still coping with zero packet loss over wireless
Dave Taht
dave.taht at gmail.com
Sun May 29 11:10:00 PDT 2011
On Tue, May 24, 2011 at 8:50 AM, John W. Linville <linville at tuxdriver.com>wrote:
> On Tue, May 24, 2011 at 02:46:59PM +0000, Dave Taht wrote:
> > yesterday I finally got a chance to move a few dozen meters out of the
> lab
> > and test the latest build of uberwrt "capetown" and debloat-testing.
> >
> > I'd hoped with the debloating techniques in place in capetown - reduced
> > buffers (4), reduced sw retries (2), hw retries (2, or so I thought) I'd
> > actually see some packet loss.
> >
> > and what I saw instead, was pings that would take as long as 1.6 seconds
> to
> > complete, and zero packet loss until I moved completely out of range of
> the
> > router.
> >
> > I never thought it would be so hard to lose a packet in my life!
> >
> > Is there some system tunable, somewhere, in the linux wireless stack that
> > I've missed, in getting packets to actually fail in 10s of ms?
>
> Not one of which I am aware. That sort of thing is going to depend
> quite a bit on the hardware itself, and it's driver. Maybe some
> ath9k folk can comment?
>
>
I finally got out from under enough to have a chance to look at this problem
with debugfs
root at io:/sys/kernel/debug# cat
./ieee80211/phy0/netdev:wlan2/stations/ce:3d:c7:b0:ae:78/tx_retry_count
291
My assumption is that this is not the value for retries, but the total
number of times a given station has had to retry.
Is there anything useful that I can poke around in down here?
(I'd actually gone into this to see if I could pull more info out on the
ag71xx issue on the other thread)
John
> --
> John W. Linville Someday the world will need a hero, and you
> linville at tuxdriver.com might be all we have. Be ready.
>
--
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
http://the-edge.blogspot.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/bloat-devel/attachments/20110529/42fc82a4/attachment.html>
More information about the Bloat-devel
mailing list