TCPIP and 2 Interfaces

I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2. The QNX
networking works perfectly and is using both interfaces as far as I can tell
from netinfo -l.

I installed TCPIP on both machines and they can ping themselves perfectly
but cannot ping each other. The setup is dead simple, no routing involved at
all, IP number range and subnet masks are the same, everything I can think
of is setup correctly.

Has anyone run across this before? I’m guessing it has something to do with
the dual ethernet interfaces but who knows…

Thanks

Sheldon Parkes
sheldon@onlinedata.com

Sheldon Parkes <sheldon@onlinedata.com> wrote:

I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2. The QNX
networking works perfectly and is using both interfaces as far as I can tell
from netinfo -l.

I installed TCPIP on both machines and they can ping themselves perfectly
but cannot ping each other. The setup is dead simple, no routing involved at
all, IP number range and subnet masks are the same, everything I can think
of is setup correctly.

Has anyone run across this before? I’m guessing it has something to do with
the dual ethernet interfaces but who knows…

One thing that has bitten us is you have to wait with starting Sock(l)et until
the Net.anything have initialised and registered with Net. Otherwise Socket
wont’t find an interface.

Thanks

Sheldon Parkes
sheldon@onlinedata.com

\

Socket is starting up ok. I can do an ifconfig en2 and get back what I would
expect from it on either of the nodes but they still cannot ping each other
even though QNX level networking is working fine between the two nodes. Its
a real head scratcher for me…

Sheldon

<Horst.Hannappel@mbs-software.de> wrote in message
news:ah0fui$nmi$1@mbs-software.de

Sheldon Parkes <> sheldon@onlinedata.com> > wrote:
I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2. The
QNX
networking works perfectly and is using both interfaces as far as I can
tell
from netinfo -l.

I installed TCPIP on both machines and they can ping themselves
perfectly
but cannot ping each other. The setup is dead simple, no routing
involved at
all, IP number range and subnet masks are the same, everything I can
think
of is setup correctly.

Has anyone run across this before? I’m guessing it has something to do
with
the dual ethernet interfaces but who knows…

One thing that has bitten us is you have to wait with starting Sock(l)et
until
the Net.anything have initialised and registered with Net. Otherwise
Socket
wont’t find an interface.


Thanks

Sheldon Parkes
sheldon@onlinedata.com

\

Sheldon Parkes <sheldon@onlinedata.com> wrote:

Socket is starting up ok. I can do an ifconfig en2 and get back what I would
expect from it on either of the nodes but they still cannot ping each other
even though QNX level networking is working fine between the two nodes. Its
a real head scratcher for me…

Could you post the output of “ifconfig [all interfaces]” and “netstat -rn”
on both nodes? Maybe someone will see a problem.

Did you try to connect just one pair of cards and see if the still QNX networking
is OK and TCP/IP not?

Are all Cards on the same hub?

Sheldon

Horst.Hannappel@mbs-software.de> > wrote in message
news:ah0fui$nmi$> 1@mbs-software.de> …
Sheldon Parkes <> sheldon@onlinedata.com> > wrote:
I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2. The
QNX
networking works perfectly and is using both interfaces as far as I can
tell
from netinfo -l.

I installed TCPIP on both machines and they can ping themselves
perfectly
but cannot ping each other. The setup is dead simple, no routing
involved at
all, IP number range and subnet masks are the same, everything I can
think
of is setup correctly.

Has anyone run across this before? I’m guessing it has something to do
with
the dual ethernet interfaces but who knows…

One thing that has bitten us is you have to wait with starting Sock(l)et
until
the Net.anything have initialised and registered with Net. Otherwise
Socket
wont’t find an interface.


Thanks

Sheldon Parkes
sheldon@onlinedata.com



\

Well unfortunately these computers are over an hour’s drive away from me so
I can’t disconnect the one network. Here is the netstar -rn info

The output for ifconfig is what your would expect.

UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST
inet 192.168.4.239 netmask ffffff00 broadcast 192.168.4.255

Here is the netstat info you wanted. There is a sl0 slip interface in there
too, but I tried my setup without the complication of the sliplogin and it
still didn’t work. I also tried starting Socket use en1 ( random guessing )
and that worked no better than on en2.

Name Mtu Network Address Ipkts Ierrs Opkts Oerrs
Coll
en2 1500 0.1.2.ed.bd.2c 1 0 18 0
0
en2 1500 192.168.4 192.168.4.239 1 0 18 0
0
en1* 1500 0.4.76.34.82.99 0 0 0 0
0
sl0* 296 0 0 0 0
0
lo0 1536 26 0 26 0
0
lo0 1536 127 127.0.0.1 26 0 26 0
0

Routing tables
Destination Gateway Flags Refs Use Interface

Route Tree for Protocol Family 2:
(root node) =>
default 192.168.4.239 UG 0 0 en2
127.0.0.1 127.0.0.1 UH 1 0 lo0
192.168.4 192.168.4.239 U 1 26 en2
(root node)

and for the other node

ame Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll
en1* 1500 0.4.76.34.80.d3 0 0 0 0
0
en2 1500 0.4.76.34.82.9e 19 0 25 0
0
en2 1500 192.168.4 192.168.4.236 19 0 25 0
0
sl0 296 66 0 57 0
0
sl0 296 192.168.4 192.168.4.236 66 0 57 0
0
lo0 1536 102 0 102 0
0
lo0 1536 127 127.0.0.1 102 0 102 0
0


Routing tables
Destination Gateway Flags Refs Use Interface

Route Tree for Protocol Family 2:
(root node) =>
default 192.168.4.236 UG 0 0 en2
127.0.0.1 127.0.0.1 UH 1 0 lo0
192.168.4 192.168.4.236 U 1 92 en2
192.168.4.231 192.168.4.236 UH 4 57 sl0
(root node)






<Horst.Hannappel@mbs-software.de> wrote in message
news:ah36iq$c7t$1@mbs-software.de

Sheldon Parkes <> sheldon@onlinedata.com> > wrote:
Socket is starting up ok. I can do an ifconfig en2 and get back what I
would
expect from it on either of the nodes but they still cannot ping each
other
even though QNX level networking is working fine between the two nodes.
Its
a real head scratcher for me…

Could you post the output of “ifconfig [all interfaces]” and “netstat -rn”
on both nodes? Maybe someone will see a problem.

Did you try to connect just one pair of cards and see if the still QNX
networking
is OK and TCP/IP not?

Are all Cards on the same hub?


Sheldon

Horst.Hannappel@mbs-software.de> > wrote in message
news:ah0fui$nmi$> 1@mbs-software.de> …
Sheldon Parkes <> sheldon@onlinedata.com> > wrote:
I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2.
The
QNX
networking works perfectly and is using both interfaces as far as I
can
tell
from netinfo -l.

I installed TCPIP on both machines and they can ping themselves
perfectly
but cannot ping each other. The setup is dead simple, no routing
involved at
all, IP number range and subnet masks are the same, everything I can
think
of is setup correctly.

Has anyone run across this before? I’m guessing it has something to
do
with
the dual ethernet interfaces but who knows…

One thing that has bitten us is you have to wait with starting
Sock(l)et
until
the Net.anything have initialised and registered with Net. Otherwise
Socket
wont’t find an interface.


Thanks

Sheldon Parkes
sheldon@onlinedata.com



\

Sheldon Parkes <sheldon@onlinedata.com> wrote:

I know you have probably already checked your configuration, but what about
your /etc/hosts file from both machines as well /.licenses (licinfo -a) from
both machines ?

ben




: Well unfortunately these computers are over an hour’s drive away from me so
: I can’t disconnect the one network. Here is the netstar -rn info

: The output for ifconfig is what your would expect.

: UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST
: inet 192.168.4.239 netmask ffffff00 broadcast 192.168.4.255

: Here is the netstat info you wanted. There is a sl0 slip interface in there
: too, but I tried my setup without the complication of the sliplogin and it
: still didn’t work. I also tried starting Socket use en1 ( random guessing )
: and that worked no better than on en2.

: Name Mtu Network Address Ipkts Ierrs Opkts Oerrs
: Coll
: en2 1500 0.1.2.ed.bd.2c 1 0 18 0
: 0
: en2 1500 192.168.4 192.168.4.239 1 0 18 0
: 0
: en1* 1500 0.4.76.34.82.99 0 0 0 0
: 0
: sl0* 296 0 0 0 0
: 0
: lo0 1536 26 0 26 0
: 0
: lo0 1536 127 127.0.0.1 26 0 26 0
: 0

: Routing tables
: Destination Gateway Flags Refs Use Interface

: Route Tree for Protocol Family 2:
: (root node) =>
: default 192.168.4.239 UG 0 0 en2
: 127.0.0.1 127.0.0.1 UH 1 0 lo0
: 192.168.4 192.168.4.239 U 1 26 en2
: (root node)

: and for the other node

: ame Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll
: en1* 1500 0.4.76.34.80.d3 0 0 0 0
: 0
: en2 1500 0.4.76.34.82.9e 19 0 25 0
: 0
: en2 1500 192.168.4 192.168.4.236 19 0 25 0
: 0
: sl0 296 66 0 57 0
: 0
: sl0 296 192.168.4 192.168.4.236 66 0 57 0
: 0
: lo0 1536 102 0 102 0
: 0
: lo0 1536 127 127.0.0.1 102 0 102 0
: 0


: Routing tables
: Destination Gateway Flags Refs Use Interface

: Route Tree for Protocol Family 2:
: (root node) =>
: default 192.168.4.236 UG 0 0 en2
: 127.0.0.1 127.0.0.1 UH 1 0 lo0
: 192.168.4 192.168.4.236 U 1 92 en2
: 192.168.4.231 192.168.4.236 UH 4 57 sl0
: (root node)






: <Horst.Hannappel@mbs-software.de> wrote in message
: news:ah36iq$c7t$1@mbs-software.de
:> Sheldon Parkes <sheldon@onlinedata.com> wrote:
:> > Socket is starting up ok. I can do an ifconfig en2 and get back what I
: would
:> > expect from it on either of the nodes but they still cannot ping each
: other
:> > even though QNX level networking is working fine between the two nodes.
: Its
:> > a real head scratcher for me…
:>
:> Could you post the output of “ifconfig [all interfaces]” and “netstat -rn”
:> on both nodes? Maybe someone will see a problem.
:>
:> Did you try to connect just one pair of cards and see if the still QNX
: networking
:> is OK and TCP/IP not?
:>
:> Are all Cards on the same hub?
:>
:> >
:> > Sheldon
:> >
:> > <Horst.Hannappel@mbs-software.de> wrote in message
:> > news:ah0fui$nmi$1@mbs-software.de
:> >> Sheldon Parkes <sheldon@onlinedata.com> wrote:
:> >> > I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2.
: The
:> > QNX
:> >> > networking works perfectly and is using both interfaces as far as I
: can
:> > tell
:> >> > from netinfo -l.
:> >> >
:> >> > I installed TCPIP on both machines and they can ping themselves
:> > perfectly
:> >> > but cannot ping each other. The setup is dead simple, no routing
:> > involved at
:> >> > all, IP number range and subnet masks are the same, everything I can
:> > think
:> >> > of is setup correctly.
:> >> >
:> >> > Has anyone run across this before? I’m guessing it has something to
: do
:> > with
:> >> > the dual ethernet interfaces but who knows…
:> >>
:> >> One thing that has bitten us is you have to wait with starting
: Sock(l)et
:> > until
:> >> the Net.anything have initialised and registered with Net. Otherwise
:> > Socket
:> >> wont’t find an interface.
:> >>
:> >> >
:> >> > Thanks
:> >> >
:> >> > Sheldon Parkes
:> >> > sheldon@onlinedata.com
:> >> >
:> >> >
:> >> >
:> >> >
:> >> >
:> >
:> >

Nope those are both good, plenty of licenses and the names are correct in
the hosts file… however I am going to post some new findings I made this
morning :slight_smile:

Thanks

Sheldon

“Brian Northgrave” <ben@qnx.com> wrote in message
news:ah3t9q$cgo$1@nntp.qnx.com

Sheldon Parkes <> sheldon@onlinedata.com> > wrote:

I know you have probably already checked your configuration, but what
about
your /etc/hosts file from both machines as well /.licenses (licinfo -a)
from
both machines ?

ben




: Well unfortunately these computers are over an hour’s drive away from me
so
: I can’t disconnect the one network. Here is the netstar -rn info

: The output for ifconfig is what your would expect.

: UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST
: inet 192.168.4.239 netmask ffffff00 broadcast 192.168.4.255

: Here is the netstat info you wanted. There is a sl0 slip interface in
there
: too, but I tried my setup without the complication of the sliplogin and
it
: still didn’t work. I also tried starting Socket use en1 ( random
guessing )
: and that worked no better than on en2.

: Name Mtu Network Address Ipkts Ierrs Opkts Oerrs
: Coll
: en2 1500 0.1.2.ed.bd.2c 1 0 18
0
: 0
: en2 1500 192.168.4 192.168.4.239 1 0 18 0
: 0
: en1* 1500 0.4.76.34.82.99 0 0 0 0
: 0
: sl0* 296 0 0 0 0
: 0
: lo0 1536 26 0 26 0
: 0
: lo0 1536 127 127.0.0.1 26 0 26 0
: 0

: Routing tables
: Destination Gateway Flags Refs Use Interface

: Route Tree for Protocol Family 2:
: (root node) =
: default 192.168.4.239 UG 0 0 en2
: 127.0.0.1 127.0.0.1 UH 1 0 lo0
: 192.168.4 192.168.4.239 U 1 26 en2
: (root node)

: and for the other node

: ame Mtu Network Address Ipkts Ierrs Opkts Oerrs
Coll
: en1* 1500 0.4.76.34.80.d3 0 0 0 0
: 0
: en2 1500 0.4.76.34.82.9e 19 0 25 0
: 0
: en2 1500 192.168.4 192.168.4.236 19 0 25 0
: 0
: sl0 296 66 0 57 0
: 0
: sl0 296 192.168.4 192.168.4.236 66 0 57 0
: 0
: lo0 1536 102 0 102 0
: 0
: lo0 1536 127 127.0.0.1 102 0 102 0
: 0


: Routing tables
: Destination Gateway Flags Refs Use Interface

: Route Tree for Protocol Family 2:
: (root node) =
: default 192.168.4.236 UG 0 0 en2
: 127.0.0.1 127.0.0.1 UH 1 0 lo0
: 192.168.4 192.168.4.236 U 1 92 en2
: 192.168.4.231 192.168.4.236 UH 4 57 sl0
: (root node)






: <> Horst.Hannappel@mbs-software.de> > wrote in message
: news:ah36iq$c7t$> 1@mbs-software.de> …
:> Sheldon Parkes <> sheldon@onlinedata.com> > wrote:
:> > Socket is starting up ok. I can do an ifconfig en2 and get back what
I
: would
:> > expect from it on either of the nodes but they still cannot ping each
: other
:> > even though QNX level networking is working fine between the two
nodes.
: Its
:> > a real head scratcher for me…
:
:> Could you post the output of “ifconfig [all interfaces]” and
“netstat -rn”
:> on both nodes? Maybe someone will see a problem.
:
:> Did you try to connect just one pair of cards and see if the still QNX
: networking
:> is OK and TCP/IP not?
:
:> Are all Cards on the same hub?
:
:
:> > Sheldon
:
:> > <> Horst.Hannappel@mbs-software.de> > wrote in message
:> > news:ah0fui$nmi$> 1@mbs-software.de> …
:> >> Sheldon Parkes <> sheldon@onlinedata.com> > wrote:
:> >> > I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and
en2.
: The
:> > QNX
:> >> > networking works perfectly and is using both interfaces as far as
I
: can
:> > tell
:> >> > from netinfo -l.
:
:> >> > I installed TCPIP on both machines and they can ping themselves
:> > perfectly
:> >> > but cannot ping each other. The setup is dead simple, no routing
:> > involved at
:> >> > all, IP number range and subnet masks are the same, everything I
can
:> > think
:> >> > of is setup correctly.
:
:> >> > Has anyone run across this before? I’m guessing it has something
to
: do
:> > with
:> >> > the dual ethernet interfaces but who knows…
:
:> >> One thing that has bitten us is you have to wait with starting
: Sock(l)et
:> > until
:> >> the Net.anything have initialised and registered with Net. Otherwise
:> > Socket
:> >> wont’t find an interface.
:
:
:> >> > Thanks
:
:> >> > Sheldon Parkes
:> >> > > sheldon@onlinedata.com
:
:
:
:
:
:
:

A follow up of things I tried to resolve why I can’t ping between 2 nodes. I
ran netsniff on the node that is 192.168.4.236 and pinged it from
192.168.4.239 and saw that the packets were actually arriving. Now can
someone please interpret what this means exactly:

In my netmap file node 8 ( which is 192.168.4.239 ) has

8 1 000102edbd2c
8 2 000476348299

Node 9 ( which is 192.168.4.236 ) has

9 1 0004763400d3
9 2 00047634829e


Now the MAC address for the Src in the netsniff output is the correct one
for en2 as I would expect. What exactly is ARP request sender? ( it contains
the MAC address for en1 for some reason - which is likely my problem ).

I tried using arp -s to map the ip address to the correct MAC address on
both nodes but still can’t ping. I seek the wisdom of the gurus out there…

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914822
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914823
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914824
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914825
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914826
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

“Sheldon Parkes” <sheldon@onlinedata.com> wrote in message
news:agv1t2$hvq$1@inn.qnx.com

I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2. The
QNX
networking works perfectly and is using both interfaces as far as I can
tell
from netinfo -l.

I installed TCPIP on both machines and they can ping themselves perfectly
but cannot ping each other. The setup is dead simple, no routing involved
at
all, IP number range and subnet masks are the same, everything I can think
of is setup correctly.

Has anyone run across this before? I’m guessing it has something to do
with
the dual ethernet interfaces but who knows…

Thanks

Sheldon Parkes
sheldon@onlinedata.com

\

Sheldon Parkes <sheldon@onlinedata.com> wrote:

Route Tree for Protocol Family 2:
(root node) =
default 192.168.4.239 UG 0 0 en2
127.0.0.1 127.0.0.1 UH 1 0 lo0
192.168.4 192.168.4.239 U 1 26 en2
(root node)

This is just a guess, but since these machines are only connected to each
and not the internet in general, you might try deleting the default route.

If you are running Socket/Socklet on QNX 4.25 then you are NOT running the
latest TCP/IP for QNX4, I would STRONGLY reccomend that you download and
install the latest “TCP/IP 4.25” update from the QNX website:

http://www.qnx.com/developer/download/updates/qnx4

In the new stack, Socket/Socklet are replaced by “Tcpip”, and the entire
stack is much more robust. The output from sin fd and netstat -nr will
also be more useful in diagnosing your problems.

Cheers,
Camz.

Sheldon Parkes <sheldon@onlinedata.com> wrote:

A follow up of things I tried to resolve why I can’t ping between 2 nodes. I
ran netsniff on the node that is 192.168.4.236 and pinged it from
192.168.4.239 and saw that the packets were actually arriving. Now can
someone please interpret what this means exactly:

This is in my mind for a while, but …

You said FLEET is working, but no TCPIP. Could it be mis-configured
that LAN 1 and 2 is actuall “crossed” ? That migh explain why ARP
didn’t get through (arp -na empty).

Can you do it one interface by one. I mean, only start one Net.etherXXXX
on each node, make sure their MAC address/LAN number matched whatever
in netmap, and make sure FLEET is working between them.

Then start Tcpip see if it works on this pair of connection.

If all good, then starting add second interface for each node …

-xtang

In my netmap file node 8 ( which is 192.168.4.239 ) has

8 1 000102edbd2c
8 2 000476348299

Node 9 ( which is 192.168.4.236 ) has

9 1 0004763400d3
9 2 00047634829e



Now the MAC address for the Src in the netsniff output is the correct one
for en2 as I would expect. What exactly is ARP request sender? ( it contains
the MAC address for en1 for some reason - which is likely my problem ).

I tried using arp -s to map the ip address to the correct MAC address on
both nodes but still can’t ping. I seek the wisdom of the gurus out there…

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914822
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914823
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914824
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914825
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914826
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target 00:00:00:00:00:00
192.168.4.236

“Sheldon Parkes” <> sheldon@onlinedata.com> > wrote in message
news:agv1t2$hvq$> 1@inn.qnx.com> …
I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2. The
QNX
networking works perfectly and is using both interfaces as far as I can
tell
from netinfo -l.

I installed TCPIP on both machines and they can ping themselves perfectly
but cannot ping each other. The setup is dead simple, no routing involved
at
all, IP number range and subnet masks are the same, everything I can think
of is setup correctly.

Has anyone run across this before? I’m guessing it has something to do
with
the dual ethernet interfaces but who knows…

Thanks

Sheldon Parkes
sheldon@onlinedata.com

\

I’m pretty sure that FLEET must be set up correctly. If I had the wires
crossed so to speak, then I would not expect node 8 to see node 9 at all
since en1 would be trying to talk to en2 and vice versa. However let me
bounce this off you:

I noticed that on the one node the Net driver for en2 is in PCI slot 0 and
is started first. On the other node the Net driver for en1 is in PCI slot 0
and is started first. Could Socket be confused by this somehow? I’ll try to
set this up here in the office and see if it causes a problem.

Sheldon

“Xiaodan Tang” <xtang@qnx.com> wrote in message
news:ah410f$f20$1@nntp.qnx.com

Sheldon Parkes <> sheldon@onlinedata.com> > wrote:
A follow up of things I tried to resolve why I can’t ping between 2
nodes. I
ran netsniff on the node that is 192.168.4.236 and pinged it from
192.168.4.239 and saw that the packets were actually arriving. Now can
someone please interpret what this means exactly:

This is in my mind for a while, but …

You said FLEET is working, but no TCPIP. Could it be mis-configured
that LAN 1 and 2 is actuall “crossed” ? That migh explain why ARP
didn’t get through (arp -na empty).

Can you do it one interface by one. I mean, only start one Net.etherXXXX
on each node, make sure their MAC address/LAN number matched whatever
in netmap, and make sure FLEET is working between them.

Then start Tcpip see if it works on this pair of connection.

If all good, then starting add second interface for each node …

-xtang

In my netmap file node 8 ( which is 192.168.4.239 ) has

8 1 000102edbd2c
8 2 000476348299

Node 9 ( which is 192.168.4.236 ) has

9 1 0004763400d3
9 2 00047634829e


Now the MAC address for the Src in the netsniff output is the correct
one
for en2 as I would expect. What exactly is ARP request sender? ( it
contains
the MAC address for en1 for some reason - which is likely my problem ).

I tried using arp -s to map the ip address to the correct MAC address on
both nodes but still can’t ping. I seek the wisdom of the gurus out
there…

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914822
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target
00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914823
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target
00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914824
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target
00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914825
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target
00:00:00:00:00:00
192.168.4.236

Src 000476 348299 [8] Dst FFFFFF FFFFFF [?] Len 48 Lan 2(Eth)
1026914826
ARP request sender 00:01:02:ed:bd:2c 192.168.4.239 target
00:00:00:00:00:00
192.168.4.236

“Sheldon Parkes” <> sheldon@onlinedata.com> > wrote in message
news:agv1t2$hvq$> 1@inn.qnx.com> …
I have 2 QNX 4 boxes set up each with 2 ethernet cards en1 and en2. The
QNX
networking works perfectly and is using both interfaces as far as I can
tell
from netinfo -l.

I installed TCPIP on both machines and they can ping themselves
perfectly
but cannot ping each other. The setup is dead simple, no routing
involved
at
all, IP number range and subnet masks are the same, everything I can
think
of is setup correctly.

Has anyone run across this before? I’m guessing it has something to do
with
the dual ethernet interfaces but who knows…

Thanks

Sheldon Parkes
sheldon@onlinedata.com



\

Sheldon Parkes <sheldon@onlinedata.com> wrote:

I noticed that on the one node the Net driver for en2 is in PCI slot 0 and
is started first. On the other node the Net driver for en1 is in PCI slot 0
and is started first. Could Socket be confused by this somehow? I’ll try to
set this up here in the office and see if it causes a problem.

I think that will impact which network “slot” they will use, and thusly, will
affect which NIC is considered en0 and which is en1

I would actually go so far as to remove the 2nd NIC from each machine and
get it working that way, then add the 2nd NIC to each machine and debug as
required. With two NICs, as xtang mentions, it’s pretty easy to get en0 and
en1 mixed up.

Cheers,
Camz.