Slow response telnet-ssh-ping

[R.O.]

I have restored a test server from ignite tape and now something strange happens with ping and telnet-ssh; they takes long time to start working. For the ping command:

test:/#timex ping localhost
PING localhost: 64 byte packets
64 bytes from 127.0.0.1: icmp_seq=0. time=0. ms

----localhost PING Statistics----
1 packets transmitted, 1 packets received, 0% packet loss
round-trip (ms) min/avg/max = 0/0/0

real 15.97
user 0.03
sys 0.02

It takes 15 seconds to start showing the first ping line and after that, it works right and show no delays. The same happens with telnet-ssh; they take about 20 seconds to show the prompt and the same time once I have entered the password.

I have looked for name resolution issues, but there is no problems:

test:/#timex nslookup localhost
Using /etc/hosts on: test

looking up FILES
Name: localhost
Address: 127.0.0.1
Aliases: loopback

real 0.04
user 0.01
sys 0.01


test:/#timex nslookup www.google.com
Using /etc/hosts on: test

looking up FILES
Trying DNS
Non-authoritative answer:
Name: www.l.google.com
Addresses: 209.85.229.104, 209.85.229.147, 209.85.229.99
Aliases: www.google.com


real 0.02
user 0.01
sys 0.01

test:/#

What could be happening?

Regards,

[Rita C Workman]

Do a traceroute...it is possible something on your network is tossing it around.

Also, confirm your gateway address because you may be sending it off so it is traveling more than it should to start with.

If you see you are hopping more than you should around routers (internally), then discuss this with your network person.

Just a thought,
Rita

[Steven Schweda]

> I have looked for name resolution issues,
> but there is no problems:
> [...]

Look harder? A Forum search for keywords
like, say:
slow telnet
should find many old threads which describe
the usual DNS problem in such cases, namely
that the server system can't look up the IP
address of the client to determine the
client's name. "nslookup " is not
where you see this problem, it's with
"nslookup " (on the server).

What's in your "resolv.conf" file? "hosts"?

[R.O.]

Hi,

I have solved the problem. In console I saw lots of messages saying that "unable to bind to NIS server" (sth like that) and I have killed the ypbind process and changed the "NIS CLIENT" entry to 0 in "/etc/rc.config.d/namesrvs". Now there is no delay in ping-telnet-ssh. Many thanks for your effort !!

Regards,

[R.O.]

See above post for resolution.

[Matti_Kurkela]

The wait of 15-20 seconds sure sounds like a DNS query timeout.

You didn't mention your HP-UX version, but if it's 11.23 or 11.31, it might be because telnet, ssh and ping are all IPv6-aware in 11.23 and above.

That means they'll be using IPv6-compatible name resolution routines, which are configured in /etc/nsswitch.conf using the "ipnodes" line, not the traditional "hosts" line.

The "ipnodes" line also has a factory default of "DNS first, then /etc/hosts".

If the nslookup command is IPv4-only, it uses the "hosts" line in /etc/nsswitch.conf (or the factory default of looking in /etc/hosts first), so it cannot detect the problem.

The current recommendation is that every DNS server should map "localhost" to 127.0.0.1 and vice versa, but not all DNS admins follow that rule.

My suggestion would be: check your /etc/nsswitch.conf. If it does not have an "ipnodes: ..." line, make a copy of your "hosts: ..." line and change the copy to "ipnodes". If you're using 11.23, you can also change it using SAM (Networking and Communications -> Name Service Switch). I don't remember how to find it in 11.31's GUI tools, sorry.

MK

[Steven Schweda]

> [...] /etc/nsswitch.conf [...]

Oops. That the one I meant. (Brain ... old.)

> [...] "unable to bind to NIS server" [...]

Revealing the use of NIS is yet another
reason for showing what's in that file.