autofs confusion

[Jason L Tibbitts III]

>>>>> "RL" == Rich Lafferty <rich+rhl at lafferty.ca> writes:

RL> Just to confirm, do you *mean* "localhost" there? (You mentioned
RL> above that you expect to mount off of a central server, so I'm not
RL> sure it's not a slip.)

Sorry for not explaining more clearly; for often used applications, we
keep a local copy updated via rsync.  This is stored in /cache on each
machine.

RL> That's only the symptom; you need to dig deeper to find the
RL> cause. Can you mount the filesystem by hand? Is an nfs server
RL> running on localhost in the first place?

An NFS server has never been required to be running on localhost in
the past.  autofs would use bind mounts to access filesystems on the
local host and bypass NFS altogether.  Remember, the exact same setup
has been running on everything from FC1 to Red Hat 6.2 (probably
earlier but I can't remember back that far).

I figured out how to increase the verbosity and got this:

Apr 29 13:31:49 compute08 automount[2010]: attempting to mount entry /usr/local/bin
Apr 29 13:31:49 compute08 automount[2020]: mount(nfs): no host elected
Apr 29 13:31:49 compute08 automount[2020]: failed to mount /usr/local/bin

So it indeed seems to be trying NFS instead of using bind mounts as it
used to do.  Now I'm off to try and figure out why; it would be a real
shame if that had stopped working.

 - J<