Our original posting:
We are running OSF/1 v2.0 on a DEC Alpha (3000/300),
and have the following problems.
1. Segmentation fault while running 'rusers'
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Detail :
Only some users with no apparent common characteristics are able
to run the "rusers" program successfully. For others, it gives
segmentation fault after printing a few entries.
2. finger problem.
^^^^^^^^^^^^^^^
Detail :
When a user uses su or login to some other account, the finger
shows the new login name in place of the old login name.
But when he logs out, it still shows the new login name rather
than the old login name. This leads to some confusion and gives
wrong information about who has logged in the system currently.
Observation :
The entry is not properly reset, when a user exits from one session
in case of su or login from one session to the other.
3) NFS problem.
^^^^^^^^^^^^
We have mounted some remote filesystems through NFS on
our machine type soft (rw, nfs, intr)
The remote systems are SUNs connected to our ethernet-based LAN
through a bridge.
When we copy large files from local directories(DEC UNIX)
to a remote directory(SUNs), sometimes, the copy fails, saying:
NFS write failed : RPC timed out for <remote host>
cp: <local file>: connection timed out.
The same thing also happens for "tar".
Whereas, when we NFS mount directories from the Alpha(DEC Unix) to
the remote SUN machines and perform "tar", "cp" etc. on them at the
SUNs, this problem doesn't occur.
----------------------------------------------------------------------
We received the following responses:
Regarding problem (2) :-
Dan Winger <winger_at_sewp.nasa.gov>,
system PRIVILEGED account <root_at_borg.med.ecu.edu>,
"Cenon B.C. Marana Jr." <bonn_at_durian.usc.edu.ph> mention that
the problem existed in v2.0 but has been corrected in later versions.
Regarding problem (3), <alan_at_nabeth.cxo.dec.com> has the following
to say:
Soft mounts allow I/O requests to timeout. If the server can't
respond in time, the client will see an read or write fail,
with errno set to ETIMEDOUT. The client software can attempt
to retry the I/O, but I don't know of many applications that
do.
Using the Sun as the server it probably isn't keeping up with
the I/O load. The Alpha system is probably better able to
keep up. Sun may also have done more work to retry I/O in
selected applications.
The solution to the I/O timeout problem is to switch to hard
mounts. If you don't want I/Os to be permanently stuck when
a server is down, use "intr". This allows the I/O to be
interrupted.
Hellebo Knut <bgk1142_at_bggfu2.nho.hydro.com> says:
Try checking out the patches for Digital Unix 2.0. As for the NFS problem,
try rerunning 'nfssetup' on the Alpha and make sure you get things right.
R. Robins (extremely silly of me to have misplaced his response) opined
that perhaps running commands on the NFS mounted filesystem with
"nohup" might help. We haven't tried that out yet.
Regarding problem (1) :-
R. Robins suggested that the segment fault might have something to do with
user quotas , but we don't have any quotas enabled.
There were no other responses. Does this mean that nobody else is facing
the problem?
Also can somebody (notably, Hellebo Knut<bgk1142_at_bggfu2.nho.hydro.com>)
please mention some sites where patches for DEC UNIX 2.0 can be ftp'ed
from, and how are they to be applied?
Thanks to everybody who responded,
Chandan Kudige and Biswajit Sain.
Received on Mon Aug 21 1995 - 19:24:35 NZST