Traceroute question

[Fabrizio Maggioni_2]

Hi,

I have an application that need to reach its database server on the same network, but seems when route to database sides, it use not the correct interface (should be lan0 with server IP).

Traceroute inexplicably use the LAN that I use for my MC/SG heartbeat. It's clearly not going to get to the other server that way! (192.168.*.*)

Is there any way to force using a specific lan card?

Thanks!

[Steven Schweda]

What are all the IP addresses and routes
involved here?

[Fabrizio Maggioni_2]

Hi,

I am working on cluster environment,these are my address:

#Hosts
10.110.60.66 hydra
10.110.60.65 medusa

#Heartbeat
192.168.0.2 hydra_hb
192.168.0.1 medusa_hb

#Packages

10.110.10.50 anagrafe
10.110.10.51 contabile
10.110.10.52 amministrazione
10.110.10.53 civile

This is the simulation about our problem:

root@medusa /# traceroute 10.110.100.11
traceroute: Warning: Multiple interfaces found; using 10.110.10.51 @ lan0:1
traceroute to 10.110.100.11 (10.110.100.11), 30 hops max, 40 byte packets
1 10.110.100.11 (10.110.100.11) 0.507 ms 0.079 ms 0.073 ms
root@medusa /# netstat -in
Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll
lan0:1 1500 10.110.0.0 10.110.10.51 3 0 0 0 0
lan2 1500 192.168.0.0 192.168.0.1 26140 0 28592 0 0
lan1* 1500 none none 0 0 0 0 0
lan0:2 1500 10.110.0.0 10.110.10.52 39684 0 36722 0 0
lan0 1500 10.110.0.0 10.110.60.65 1222928 0 894210 0 0
lo0 4136 127.0.0.0 127.0.0.1 128316 0 128316 0 0
root@medusa /# ifconfig lan0
lan0: flags=1843
inet 10.110.60.65 netmask ffff0000 broadcast 10.110.255.255

It is using Package 's IP, and if the packages are down it use Heartbeat IP.I would like to force it using host IP, for example 10.110.60.65 .

Thanks.
Fabrizio

[Bill Hassell]

What is your routing table? Use:

netstat -rn

[Michael Steele_2]

Hi Fabrizio:

a) "...I have an application that need to reach its database server on the same network..."

Since you're using MC/SG you'll have to work at two levels using the 'cmmodnet' command and the 'ifalias' command.

Verify that

cmmodnet -a -i ip subnet, for example,
cmmodnet -a -i 15.13.169.16 15.13.143

...is the same order as /etc/rc.config.d/netconf.

Use 'netstat -in'.

b) Forcing traceeroute "...Is there any way to force using a specific lan card?...".

'traceroute -s ip.address'

-s source_addr
Use the following IP address (which must be given as an IP
number, not a hostname) as the source address in outgoing probe
packets. On hosts with more than one IP address, this option
can be used to force the source address to be something other
than the IP address of the interface the probe packet is sent
on. If the IP address is not one of this machine's interface
addresses, an error is returned and nothing is sent.

http://hpux.connect.org.uk/hppd/hpux/Networking/Admin/traceroute-991603/man.html

[Matti_Kurkela]

HP-UX traceroute won't really use the routing table at all. It uses some dumb algorithm to pick an interface to use. That algorithm often does the wrong thing when the server has multiple network interfaces.

Whenever using HP-UX traceroute on a server with multiple NICs, *always* use the -i option to explicitly select the correct interface to test.

A similar problem is discussed in this thread:
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1171990

MK

[Fabrizio Maggioni_2]

Hi all,

Now i am not to the customer, and i can't provide you out of command (for example netstat as suggested).
My focus is not to use traceroute but to set globally on the system using lan0 (primary IP) to "go out".

About cluster, it execute cmmodnet -a -i with virtual ip correctly during packages startup.

Thanks a Lot.
Fabrizio

[Franz F. Steininger · Accepted solution]

vou may try an "route add host "
count 1 if database-server is not on this host.

Franz