Wed, 03 Jan 2018 17:33:12 +0200 було написано David Wolfskill
<da...@catwhisker.org>:
Hi,
https://reviews.freebsd.org/D10728 fixes output stats (I cannot check it
with Intel 4965 - sometimes firmware throws an exception - with or without
this patch - that's why it is not in the tree yet).
I recently configured my "perimeter device" (a 4-homed FreeBSD box
named "bats"; see <http://www.catwhisker.org/~david/FreeBSD/upgrade.html>
for a diagram) to make use of FQ-CoDel for outbound traffic on the
Internet-facing NIC (em2, in this case).
I failed to find a way to monitor the traffic rates in a way that I
found suitably convenient, so I cobbled up a Perl script to grab the
output of "netstat -nibdf link" at intervals (timed via Time::HiRes), do
the arithmetic, and display the results.
For the perimeter device, it seems to work as expected.
For my laptop, using the wlan0 (iwn0) NIC, the report of input traffic
looks reasonable, but output ... makes no sense at all to me.
As illustration, I'll show the reported results for both bats and my
laptop while I run a "speed" test from
<http://beta.speedtest.net/run?r=f>;
note that my Internet connection is (residential) ADSL. Here's what my
ISP stated about it (from a conversation some time back):
| The line is at 6M/768K with good synch readings and I don't see
| any reason you wouldn't be able to get 5.1Mbps/640Kbps out of
| it which is about the yield after overhead.
Note that I start the test just after the script has reported fro the
first interval (which I have specified as 5 seconds for this):
bats(11.1-S)[18] net_mon -e -w 5 -c 15 em{1,2}
em1: 172.17.0.1/16 Ethernet autoselect (1000baseT <full-duplex>)
em2: 198.144.209.73/24 Ethernet autoselect (10baseT/UTP
<half-duplex>)
em1
em2
Time In
Out In Out
Rate/sec Bits Errs Drops Bits Errs Drops Coll
Bits Errs Drops Bits Errs Drops Coll
1514993320.453 3.5K 0.0 0.0 3.3K 0.0 0.0 0.0
403.0K 0.0 0.0 173.2K 0.0 0.0 1.8
1514993325.454 24.1K 0.0 0.0 16.4K 0.0 0.0 0.0
31.0K 0.0 0.0 27.9K 0.0 0.0 0.0
1514993330.454 22.4K 0.0 0.0 525.3K 0.0 0.0 0.0
1.5M 0.0 0.0 142.2K 0.0 0.0 1.4
1514993335.455 143.9K 0.0 0.0 5.1M 0.0 0.0 0.0
5.3M 0.0 0.0 340.5K 0.0 0.0 34.2
1514993340.455 135.4K 0.0 0.0 5.0M 0.0 0.0 0.0
5.4M 0.0 0.0 227.8K 0.0 0.0 8.8
1514993345.454 104.1K 0.0 0.0 4.2M 0.0 0.0 0.0
5.4M 0.0 0.0 249.2K 0.0 0.0 25.6
1514993350.453 544.1K 0.0 0.0 544.4K 0.0 0.0 0.0
858.5K 0.0 0.0 644.5K 0.0 0.0 5.4
1514993355.454 585.0K 0.0 0.0 34.1K 0.0 0.0 0.0
868.0K 0.0 0.0 658.0K 0.0 0.0 17.2
1514993360.454 483.4K 0.0 0.0 31.8K 0.0 0.0 0.0
1.4M 0.0 0.0 635.2K 0.0 0.0 22.2
1514993365.454 254.9K 0.0 0.0 25.1K 0.0 0.0 0.0
115.6K 0.0 0.0 317.2K 0.0 0.0 2.8
1514993370.454 822.4 0.0 0.0 1.6K 0.0 0.0 0.0
9.6K 0.0 0.0 14.8K 0.0 0.0 0.2
1514993375.454 2.3K 0.0 0.0 2.3K 0.0 0.0 0.0
720.5K 0.0 0.0 108.0K 0.0 0.0 0.2
1514993380.455 3.2K 0.0 0.0 2.8K 0.0 0.0 0.0
358.9K 0.0 0.0 210.6K 0.0 0.0 0.6
1514993385.454 1.0K 0.0 0.0 1.9K 0.0 0.0 0.0
1.1M 0.0 0.0 97.2K 0.0 0.0 2.8
1514993390.455 566.3 0.0 0.0 1.2K 0.0 0.0 0.0
433.5K 0.0 0.0 219.9K 0.0 0.0 1.0
bats(11.1-S)[19]
g1-252(11.1-S)[22] net_mon -e -w 5 -c 15
wlan0: 172.17.1.252/16 IEEE 802.11 Wireless Ethernet MCS mode 11ng
wlan0
Time In Out
Rate/sec Bits Errs Drops Bits Errs Drops Coll
1514993320.941 4.2K 0.0 0.0 0.0 5.8 0.0 0.0
1514993325.924 14.3K 0.0 0.0 0.0 12.0 0.0 0.0
1514993330.906 985.3K 0.0 0.0 160.6 55.6 0.0 0.0
1514993335.911 5.1M 0.0 0.0 0.0 246.9 0.0 0.0
1514993340.938 5.0M 0.0 0.0 0.0 241.9 0.0 0.0
1514993345.903 4.0M 0.0 0.0 0.0 184.3 0.0 0.0
1514993350.903 235.2K 0.0 0.0 0.0 62.6 0.0 0.0
1514993355.919 35.5K 0.0 0.0 0.0 51.2 0.0 0.0
1514993360.904 29.1K 0.0 0.0 0.0 45.3 0.0 0.0
1514993365.975 19.9K 0.0 0.0 0.0 21.5 0.0 0.0
1514993370.922 1.6K 0.0 0.0 0.0 1.6 0.0 0.0
1514993375.905 2.8K 0.0 0.0 0.0 2.8 0.0 0.0
1514993380.968 897.5 0.0 0.0 0.0 1.0 0.0 0.0
1514993386.035 1.7K 0.0 0.0 0.0 1.4 0.0 0.0
1514993390.957 1.6K 0.0 0.0 0.0 1.2 0.0 0.0
g1-252(11.1-S)[23]
Note that even for the "upload" phase of the test, the laptop is
showing negligible outbound *traffic* -- but a *lot* of outbound
"errors."
Can someone make sense of this?
Thanks!
Peace,
david
_______________________________________________
freebsd-wireless@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-wireless
To unsubscribe, send any mail to "freebsd-wireless-unsubscr...@freebsd.org"
