[prev in list] [next in list] [prev in thread] [next in thread] 

List:       hpux-admin
Subject:    [HPADM] vuelogin, DNS and hung logins
From:       bdmcclu () sandia ! gov
Date:       1995-06-28 13:15:03
[Download RAW message or body]

We are running a number of HP 400's and 700's running 9.X.  Most of
our machines run vuelogin as well as DNS.

Periodically, our connection to the subnets of our parent domain ( and
external connection ) get flaky, and during these times logins via 
vuelogin hang and eventually fail because of this.

I've seen a couple solutions to the problem posted:

1. Set up a local secondary nameserver - This doesn't work.  We're doing it
	and it and logins still hang when the connection to the external network
	goes down.  

2. set up nsswitch.conf. - Yes, this helps and some of us have done it, but
	I think its only a workaround for something that I think is misconfigured
	in vuelogin

We've also found that using the "search" directive in our resolv.conf
instead of "domain" takes care of the hanging as well.  However, it appears
what really should be fixed is that vuelogin when looking up the hostname
for the machine its running on should not be applying the DNS search path
to the hostname.

This is the debugging output from named on a local nameserver that is 
authoritative for "ms.sandia.gov".  The output occurred while I was logging
into the machine "scooby" (132.175.27.105) using vuelogin:

-------------------------
Debug turned ON, Level 1

datagram from 132.175.27.105 port 3455, fd 6, len 38
req: nlookup(scooby.ms.sandia.gov) id 1 type=1
req: found 'scooby.ms.sandia.gov' as 'scooby.ms.sandia.gov' (cname=0)
req: answer -> 132.175.27.105 6 (3455) id=1 Local

datagram from 132.175.27.105 port 3460, fd 6, len 38
req: nlookup(scooby.ms.sandia.gov) id 1 type=1
req: found 'scooby.ms.sandia.gov' as 'scooby.ms.sandia.gov' (cname=0)
req: answer -> 132.175.27.105 6 (3460) id=1 Local

datagram from 132.175.27.105 port 3461, fd 6, len 52
req: nlookup(scooby.ms.sandia.gov.ms.sandia.gov) id 2 type=1
req: found 'scooby.ms.sandia.gov.ms.sandia.gov' as 'ms.sandia.gov' (cname=0)
req: answer -> 132.175.27.105 6 (3461) id=2 Local

datagram from 132.175.27.105 port 3462, fd 6, len 49
req: nlookup(scooby.ms.sandia.gov.sandia.gov) id 3 type=1
req: found 'scooby.ms.sandia.gov.sandia.gov' as 'sandia.gov' (cname=0)
forw: forw -> 132.175.109.2 7 (53) nsid=9 id=3 0ms retry 4 sec
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

datagram from 132.175.109.2 port 53, fd 6, len 113
send_msg -> 132.175.27.105 (UDP 6 3462) id=3

datagram from 132.175.27.105 port 3463, fd 6, len 38
req: nlookup(scooby.ms.sandia.gov) id 4 type=1
req: found 'scooby.ms.sandia.gov' as 'scooby.ms.sandia.gov' (cname=0)
req: answer -> 132.175.27.105 6 (3463) id=4 Local
 ...
[more of the same deleted]
-------------------------

Note that even though this nameserver is on our local network, the
query for "scooby.ms.sandia.gov.sandia.gov" gets sent off to our
forwarder.  So if connectivity to that network is down, then 
the queries by vuelogin for its own hostname end up hanging anyway.

My gripe is that the DNS search path shouldn't be applied when the
hostname is already fully qualified anyway.  As I said, we can alter
this behavior by using "search ms.sandia.gov" in resolv.conf, but 
is there another way of getting vuelogin to not apply the search path?

thanks, Brent


--
             ---> Please post QUESTIONS and SUMMARIES only!! <---
        To subscribe/unsubscribe to this list, contact majordomo@cv.ruu.nl
  Name: hpux-admin@cv.ruu.nl                 Owner: owner-hpux-admin@cv.ruu.nl
 Archive: (FTP, browse only):     ftp.cv.ruu.nl/pub/digests/hpux-admin 
          (Web, browse & search): http://hpux.ced.tudelft.nl/HPUX_ADMIN_ARCHIVE

[prev in list] [next in list] [prev in thread] [next in thread]