[5084] in Hotline Meeting
AFS cache scan and NFS server use by Athena Ops people
daemon@ATHENA.MIT.EDU (Joe Harrington)
Thu May 16 12:44:58 1991
Date: Thu, 16 May 91 12:44:42 -0400
From: jh@saturn.mit.edu (Joe Harrington)
To: hotline@ATHENA.MIT.EDU, f_l@ATHENA.MIT.EDU,
Cc: jh@mit.edu
Reply-To: jh@mit.edu
Just like the rest of the campus, we got a power hit around noon. Our
machines have all been waiting for half an hour for the AFS servers to
come back up, probably like the rest of MIT. There has to be a
solution to this problem; if we're cut off from the campus net or if
the AFS servers die for any period of time and a machine reboots, it's
useless for the rest of that period of time. This is particularly a
problem with our NFS server, whose operation should never depend on
files not local to the machine anyway. Can the AFS cache scan have
timeouts or be backgrounded so the rest of the machine can come back
and function? Is there some other waywe can locally reconfigure the
machines not to hang in the boot until AFS decides to wake up?
Until this problem is permanently fixed, we have to request that
Operations people not leave the server (earth.mit.edu) in a state
where the boot will hang because of AFS outages after they logout.
Not knowing a lot about AFS, I realize that this could mean they can't
use AFS at all and will have to get their file access through some
other means (though some sort of cache flush in root's .logout would
do the trick, I think, if such exists). But right now we can't use
our machine at all and power hits have happened too often lately for
us to be able to afford hour-long midday outages.
--jh--
