[26331] in Athena Bugs

home help back first fref pref prev next nref lref last post

Slow AFS timeouts when Linux machine is booted without network

daemon@ATHENA.MIT.EDU (Greg Hudson)
Tue Jan 25 16:26:33 2005

Date: Tue, 25 Jan 2005 16:25:59 -0500
Message-Id: <200501252125.j0PLPx8X025572@equal-rites.mit.edu>
From: Greg Hudson <ghudson@mit.edu>
To: bugs@mit.edu
Errors-To: bugs-bounces@mit.edu

If a 9.3.18 Linux machine is booted with no network interfaces
configured, afsd spits out the following message:

  # /usr/vice/etc/afsd -stat 10000 -daemosn 4 -volumes 200 -afsdb -dynroot
  ADVISEADDR: Error in specifying interface addresses:No existing IP interfaces found

If AFS is started in this state, every AFS access to the athena cell
pauses for 27 seconds (generating a "lost contact" message for each
VLDB server) and reponds with a message like:

  # ls /afs/athena/project
  ls: /afs/athena/project: No such device

Note that different behavior is seen if the cell has been successfully
accessed before the network interface goes away.  In that case, the
first access to the cell takes the same 27 seconds to fail (yielding
"Connection timed out") and subsequent accesses fail immediately
(yielding "No such file or directory") until network is restored and
the servers are re-checked with "fs checks -a".

It is unlikely that Athena machines will be meaningfully bootable off
network until this problem is addressed.  Even the 27 seconds to time
out for the first access to a cell is pretty unfriendly.

home help back first fref pref prev next nref lref last post