Is it normal to have dropped packets on the interface that SQM creates? What are the reasons for this?
ifb4pppoe-wan Link encap:Ethernet HWaddr XX:XX:XX:XX:XX:XX
inet6 addr: fe80::xxxx:xxxx:xxxx:xxxx/64 Scope:Link
UP BROADCAST RUNNING NOARP MTU:1500 Metric:1
RX packets:5948130 errors:0 dropped:23 overruns:0 frame:0
TX packets:5948106 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:32
RX bytes:6514925414 (6.0 GiB) TX bytes:6514922312 (6.0 GiB)
fuller
March 4, 2018, 6:56pm
2
i'd say it is not normal and suspect some kind of overflow (packets coming in too fast)
Not likely, the router is under very little use.
fuller
March 4, 2018, 8:31pm
4
yea, it is a small amount...
you could monitor the number and do some abusive tests or just wait and see if increases correlate with any activity.
Dropping packets is how SQM controls bufferbloat.
Read the first paragraph in this article...
https://www.bufferbloat.net/projects/bloat/wiki/More_about_Bufferbloat/
My understanding is that drops from qdisc like sqm don't show up in those stats.
I've never seen dropped packets on the interface even after a speed test.
Neither did I when running the 17.01. I think it started when I installed master . But I cannot be sure nor I can say if it is affecting anything.
fuller
March 5, 2018, 5:58am
9
could be unhandled packets?
try restarting the connection and tcpdump it
@jwoods not like this
Do you mean restarting the sqm interface?
fuller
March 5, 2018, 2:57pm
12
the pppoe connection and yes (implicitly) the ifb.
thought is, those 23 pkts could also be some discover-something at startup.
but i really suggest you try to saturate the link with small packets and see if this increases drops. eg:
# iperf -c <dest-ip> -u -b1M -P<bw-in-megabit> -l100
(from another machine)