[Olsr-users] Discrepancy between ETX value and ping packet loss

Henning Rogge (spam-protected)
Wed Mar 18 21:31:54 CET 2015


I would suggest not using the 802.11b datarates... use 6 MBit/s... or
maybe even 9/12 MBit/s.

Nothing that's easily available.

Henning Rogge

On Wed, Mar 18, 2015 at 6:25 PM, O'Keeffe, Daniel
<(spam-protected)> wrote:
> Hey Henning,
> Thanks for the confirmation. Would there be any disadvantage to just pushing
> the broadcast rate all the way to 11 (spinal tap style :))? Also, off the
> top of your head can you think of any other typical Mac or transport layer
> tuning that people do for high throughput TCP?
> Cheers,
> Dan
>
> On 18 Mar 2015 05:28, Henning Rogge <(spam-protected)> wrote:
>
> Yes,
>
> that is the reason... many mesh networks increase the basic multicast
> rate to 6 MBit/s to ignore the old 802.11b rates.
>
> Henning Rogge
>
> On Tue, Mar 17, 2015 at 11:06 PM, Dan O'Keeffe <(spam-protected)>
> wrote:
>> Hey again,
>> Actually it's ok, I figured it out. Because the OLSR HELLO messages are
>> broadcast at the 1Mbps data rate for 802.11, they experience a different
>> packet loss to the 11Mbps unicast traffic.
>> Cheers,
>> Dan
>> On 17/03/15 20:43, Dan O'Keeffe wrote:
>>> Hello,
>>> I'm trying to use the OLSR ETX extensions to determine the path quality
>>> to different destinations (so my application can send data to the
>>> destination with the best path as the quality changes over time).
>>> However the ETX values I'm seeing don't seem to correspond particularly
>>> well to the packet loss I observe when I manually ping the different
>>> destinations.
>>>
>>> More specifically, I have a two node network with each node running
>>> olsrd and one node pinging the other (using a ping payload size of 0
>>> so that the actual ping is just a 28 byte header). At a certain distance
>>> the ping packet loss goes to 40%, but the ETX I observe using the
>>> txtinfo plugin is 1.000 (i.e. from the output of 'echo "/all" | nc
>>> localhost 2006).
>>>
>>> Does anyone have an idea as to why I'm not seeing any packet loss
>>> affecting the ETX? Using tshark I can see the OLSR control
>>> messages being sent and received. However it's hard to tell whether
>>> any of the control messages are actually being dropped, and so
>>> I'm not sure whether the issue is that the OLSR control messages just
>>> aren't experiencing packet loss or I haven't configured the ETX
>>> calculation properly. From what I understand the ETX calculation should
>>>  be based on the number of hello messages sucessfully sent and received
>>> over a sliding window, but perhaps there's more to the implementation?
>>>
>>> The output of olsrd --version is:
>>>
>>> *** olsr.org -  0.6.3-git_-hash_e334b612adcb7c22239a9dcafaf5b7ba  - ***
>>>  Build date: 2013-03-28 03:47:37 on marid
>>>
>>>
>>> I have made a few tweaks to the default config (see attached), but
>>> I don't see that they should make a difference.
>>>
>>>
>>> Finally, I should note that I'm running in an emulated environment
>>> (using the CORE EMANE wireless network emulator). The only way I could
>>> see that making a difference is that perhaps the same packet loss isn't
>>> being applied to broadcast traffic as to normal traffic, but before I
>>> check that on the CORE EMANE mailing list I'd just like to be sure there
>>> isn't an obvious issue with my olsrd configuration.
>>>
>>> Thanks for any help!
>>> Dan
>>>
>>>
>>>
>>
>>
>> --
>> Olsr-users mailing list
>> (spam-protected)
>> https://lists.olsr.org/mailman/listinfo/olsr-users




More information about the Olsr-users mailing list