Olsrd cann't ping two-hops node - Nexus7

[Abubakr Ali]

Hello everyone,

I have an ad-hoc network that is up-and-running on three tablets (all are
Nexus7 that are rooted with CyanogenMod (cm-10.2-20130919-NIGHTLY-grouper)).

I downloaded, built, installed and run OLSRd stable version 0.6.6. In the
"olsrd.conf" file, I have set:
Willingness to 7,
TCRedundancy to 2 (as 0 and 1 are not allowed)
LinkQualityLevel to 2 (as 1 is not allowed).
Note: any other configuration option has not been touched.

Now when running OLSRd with debug-level 1, I can see three tables:
Links,
Neighbors
Two-hop Neighbors tables, and my network looks like the following:
node1 ----> node2 <---- node3
node1 can only see node2,
node3 can only see node2
node2 see both: node1 and node3
(note: The routes-table changes as I change the location of the tablets)

However, when trying to *ping node3 (from node1*'s terminal) and monitor
the traffic that is coming-and-leaving node2 (from node2 itself using
wireshark), I can see that the* *ping-request is coming to node2 and see
node2 as it is forwarding the request to node3. But *there is no
response*from node3 to node2 and node2 to node1. it seems to drop
packets and the
sent-packets never reach node3.

am I missing any configuration that should be set in olsrd.conf ?

Ali, Abubakr
Lancaster, UK

[Henning Rogge]

Post by Abubakr Ali
Hello everyone,
I have an ad-hoc network that is up-and-running on three tablets (all
are Nexus7 that are rooted with CyanogenMod
(cm-10.2-20130919-NIGHTLY-grouper)).
I downloaded, built, installed and run OLSRd stable version 0.6.6. In
Willingness to 7,
TCRedundancy to 2 (as 0 and 1 are not allowed)
LinkQualityLevel to 2 (as 1 is not allowed).
Note: any other configuration option has not been touched.
Links,
Neighbors
node1 ----> node2 <---- node3
node1 can only see node2,
node3 can only see node2
node2 see both: node1 and node3
(note: The routes-table changes as I change the location of the tablets)
However, when trying to *ping node3 (from node1*'s terminal) and monitor
the traffic that is coming-and-leaving node2 (from node2 itself using
wireshark), I can see that the**ping-request is coming to node2 and see
node2 as it is forwarding the request to node3. But *there is no
response* from node3 to node2 and node2 to node1. it seems to drop
packets and the sent-packets never reach node3.
am I missing any configuration that should be set in olsrd.conf ?

Default configuration would be a nearly empty configuration file, just
with the interface definition (but without any timings of Hellos/TCs).

Do you see a route on node 3 for node 1?

Henning Rogge

[Abubakr Ali]

Thanks Henning Rogge for your reply ..

Regards the route from node3 to node1, and at the beginning where all the
three tablets are close to, and in-range of, each others, I see the
following in the */proc/net/route* :

Destination==node1
Gateway==node1,
and ...
Destination==node2
Gateway==node2,

when changing the location of the tablets, then the routes change as follow:

Destination==node1
Gateway==node2,
and ...
Destination==node2
Gateway==node2,

accordingly, this mean that the HELLO and TC msgs are exchanged (i.e.: the
number-of-hops was calculated and that node1 is reachable through node2),
right ?

you sounded like if the default-values in the .conf file do not do me any
favor and that I have to hard-coded all the default-values myself, is that
true ?

Thanks,
Abubakr


Ali, Abubakr
Lancaster, UK

Post by Henning Rogge

Post by Abubakr Ali
Hello everyone,
I have an ad-hoc network that is up-and-running on three tablets (all
are Nexus7 that are rooted with CyanogenMod
(cm-10.2-20130919-NIGHTLY-grouper)).
I downloaded, built, installed and run OLSRd stable version 0.6.6. In
Willingness to 7,
TCRedundancy to 2 (as 0 and 1 are not allowed)
LinkQualityLevel to 2 (as 1 is not allowed).
Note: any other configuration option has not been touched.
Links,
Neighbors
node1 ----> node2 <---- node3
node1 can only see node2,
node3 can only see node2
node2 see both: node1 and node3
(note: The routes-table changes as I change the location of the tablets)
However, when trying to *ping node3 (from node1*'s terminal) and monitor
the traffic that is coming-and-leaving node2 (from node2 itself using
wireshark), I can see that the**ping-request is coming to node2 and see
node2 as it is forwarding the request to node3. But *there is no
response* from node3 to node2 and node2 to node1. it seems to drop
packets and the sent-packets never reach node3.
am I missing any configuration that should be set in olsrd.conf ?

Default configuration would be a nearly empty configuration file, just
with the interface definition (but without any timings of Hellos/TCs).
Do you see a route on node 3 for node 1?
Henning Rogge
--
Diplom-Informatiker Henning Rogge , Fraunhofer-Institut für
Kommunikation, Informationsverarbeitung und Ergonomie FKIE
Kommunikationssysteme (KOM)
Fraunhofer Straße 20, 53343 Wachtberg, Germany
Telefon [phone], Fax [phone]
--
Olsr-users mailing list
https://lists.olsr.org/mailman/listinfo/olsr-users

[Henning Rogge]

Post by Abubakr Ali
Thanks Henning Rogge for your reply ..
Regards the route from node3 to node1, and at the beginning where all
the three tablets are close to, and in-range of, each others, I see the
Destination==node1
Gateway==node1,
and ...
Destination==node2
Gateway==node2,
Destination==node1
Gateway==node2,
and ...
Destination==node2
Gateway==node2,
the number-of-hops was calculated and that node1 is reachable through
node2), right ?

If this is true in both directions on all three nodes, your should be
able to ping node3 from node1 (at least from the routing table point of
view).

You can use tcpdump

Post by Abubakr Ali
you sounded like if the default-values in the .conf file do not do me
any favor and that I have to hard-coded all the default-values myself,
is that true ?

The problem with huge config files with lots of "default" values is that
its easy to overlook a changed value. I would suggest only adding things
to the configuration file that are not using the default value.

Henning Rogge

[Abubakr Ali]

Hi Henning Rogge,

I've finally figured out what was the problem. It was simply the "Iptables
Rules" that were configured, in the Nexus7, to drop all the forwarded
packets. Now, after adding another rule that enables node2 to forward
packet that aren't meant to it, the pinging request from node1 to node3
and the pinging reply is done through node2 smoothly.
My next step is to test the network throughput. Therefore, I am building an
android app as packets-generator, wish me luck ;-)

Thanks


Ali, Abubakr

Post by Henning Rogge

Post by Abubakr Ali
Thanks Henning Rogge for your reply ..
Regards the route from node3 to node1, and at the beginning where all
the three tablets are close to, and in-range of, each others, I see the
Destination==node1
Gateway==node1,
and ...
Destination==node2
Gateway==node2,
when changing the location of the tablets, then the routes change as
Destination==node1
Gateway==node2,
and ...
Destination==node2
Gateway==node2,
the number-of-hops was calculated and that node1 is reachable through
node2), right ?

If this is true in both directions on all three nodes, your should be
able to ping node3 from node1 (at least from the routing table point of
view).
You can use tcpdump

Post by Abubakr Ali
you sounded like if the default-values in the .conf file do not do me
any favor and that I have to hard-coded all the default-values myself,
is that true ?

The problem with huge config files with lots of "default" values is that
its easy to overlook a changed value. I would suggest only adding things
to the configuration file that are not using the default value.
Henning Rogge
--
Diplom-Informatiker Henning Rogge , Fraunhofer-Institut für
Kommunikation, Informationsverarbeitung und Ergonomie FKIE
Kommunikationssysteme (KOM)
Fraunhofer Straße 20, 53343 Wachtberg, Germany
Telefon [phone], Fax [phone]

[Henning Rogge]

Good to hear this... :)

Henning

Post by Abubakr Ali
Hi Henning Rogge,
I've finally figured out what was the problem. It was simply the "Iptables
Rules" that were configured, in the Nexus7, to drop all the forwarded
packets. Now, after adding another rule that enables node2 to forward packet
that aren't meant to it, the pinging request from node1 to node3 and the
pinging reply is done through node2 smoothly.
My next step is to test the network throughput. Therefore, I am building an
android app as packets-generator, wish me luck ;-)
Thanks
Ali, Abubakr

Post by Henning Rogge

Post by Abubakr Ali
Thanks Henning Rogge for your reply ..
Regards the route from node3 to node1, and at the beginning where all
the three tablets are close to, and in-range of, each others, I see the
Destination==node1
Gateway==node1,
and ...
Destination==node2
Gateway==node2,
Destination==node1
Gateway==node2,
and ...
Destination==node2
Gateway==node2,
the number-of-hops was calculated and that node1 is reachable through
node2), right ?

If this is true in both directions on all three nodes, your should be
able to ping node3 from node1 (at least from the routing table point of
view).
You can use tcpdump

Post by Abubakr Ali
you sounded like if the default-values in the .conf file do not do me
any favor and that I have to hard-coded all the default-values myself,
is that true ?

The problem with huge config files with lots of "default" values is that
its easy to overlook a changed value. I would suggest only adding things
to the configuration file that are not using the default value.
Henning Rogge
--
Diplom-Informatiker Henning Rogge , Fraunhofer-Institut für
Kommunikation, Informationsverarbeitung und Ergonomie FKIE
Kommunikationssysteme (KOM)
Fraunhofer Straße 20, 53343 Wachtberg, Germany
Telefon [phone], Fax [phone]

--
Olsr-users mailing list
https://lists.olsr.org/mailman/listinfo/olsr-users