by I administer several clusters of RS/6000's (all at 3.2.2.xx).
In each of the clusters, the info database is NFS mounted from
a cdrom filesystem on a server.
On all of the clients in one cluster, info (startinfo) spits
out the message:
"error writing display id to socket -1" when
when it starts up. After the error message, info goes ahead and
starts fine. I've done a lot of checking on /tmp and
/tmp/.info-help to make sure ownerships, etc are set correctly.
I've even stopped infod, deleted the socket, and restarted infod.
It doesn't make any difference whether I run info from the
console or display it remotely - the message is the same.
For the life of me, I can't pinpoint where/why this message
is being generated. Does anyone know the circumstances under
which startinfo generates this message? Something is out of
sorts somewhere.
I've never seen this in any of the other clusters.
Thanks.
Charlie
*************************************************************************
( Charlie McGuire | Tel. : (406) 243-4618 )
( Research Sys Mgr | Fax : (406) 243-4076 )
( The University of Montana | )
( SS409 | )
*************************************************************************