[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
It's Alive
Date: Wed, 20 Sep 89 11:23:04 EST
From: bouma@cs.purdue.edu
Why is it that when I make changes to the NAMESPACE some of the
machines don't notice? For instance, one machine on the network went
away, and we got another (unix machines), so I edited the namespace
appropriately. But some of the machines still list that deleted machine
as a dead connection in PEEK on NETWORK? Rebooting does not help, but
Reset Network gets rid of it until the next time I reboot.
How is it that all the lisp machines know about the new machine I
added, but some still think the one I deleted is out there? Is info
cached in a world file overriding the present namespace? Does this
behavior conform with the way the namespace works, or it is a local
bug in our system?
Machines will only query the namespace server about a host if they
believe that it has changed, or if you force them to by using the
:CHECK-VALIDITY message (the Show Namespace Object command does this).
Deleting a host doesn't seem to count as a change for these purposes.
Therefore, hosts that existed when the band was saved will continue to
exist until you build a new band.
In the case of new machines, whenever you type a host name that the
local host has never heard of, it queries the namespace server. So, new
hosts always show up immediately.
Normally a Lispm will only query automatically about a particular host
once per boot. The variable NETI:*VALIDATE-ONCE-PER-BOOT* can be set to
NIL to cause it to query more often. If machines frequently go more
than a few days between boots I suggest you do this.
A few days ago I logged into a freshly booted machine as
LISP-MACHINE. I then edited and attempted to save a file. Instead
of saving the file, the machine writes, "Who are you really?"!
Zmacs knows that LISP-MACHINE is a fake user, so when you save out a
file it asks for a real user name to use when setting the author of the
file.
barmar