2.x kernel so I am optimistic that things are moving in the right direction. This is running Apache 1.3.3p3 at present. When it "crashed" this time, it was the weird mystery hang again. Absolutely no error messages on the console, you could switch virtual consoles still, and get a login prompt. Whatever you typed would be echoed but you'd never get a password prompt. It stops responding to the net, stops responding to a ping, but you can still type and see what you type at the console and switch virtual consoles. Since there's no OOPS or other error messages spin_lock deadlock messages, I'm not sure what to make of this. Prior to 2.2.15 it would usually spinlock deadlock within two days, probably averaging about once a day. With 2.2.15 that went away but then we got mystery hard hangs, and the above failure mode. Still under 2.2.15 the longest we made it was three days. Anton, I never heard anything with respect to the problems we experienced when trying to boot 2.4.0-test1 where it would freak when it came to executing the init scripts. Any ideas?