Very Computer

One of our larger NIS domains, recently turned up an odd problem.

The domain contains many Sun's, some with 4.1.x, some 5.3, a handful
of alpha's (1.1 and 1.2), and a single sgi (with Irix 5.2).

One of the 4.1.x machines (ang1) lost a harddisk.  The Irix machine's
(puffin's) automounter became confused while trying to nfs mount a
filesystem from ang1.  The other machines seemed fine.  I don't know
if this is an Irix 5.2 problem, or a local config problem; that's why
I'm posting.

The users on puffin have no need for the data on ang1.  There may have
been someone on puffin trying to cd to one of the directories on ang1,
but it wouldn't make much sense; it's not a high-end machine, and the
machines belong to different departments.

Does anyone have any suggestions?  I'd prefer to avoid local
/etc/auto.master's if possible; it seems that in such usage, one may
almost just as well use /etc/{{v,}fstab,checklist}.

Out of curiosity: did SGI purchase source to sun's automounter at one
point (if so, presumably prior to the multithreaded autofs-version?),
or is this a rewrite?

The error that seems most critical is:

        Oct  7 15:32:02 automount[190]: host ang1.eng.uci.edu not responding

But there were tons of similar messages about mount attempts against
filesystems that were alive, on which access was denied.

Does Irix just do that, or was someone digging around looking for
interesting stuff?