[Bloat] Possibly FreeBSD's Codel is misconfigued

[Dave Taht]

Sigh. No matter how hard we tried for something almost impossible to
configure wrong, it happens.

1) A target needs to be bigger than the time it takes for a single
full size MTU. E.g. at 1mbit, target should be about 13ms on 1500 byte
mtu.


on that thread:
is this wrong?

altq on em0 codelq( qlimit 5, interval 105 ) bandwidth 630Kb queue { root_em0 }
Edit: To clarify, qlimit = target.

2) There are people there complaining about ping loss in codel under
load. That's expected. (with or without an appropriate target). It
shows it's working.

3) With hfsc, however, sparser flows are prioritized, so you would
probably see less ping loss.


On Thu, Jun 18, 2015 at 5:03 PM, Benjamin Cronce <bcronce at gmail.com> wrote:
> https://forum.pfsense.org/index.php?topic=88162.msg519194#msg519194
>
> The source code looks like it is set for a target of 100 and an interval of
> 5. I use PFSense and CoDel is working great for me right now. Pretty much
> 0ms of bufferbloat on the DSLReports tests. If it actually is misconfigured,
> it's working great for me.
>
> codel_alloc(int target, int interval, int ecn)
> cl->cl_codel = codel_alloc(100, 5, 0);
>
> _______________________________________________
> Bloat mailing list
> Bloat at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat
>



-- 
Dave Täht
worldwide bufferbloat report:
http://www.dslreports.com/speedtest/results/bufferbloat
And:
What will it take to vastly improve wifi for everyone?
https://plus.google.com/u/0/explore/makewififast