Ping losts network connexion

[MR VILLOT MR MONTAGNE]

Hi,

I ping one machine and I get sometimes some trouble (disconnection !):
Délai d'attente de la demande dépassé.
Délai d'attente de la demande dépassé.
Délai d'attente de la demande dépassé.
Réponse de 43.XXX.XXX.XXX : octets=32 temps<10 ms TTL=255
Réponse de 43.XXX.XXX.XXX : octets=32 temps<10 ms TTL=255
Réponse de 43.XXX.XXX.XXX : octets=32 temps<10 ms TTL=255

I think it could be hardware problem with network card but I would like to have some advices.
Server : HP-UX 10.20
#ioscan -fnC lan
Class I H/W Path Driver S/W State H/W Type Description
========================================================================
lan 0 2/0/2 lan2 CLAIMED INTERFACE Built-in LAN
/dev/diag/lan0 /dev/
#lanscan
Hardware Station Crd Hardware Net-Interface NM MAC HP DLPI Mjr
Path Address In# State NameUnit State ID Type Support Num
2/0/2 0x080009D2A66A 0 UP lan0 UP 4 ETHER Yes 52
#[alse02:/] ifconfig lan0
lan0: flags=863
#lanadmin :
Network Management ID = 4
Description = lan0 Hewlett-Packard LAN Interface Hw Rev 0
Type (value) = ethernet-csmacd(6)
MTU Size = 1500
Speed = 10000000
Station Address = 0x80009d2a66a
Administration Status (value) = up(1)
Operation Status (value) = up(1)
Last Change = 2688
Inbound Octets = 495072723
Inbound Unicast Packets = 438058
Inbound Non-Unicast Packets = 2903330
Inbound Discards = 0
Inbound Errors = 0
Inbound Unknown Protocols = 1320490
Outbound Octets = 2048451346
Outbound Unicast Packets = 1595618
Outbound Non-Unicast Packets = 52
Outbound Discards = 0
Outbound Errors = 0
Outbound Queue Length = 0
Specific = 655367

Ethernet-like Statistics Group

Index = 4
Alignment Errors = 0
FCS Errors = 0
Single Collision Frames = 834
Multiple Collision Frames = 964
Deferred Transmissions = 234349
Late Collisions = 0
Excessive Collisions = 0
Internal MAC Transmit Errors = 0
Carrier Sense Errors = 0
Frames Too Long = 0
Internal MAC Receive Errors = 0

Statistics were created during normal behaviour.

what /stand/vmunix | grep btlan
btlan3.c Core 100BT 10.20-PHNE_13675 $Revision: 1.71 $ $Date: 98/01/05 15:18:53 $

I saw that patch must be : PHNE_22405
Speed Card : 10000000
Switch is set up to 100 Mb.

The problem must be speed switch (Must i set up port to 10 Mb) or do you think that it is a harw

[RAC_1]

As you may be aware 10.20 is absolute now.

Things to check.

1. Speed settings on NIC and switch/router.
They should be same.

2. NDD parameter - ip_ire_gw_probe
ndd -h ip_ir_gw_probe. If set to 1, disable it. Set it to 0
nettune -get /dev/ip ip_ire_gw_probe
nettune -set /dev/ip ip_ire_gw_probe 0

Anil

[Massimo Bianchi]

Hi,
with so many collisions first i would look at the switch, as you said, check network cable and port, and fix speed.

Single Collision Frames = 834
Multiple Collision Frames = 964
Deferred Transmissions = 234349

Too many collisions...

Regards,
Massimo

[Ron Kinner]

I see no obvious problems with the NIC. Your collision rate is quite acceptable (834/2048451346 <<.1%) so we can rule out a duplex mismatch. The builtin NIC on the old 10.20's was always limited to 10 half anyway and lanadmin confirms that this is a 10Meg circuit with no mention of duplex so it has to be a half. Your switch should not be set to 100M if you want it to work at all. Lanadmins says:
Speed = 10000000 which if we add commas is:
Speed = 10,000,000 or 10 MEG! Set your switch to 10 Half or AutoNegotiate and it should work OK. I've never seen one that could talk with one end set to 10 and the other 100 so I doubt that's your problem but just in case please verify that the switch port is set to 10 Meg Half Duplex.

If this gap in connectivity always happens at the beginning of a connection and doesn't happen in the middle then it's probably just a delay in getting the MAC address with an ARP request. Possibly you are using proxy arp instead of a default route.

Is the device we are pinging on the same LAN or is it over a router? If over a router you could have a dirty serial link or a full queue dropping packets (in either direction).

If on the LAN then check the connections of the device you are pinging. Perhaps it is the guilty party. You can try linkloop (see the man) to verify that you have a good layer 2 connection. It doesn't always work with non HP stuff but I believe it was part of 10.2.

Ron

[Jeff Schussele]

Hi,

Some routers can be set up to discard ICMP packets under times of heavy load. You should check with the Net Ops people to check the router logs to see if this might be occurring in your case.

Rgds,
Jeff