routing misery

[Dermot Beirne]

Hello all,
I have been having problems with routing on my server for a long time now. It is running HPUX 10.20. If a leased line goes down between our main and a remote site, all connectivity stops obviously. However, when the line comes back up, it almost always necessary to do a "route add" to a series of IP addresses on the remote site before thoses clients can reconnect to the box, even though the default gateway the box is using can see everything fine. This is a continuing problem, as the box is not see when original routes have been restored, and we usually have to tell it to query it's gateway using route add before it can ping them. Can anyone tell me where I might need to look to stop this behaviour. Thanks in advance.
Dermot.

[Paula J Frazer-Campbell]

Hi Dermot,

If I understand you problem correctly then if a leased line drops and restarts then the machines on the remote sites cannot see your server?

Are these unix machines?
Can you set up a ping routine that monitors from the remote site so that if the line drops and comes back then it runs a routing routine?


Paula

[Dermot Beirne]

Hi Paula,
Yes it is a unix machine running hpux 10.20. I understand that I can write a script to ping every so often and readd a route, but why do I need to do this, is there somewhere I can find out how routing works on HPUX, or what it cannot find a route back even though it's default gateway router has recovered ok.
Thanks.
Dermot

[Praveen Bezawada]

Hi
Routing tables can be configured dynamically,statically or a combination of both.
A static route is one we explictly add with a route command. In a stable network this is quite efficient.
But for case like yours, dynamic routing is required. Dynamic routing is performed by daemons on different hosts communication todiscover the topology of the network.
Generally routed and gated are the daemons that do the job. Please check if these are running on your machines.
Hope this helps.

...BPK...

[Paula J Frazer-Campbell]

Hi
Try these:-
Do a traceroute to the remote sites and identify all devices that you have control over - the problem may ie there.

Next time the line drops do a netstat -r -v and compare it with a a working netstat -r -v.

BTW - Are these remote sites in the netconf file?

Paula

[John Palmer · Accepted solution]

Hi Dermot,

The problem is probably due to ICMP redirects being issued by your routers.

Have a look at this previous thread...
http://forums.itrc.hp.com/cm/QuestionAnswer/1,1150,0x629f0559ff7cd4118fef0090279cd0f9,00.html

Regards,
John