20 seconds to login at console?? (ds3100)

[Mike Iglesias]

In article <1213 at psych.stanford.edu> connie at psych.UUCP (Constance Stillinger) writes:
>
>Much to my surprise, this nasty behavior stopped when my local
>mail gateway/nameserver/etc. came back up and started behaving
>itself again (after a recent thunderstorm).
>
>I didn't think this could be the source of login problems,
>especially at the console, but it seems like more than a
>coincindence.
>
>Any comments from people more knowledgeable than I?  What kinds
>of flaky behavior can I expect from my workstation when my gateways and nameservers
>freak out?

It all depends on what happens when you log in.  If your system has to
do any hostname<->ip address translations, it's going to try using the
first nameserver in your /etc/resolv.conf file.  If it doesn't get
an answer from that one, it will try any others (if more than one is
listed) in order.  If you have your system setup via /etc/svcorder
(/etc/svc.conf in Ultrix 4.0) to use BIND first and then LOCAL, it
will look in your /etc/hosts file when it can't contact a nameserver.
If your nameservers are on the other side of a gateway and the gateway
is down, it will take a while to resolve the hostnames (it usually tries
for about 5-10 seconds per nameserver).

If you're mounting partitions from a fileserver on the other side of
the gateway, then you'll get NFS server not responding messages, and
depending on whether the partition is mounted hard or soft it will
keep trying (hard) or give up eventually (soft).

You could change the order that you look for hostnames to local, bind
so that it looks in /etc/hosts first.  The danger there is that if one
of your "necessary" hosts changes it's IP address and you don't change
the hosts file, you'll have problems contacting that host.


Mike Iglesias
University of California, Irvine
Internet:    iglesias at orion.oac.uci.edu
BITNET:      iglesias at uci
uucp:        ...!ucbvax!ucivax!iglesias