Tx_bytes

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Tx_bytes

Mister_New
Hello everybody, I have a question and I hope that somebody can help me. I know that in netdevice.h file there is the struct net_device_stats with rx_bytes and tx_bytes fields. I wrote a program that prints on stdout (every x seconds) these two numbers. While for rx_bytes I know that this is the number of correctly received bytes, I was wondering: how can I know the number of total successfully transmitted bytes? This question comes from the fact that tx_bytes is incremented when is put into the buffer and so it counts also retransmissions (and by the way some of them are performed by hardware...), but I'm interested in just the correct tx_bytes, because I have to compute the goodput. Is there an expert guy (or maybe more than one) that knows how to help me?

Thanks in advance.

Regards.
Reply | Threaded
Open this post in threaded view
|

Re: Tx_bytes

Eric Malkowski
I may be completely wrong, but if you were to look at the tx_bytes that
you are already looking at, and then look at the statistics that the
"athstats" utility reports for the wifiX interface you may be able to
use those numbers to get out of total attempted tx_bytes how many were
retries and such or at least some sort of better insight into how the
interface is actually doing with data transmitted.  I know for long
distance links the ack timeouts need to be increased or the data rate
never goes up.  You might want to find where the code dispositions data
it needs to get across but hasn't due to no ack w/in the ack timeout
time -- see if any stats anywhere are incremented or something.  Have a
look at the long distance tuning in the FAQs and then search for the
timeouts in the sourcecode and see where it drops data when acks aren't
seen etc...

Again -- I could be way off here as I don't know the guts of MADWIFI,
but I've been building and using it for kernel 2.4 and 2.6 for several
years now -- since 2005 since it's got major flexability for doing all
kinds of cool things!  I really need to switch over to ath5k and ath9k
now that these are pretty feature rich and fully open etc.

Good luck.

-Eric Malkowski

Mister_New wrote:

> Hello everybody, I have a question and I hope that somebody can help me. I
> know that in netdevice.h file there is the struct net_device_stats with
> rx_bytes and tx_bytes fields. I wrote a program that prints on stdout (every
> x seconds) these two numbers. While for rx_bytes I know that this is the
> number of *correctly* received bytes, I was wondering: how can I know the
> number of total *successfully* transmitted bytes? This question comes from
> the fact that tx_bytes is incremented when is put into the buffer and so it
> counts also retransmissions (and by the way some of them are performed by
> hardware...), but I'm interested in just the correct tx_bytes, because I
> have to compute the goodput. Is there an expert guy (or maybe more than one)
> that knows how to help me?
>
> Thanks in advance.
>
> Regards.
>
> --
> View this message in context: http://madwifi-users.20070.n2.nabble.com/Tx-bytes-tp7015933p7015933.html
> Sent from the Madwifi Users mailing list archive at Nabble.com.
>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure
> contains a definitive record of customers, application performance,
> security threats, fraudulent activity, and more. Splunk takes this
> data and makes sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-novd2d
> _______________________________________________
> Madwifi-users mailing list
> [hidden email]
> https://lists.sourceforge.net/lists/listinfo/madwifi-users
>  


------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure
contains a definitive record of customers, application performance,
security threats, fraudulent activity, and more. Splunk takes this
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d
_______________________________________________
Madwifi-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/madwifi-users
Reply | Threaded
Open this post in threaded view
|

Re: Tx_bytes

Mister_New
Dear Mr. Eric Malkowski, first of all, thanks for your help. I really appreciated that. I can say that your solution could be acceptable, and I'll try it, but I think that a deeper job should be done for my task. I mean, I should work at bytes level, and it means adding a new field tx_bytes_success and inserting it somewhere into the code so that (after some modifications) I can print it beside tx_bytes and compare them. All of this launching just one program. This is my idea, but I don't know if it's sustainable. Moreover, the code is pretty difficult for my level... Could it be possible the operation that I mentioned above? Thanks again for your time.

Regards.