[LARTC] htb random weirdness

[Johannes Erdfelt <johannes@erdfelt.com>]

Today I setup htb on a system of mine that I needed to control bandwidth
for.

I read the documentation and setup a config that seems to be working
fine.

Unfortunately, there's a problem where randomly for between 1 and a
couple of seconds, almost no packets are sent out, or are delayed
significantly.

This is the output from a script I got from a friend. It does some
simple difference averaging with a 10 second period. Notice the period
of 0.13 Mb/s. That's not supposed to happen.

  eth0:    0.04 Mb/s In     1.68 Mb/s Out -    116.3 p/s In     180.7 p/s Out
  eth0:    0.04 Mb/s In     1.59 Mb/s Out -     82.1 p/s In     130.4 p/s Out
  eth0:    0.04 Mb/s In     1.60 Mb/s Out -    123.1 p/s In     184.8 p/s Out
  eth0:    0.01 Mb/s In     0.13 Mb/s Out -     33.3 p/s In      53.9 p/s Out
  eth0:    0.05 Mb/s In     1.53 Mb/s Out -     63.2 p/s In      89.6 p/s Out
  eth0:    0.05 Mb/s In     1.76 Mb/s Out -    119.4 p/s In     180.9 p/s Out

At the same time I ran a ping to the host as well:

64 bytes from xxx.xxx.xxx.20: icmp_seqH8 ttlX timeu.8 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqH9 ttlX timeu.6 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqI0 ttlX timeu.6 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqI1 ttlX timeÄ.4 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqI2 ttlX timeu.7 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqI3 ttlX time\x111 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqI5 ttlX time\x1370 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqI7 ttlX time&95 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqI9 ttlX timep0 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqP4 ttlX time\x135 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqP6 ttlX timeÅ.8 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqP7 ttlX timeu.4 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqP8 ttlX timeu.6 ms
64 bytes from xxx.xxx.xxx.20: icmp_seqP9 ttlX timeu.7 ms

I also ran a ping to a host sitting right next to it on the same switch.
It did not see the same increase in latency.

Load on the system was negligble at the time of the "blip".

Also, in my logs was this message:

HTB: mindelayP0, report it please !

Here's the jist of my config:

tc add dev eth0 root handle 1: htb default 12

tc class add dev eth0 parent 1: classid 1:1 htb \
	rate 2016kbit ceil 2016kbit
tc class add dev eth0 parent 1:1 classid 1:10 htb \
	rate 768kbit ceil 2016kbit prio 0
tc class add dev eth0 parent 1:1 classid 1:11 htb \
	rate 256kbit ceil 1920kbit prio 2
tc class add dev eth0 parent 1:1 classid 1:12 htb \
	rate 1024kbit ceil 2016kbit prio 1

tc filter add dev eth0 protocol ip parent 1:0 prio 1 u32 \
        match ip tos 0x10 0xff flowid 1:10
tc filter add dev eth0 protocol ip parent 1:0 prio 1 u32 \
        match ip protocol 1 0xff flowid 1:10
tc filter add dev eth0 protocol ip parent 1:0 prio 1 u32 \
        match ip src xxx.xxx.xxx.119 flowid 1:11

tc qdisc add dev eth0 parent 1:10 handle 20: sfq perturb 10
tc qdisc add dev eth0 parent 1:11 handle 21: sfq perturb 10
tc qdisc add dev eth0 parent 1:12 handle 22: sfq perturb 10

Anyone have any idea what be causing the "blip"? It's noticable because
for that period of time, all of my ssh sessions become very slow and
laggy, and then everything speeds up again. In fact, it happened as I
was writing this email too.

JE

_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/