ping slow intermittent

[Printaporn_1]

Hi folk,

Following is senario
I have problem in hostA that ping HostC
# ping 10.196.16.158
PING 10.196.16.158: 64 byte packets
+++++ wait for > 2 mins ++++++++
64 bytes from 10.196.16.158: icmp_seq=0. time=17. ms
64 bytes from 10.196.16.158: icmp_seq=1. time=19. ms
64 bytes from 10.196.16.158: icmp_seq=2. time=50. ms

question is what happen in ping why it slow before get response.

hostA can ping HostB and HostB can ping HostC
without slowly.

all host is in same segment.
also I notice that when HostB ping HostC
ping that hang at HostA will get response.
what can be the cause or what else information have to check ?
no duplicate IP problem and problem is intermittent sometime it not slow but most of the ping will got this problem , telnet also slow.
Thanks in advance,

[Volker Borowski]

Hello,

due to the time (>10ms) I suspect a WAN line with a router to be involved.

I think there need to be an adjustment on that router, because it seems to do no dial for packets of your hostA. For the other hostB it does a dial, and after the line is opened, you first hostA can use it as well.

If the router is a cisco, you should check the WAN interface configuration for the
dialer-group command.
Cisco-Routers can decide for a
"permitted packet" and
"interesting packet".

A "permitted packet" is transmitted, if the line is up.
An "interesting packet" will cause a WAN interface to execute a dial-command and after this the line is up.

Hope this helps.
Volker

[Printaporn_1]

Hi,

thanks but for
More information , in this configuration it's not WAN and all host is in same LAN segment.

[Vincenzo Restuccia]

traceroute 10.196.16.158

[Volker Borowski]

To get more information,
in addition to vincenzos question,
I would like to see the output of

netstat -nr
ifconfig lan0
ifconfig lan1
...

of hostA B and C
Volker

[Printaporn_1]

that is

# traceroute 10.196.16.158
traceroute to 10.196.16.158 (10.196.16.158), 30 hops max, 20 byte packets
1 ict-server (10.196.16.158) 1 ms 1 ms 1 ms
# netstat -nr
Routing tables
Destination Gateway Flags Refs Use Interface Pmtu PmtuTime
10.3.112.1 127.0.0.1 UH 0 2188 lo0 4608
10.196.16.189 127.0.0.1 UH 44 1510701 lo0 4608
127.0.0.1 127.0.0.1 UH 0 324 lo0 4608
default 10.196.16.1 UG 1 36887 lan0 1497
10.3.112.0 10.3.112.1 U 2 456809 lan1 1500
10.23 10.196.16.9 UG 0 0 lan0 1497
10.196.16.0 10.196.16.189 U 8 620234 lan0 1497
32.76 10.196.16.9 UG 3 374325 lan0 1497
# ifconfig lan0
lan0: flags=863
inet 10.196.16.189 netmask fffffc00 broadcast 10.196.19.255
# ifconfig lan1
lan1: flags=863
inet 10.3.112.1 netmask fffff800 broadcast 10.3.119.255

-------
I will very appreciate on help.

[Printaporn_1]

sorry above information is for hostA that got problem , following is information on HostB.

I don't have information for HostC

A --> C slow
A ---> B fast
B ----> C fast

# netstat -rn
Routing tables
Destination Gateway Flags Refs Use Interface Pmtu PmtuTime
10.196.16.185 127.0.0.1 UH 72 676509 lo0 4608
15.3.112.1 127.0.0.1 UH 0 108 lo0 4608
127.0.0.1 127.0.0.1 UH 0 514 lo0 4608
default 10.196.16.1 UG 2 363438 lan0 1497
10.23 10.196.16.9 UG 0 0 lan0 1497
10.196.16.0 10.196.16.185 U 34 1150919 lan0 1497
15.3.112.0 15.3.112.1 U 2 514037 lan1 1500
32.76 10.196.16.9 UG 4 213521 lan0 1497
# ifconfig lan0
lan0: flags=863
inet 10.196.16.185 netmask fffffc00 broadcast 10.196.19.255
# ifconfig lan1
lan1: flags=863
inet 15.3.112.1 netmask fffff800 broadcast 15.3.119.255

---------

one diffrent on both host is lan1

thanks again.

[Volker Borowski]

Hmm, this is really strange.

Routing table is ok
10.196.16.0 10.196.16.189 U
sees the target to be directly connected.

Please check this on the target site as well.

I did not notice at first, that you do not loose a packet. ICMP-SEQNR is 0 for the first packet you receive after the 2 minutes.

This means, it is NOT slow ! It is not send at all(!) before this first packet....

# ifconfig lan0
lan0: flags=863
inet 10.196.16.189 netmask fffffc00 broadcast 10.196.19.255

Check the coresponding interface of the other host. Does the netmask match fffffc00 there as well ?

In this setup this is supposed to work OK.

How is this segment made physically ? Switch ? Hub ? Did you try to change the port of Switch/HUB ? Can you access the Switch/Hub to see, if there are error-logs on it. Can you check half/full duplex connection to the switch/hub ?

No real smashing idea any more.
Volker

[Juan González]

Hi,
mmmm, could you include too the information about HostC?

Best Regards
JGM

[Printaporn_1]

Hi,

Thanks all , espically Volker that confirm that configuration should work fine.

subnetmask in HostC is correct.
I will proceed with Hardware checking.

thanks a lot,

[Brian Hackley]

Hi,

We can all keep guessing as to what is going on, but what else I can think of is:
1. Host C is sending ICMP Source Quench (HPUX 11.0). Disable on host c with ndd -set /dev/ip ip_send_source_quench 0 (default is 1). To make permanent edit /etc/rc.config.d/nddconf

2. You should get a concurrent nettl trace at both ends of the connection. See ITRC Knowledge Base document KBRC00001357 or KBRC00001272 for more info on collecting a nettl trace.

I hope this helps you pinpoint the root cause of the problem.

Regards,

-> Brian Hackley