Operating System - HP-UX
1820201 Members
3791 Online
109620 Solutions
New Discussion юеВ

Periodic traceroute problems

 

Periodic traceroute problems

I am having a strange problem on my private network. Periodicaly we seem to lose connection between the client and the server using Oracle. I can ping my client but traceroute runs and never connects. I can reset the card on the client side and then everything is good again. There are 4 servers connected to the private network. The first 3 have worked fine but this next server seems to have issues.

One more note was when we tried NFS to the client it would work fine for a while then start with the NFS server "IP" not responding ... error. Eventually it would come back or I could speed it along by reseting the card with landiag.
5 REPLIES 5
Bill Hassell
Honored Contributor

Re: Periodic traceroute problems

Can you ping your router? Your network admin may have turned off ping responses and HP-UX has a default setting to disable any connection where the router cannot be pinged every few minutes.

To see if the setting is turned on or off, use:

ndd -get /dev/ip ip_ire_gw_probe

1 means the detection feature is enabled. To turn off this feature:

ndd -set /dev/ip ip_ire_gw_probe 0

This is a temporary change. To make it permanent, edit your /etc/rc.config.d/nddconf file to add:

TRANSPORT_NAME[0]=ip
NDD_NAME[0]=ip_ire_gw_probe
NDD_VALUE[0]=0

Be sure to substitute the next available array element for [0] in case [0] is already in use.

Note also that for 11.00, early versions of ndd had a bug that prevented the changes in nddconf from taking effect. Get the latest ndd patch.


Bill Hassell, sysadmin
Tor-Arne Nostdal
Trusted Contributor

Re: Periodic traceroute problems

Make sure that you have same link settings on NIC as on switch.
If you use Auto on NIC, then use the same on switch port.
If you use fixed speed/duplex on server... do the same on the switch port.

/Tor-Arne
I'm trying to become President of the state I'm in...

Re: Periodic traceroute problems

It turns out we had a duplicate IP address! Some nice person added IP adresses to the "dumb" hubs.

I do have a syntax question about the command:

ndd -set /dev/ip ip_ire_gw_probe 0

If my card IP is 25.25.25.4 should it read:

ndd -set 25.25.25.4 ip_ire_gw_probe 0?

I get back:

*ERROR* module name length (10) > FMNAMESZ (8)
ndd: couldn't push module '25.25.25.4', Invalid argument

I suspect I don't quite get the command you suggested.

Bill Hassell
Honored Contributor

Re: Periodic traceroute problems

> If my card IP is 25.25.25.4 should it read:

> ndd -set 25.25.25.4 ip_ire_gw_probe 0?

> I get back:

> *ERROR* module name length (10) > FMNAMESZ (8)
ndd: couldn't push module '25.25.25.4', Invalid argument

No. The /dev/ip has nothing to do with any IP address. This is a system-wide setting and affects all LAN cards. The correct syntax is exactly as shown...just copy and paste. The error you saw is because ndd is trying to use the numbers as a device file.


Bill Hassell, sysadmin

Re: Periodic traceroute problems

Thanks. Everything is working fine.