In <[log in to unmask]> [log in to unmask] writes:

> we're working on bringing up a new application using m.b. foster's
> odbc software.  recently, we started getting netipc 116 errors.
> this has got both the rc and m.b. foster stumped.  116 means the
> 3000 can't reach/find a given ip address/pc to send information
> (back to the requestor).  i'm wondering if anyone else has had
> trouble with this particular error?  just so you'll know, hp has
> opened an sr for this problem because is causes the listener job
> to abort and (we think) it hoses arp cache - so that means stopping
> and restarting the network <- not a good thing.  also - please
> don't take this message as a complaint towards m.b. foster!  we've
> been real pleased so far with the company and the software.  - d

We've seen it many times. From the NetIPC manual:

116          MESSAGE      DESTINATION UNREACHABLE.

             CAUSE        The transport was unable to route the packet to the
                          destination.

             ACTION       Notify your operator.

As Bruce mentioned, it usually means that you're (your machine) is trying
to connect to an IP address that the network software doesn't know how to
get to (i.e. if you don't have a default gateway in NMMGR and you try to
connect to a network that's not listed under a specific gateway in NMMGR).
I believe it can also occur if your router knows a route is down or
unavailable.

I've also seen it occasionally for machines on the local network; usually
another 3000 which was down (or it's network was down) at the time. I suspect
this might happen because the MAC address is in cache on the (source) 3000
-so it knows it HAS talked to the host before- but suddenly can't reach it.

It's a normal NetIPC error though and the application should recover normally
from it. Our e-mail system knows, for instance, when it hits a 116 when trying
to send mail somewhere, to just give up and try again later.

*HOWEVER* This normally occurs on an outbound connection request (i.e. your
system is trying to connect to some other system and can't get through). If
the process getting this error is a *listening* socket, then that means that
you have a network-config or router problem. This happens when an inbound
connection-request packet can get TO the 3000, but the 3000 doesn't know how
to route the return packet. Either you don't have a default gateway (or a
specific gateway) for that network in NMMGR, or your router(s) is having
trouble. [It would only be a router if you have more than one on your network,
such that inbound packets might come in via one router and the 3000 only knows
about another router(s) --- and those router(s) aren't talking to the other
router and exchanging route info.]

                    -Chris Bartram