Operating System - HP-UX
1748156 Members
4237 Online
108758 Solutions
New Discussion

Re: Different subnet's IPs are routing server's own IP

 
Taurian
Valued Contributor

Different subnet's IPs are routing server's own IP

 

Hi All,

 

I have a strange problem in my HP-UX 11.31 box. When I use traceroute command for an IP of different subnet than it's own, the trace ends up on it's own IP:

 

# traceroute 172.20.11.1
      traceroute to 172.20.11.1 (172.20.11.1), 30 hops max, 40 byte packets
      1  agbdb (172.22.25.11)  0.077 ms !N  0.029 ms !N  0.011 ms !N

 

 

 

I have tried deleting the IP and resetting it after restart. The gateway setting is also present in the server. Default route is also present in the routing table:

 

     default        0.0.0.0           172.22.25.1    Yes              lan0   

 

 

All these same settings are working fine on other servers. But traffic on this server ends up on it's own IP.

 

Why is the traffic not moving out from the it's network card? What can be the issue?

 

 

Regards,

Taurian.

6 REPLIES 6
paulgear
Esteemed Contributor

Re: Different subnet's IPs are routing server's own IP

You're probably better off asking this in the HP-UX forum instead of the HP Networking forums, but if you post the output of 'netstat -in' and 'netstat -rn', we can probably point you in the right direction.

Regards,
Paul
Taurian
Valued Contributor

Re: Different subnet's IPs are routing server's own IP

 

Hi Paul,

 

Thanks for your reply.

 

Following is the output from a server that is NOT working:

 

# netstat -in
Name      Mtu  Network         Address         Ipkts              Ierrs Opkts              Oerrs Coll
lo0      32808 127.0.0.0       127.0.0.1       12305              0     12305              0     0  
lan0      1500 172.22.25.0     172.22.25.11    118016             0     11609              0     0

 

# netstat -rn
Routing tables
Destination           Gateway            Flags Refs Interface  Pmtu
127.0.0.1             127.0.0.1          UH    0    lo0       32808
172.22.25.11          172.22.25.11       UH    0    lan0      32808
172.22.25.0           172.22.25.11       U     2    lan0       1500
127.0.0.0             127.0.0.1          U     0    lo0       32808
default               172.22.25.1        UG    0    lan0       1500

 

 

 

 

 

 

 

 

Following is the output from a WORKING server:

 

# netstat -in
Name      Mtu  Network         Address         Ipkts              Ierrs Opkts              Oerrs Coll
lan0      1500 172.22.25.0     172.22.25.19    378832             0     278590             0     0  
lo0      32808 127.0.0.0       127.0.0.1       17816              0     17816              0     0  
lan0:131  1500 172.20.24.0     172.20.24.23    120                0     0                  0     0  
Warning: The above name 'lan0:1316887' is truncated, use -w to show the output in wide format


# netstat -rn
Routing tables
Destination           Gateway            Flags Refs Interface  Pmtu
127.0.0.1             127.0.0.1          UH    0    lo0       32808
172.20.24.23          172.20.24.23       UH    0    lan0:1316887 32808
172.22.25.19          172.22.25.19       UH    0    lan0      32808
172.20.24.0           172.20.24.23       U     3    lan0:1316887  1500
172.22.25.0           172.22.25.19       U     3    lan0       1500
127.0.0.0             127.0.0.1          U     0    lo0       32808
default               172.22.25.1        UG    0    lan0       1500

 

 

 

 

 

Regards,

Taurian.

paulgear
Esteemed Contributor

Re: Different subnet's IPs are routing server's own IP

I'm not exactly sure about how HP-UX responds to ICMP redirects or connection refusals, but my guess is that one or other of those is preventing your host from getting through.  What does the traceroute output look like on both servers?

Regards,
Paul
Taurian
Valued Contributor

Re: Different subnet's IPs are routing server's own IP

 

Hi Paul,

 

Traceroute from NOT working server:

 

# traceroute 172.20.11.1
traceroute to 172.20.11.1 (172.20.11.1), 30 hops max, 40 byte packets
1 agbdb (172.22.25.11) 0.077 ms !N 0.029 ms !N 0.011 ms !N

 

 

 

 

Traceroute from WORKING server:

 

# traceroute 172.20.11.1   
traceroute to 172.20.11.1 (172.20.11.1), 30 hops max, 40 byte packets
 1  172.22.0.5 (172.22.0.5)  1.220 ms  0.711 ms  0.707 ms
 2  172.22.5.242 (172.22.5.242)  0.789 ms  0.736 ms  0.726 ms
 3  172.22.0.9 (172.22.0.9)  8.847 ms  9.829 ms  9.918 ms
 4  8.8.8.2 (8.8.8.2)  44.848 ms  45.020 ms  44.818 ms

 

 

 

 

The thing is that I have already reinstalled HP-UX from DVD in exactly the same manner as I did the working servers. The issue still exists. This eliminates any issues at HP-UX's end.

 

 

Reagrds,

Taurian.

paulgear
Esteemed Contributor

Re: Different subnet's IPs are routing server's own IP

Those traceroutes are hitting a different router for the same destination.  I've just realised that your netstat -rn output did not include any netmasks.  If you can find the flag on HP-UX which shows the mask for each route and repost the netstat -rn output, we should be able to pinpoint what's happening.  Do the HP-UX boxes run any dynamic routing daemons such as ripd or quagga?

Regards,
Paul
Matti_Kurkela
Honored Contributor

Re: Different subnet's IPs are routing server's own IP

> 1 agbdb (172.22.25.11) 0.077 ms !N 0.029 ms !N 0.011 ms !N

 

The !N in the traceroute output means that a "network unreachable" error was generated.


Does the gateway 172.22.25.1 answer to pings? If not, have you disabled the Dead Gateway Detection function of HP-UX?

 

By default, HP-UX tends to expect that gateways should answer to pings, and if they don't, HP-UX assumes the gateway has failed and marks the gateway entry in the routing table as disabled. That would explain the "network unreachable" error generated at the sending host itself: the host thinks it has no valid gateway for sending to the specified destination, so it has no choice but to generate error responses.

 

To disable Dead Gateway Detection, you'll need to change a ndd setting:

ndd -set /dev/ip ip_ire_gw_probe 0

 To make the setting persistent, edit /etc/rc.config.d/nddconf.

 

After switching off Dead Gateway Detection, you may need to delete the default gateway entry from the routing table and add it back to re-enable it. (Unfortunately the "netstat -rnv" output does not indicate whether any gateways listed are disabled by DGD or not.)

 

> If you can find the flag on HP-UX which shows the mask for each route and repost the netstat -rn output,

 

The command for that is "netstat -rnv".

MK