SUMMARY: Netstat causes telnetd to lose the plot

From: John Waterson, ITS, SENM (john.waterson@scotent.co.uk)
Date: Wed Oct 14 1998 - 13:52:20 CDT


Original question is below. Many thanks to the following people for
their speedy replies:

Bill Crane [bcrane@weather.com]
Frank Smith [franks@tradeonemktg-aus.com]
Bruce R. Zimmer [bzimmer@all-phase.com]

The problem is caused by the fact that our DNS server was (and actually
still is) a bit broken, and wasn't doing reverse lookups properly.
netstat -rn doesn't attempt to resolve hostnames and ran correctly,
which led to the heart of the matter.

The post machine which was still able to telnet in successfully is
actually the broken DNS server, which probably explains (in some strange
way) why it could get through. The Subnet thing was most likely a red
herring.

Anyway, easily sorted as I don't actually need DNS on this machine just
now. Therefore I just renamed /etc/resolv.conf and rebooted. I'll sort
the wider DNS problem out later.

Many thanks.

John

---
john.waterson@scotent.co.uk

> Original problem: > > This is a very strange problem, and I'm only half expecting an answer. > The short description of the problem is this: netstat -r > produces a few > lines of output and then hangs, and from that point onwards telnetd > stops responding to connections from most (but not all) hosts. > > More detail: > > Here is the output from netstat -r: > > # netstat -r > > Routing Table: > Destination Gateway Flags Ref Use Interface > -------------------- -------------------- ----- ----- ------ --------- > localhost localhost UH 0 0 lo0 > arpanet news U 3 126 le0 > > ...and that's all she wrote - I never get my prompt back again. I can > kill the netstat from another window and my shell comes back without > problems, but now I can't log in again over telnet from my (Win95) > workstation. Telnet connects okay, but the login prompt never appears. > The telnet windows I already have open continue to function normally. > > Telnetd still responds to some hosts. From the machine's > console, I can > successfully telnet to either 127.0.0.1 or the normal IP > address of the > Ethernet card. I also have one other Sparc (called post) which is able > to telnet in without problems, but all the other Unixen and Windows > machines behave the same way as my PC. It may be significant that the > "post" machine is the only one on a different Subnet from the affected > server. > > All of this under Solaris 2.5.1 on a Sparc 10. I have applied patch > numbers 103630-10 and 103582-16 and rebooted, but without any success. > Every time I run netstat -r again, I get the same problem. Probably > worth noting that the machine also runs a couple of HTTP servers, and > these are unaffected by this problem. > > Addendum - still working on this and just noticed that arp -a > now hangs > after printing the first two header lines of output. I assume > that this > is also related.... > > Any advice on this gratefully received. Will summarise. > > John > --- > john.waterson@scotent.co.uk



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:50 CDT