> I had seen similar problems with the startup, but in my case, it
> was klogd that tended to die. It's inelegant, but adding a "sleep 1"
> between the 2 invocations in the startup script has solved the problem
> here. Someone needs to trace down the race condition.
May it's because these processes "auto-background" themselves. If
they do that before opening their sockets, the second process might
try to access the socket that's not open yet. But wouldn't that give
an error message?
>
> Bill Earnest wde@fast.net
> Fax (610) 432-7477
Ulrich