* [LARTC] Confusion about MTU
@ 2002-12-13 18:54 Daniel
2002-12-14 15:21 ` Daniel
0 siblings, 1 reply; 2+ messages in thread
From: Daniel @ 2002-12-13 18:54 UTC (permalink / raw)
To: lartc
Hi!
This is acctually my first mail to the list, I never really
planned on joining but I could find any answers so, well...
I'm kind of a perfecttionist... or rather I like to play around
alot with stuff so that they work the best that they possibly can.
I know most of the stuff about HTB and IMQ and the setup I use now
works great, not to say that there aren't room for improvement, there
always is. Anyways I'm really enjoying the setup, finally I can play
online games not having to constantly worrying about my siblings destroying
my game using som filesharing app or whatever.
Ok, this is my question : What should I consider to be the maximum
packetsize
inside an IMQ and using the HTB queue?
I'm thinking 1500b i.e without the 14b ethernet encap, small proof of this
is that
an sfq queue withing the IMQ device defaults it's quantum to 1500b instead
of the
1514b as it does if attached to a real ethernet device. This could also be a
bug, dunno.
I want to use the smallest possible value for burst, cburst and quantum, so
what
would that be? Note I'm shaping traffic that is small enough that setting
those settings to the mtu wont affect troughoutput.
//Daniel
ps. Sorry if the mail got a little messy, I wrote it in a hurry
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [LARTC] Confusion about MTU
2002-12-13 18:54 [LARTC] Confusion about MTU Daniel
@ 2002-12-14 15:21 ` Daniel
0 siblings, 0 replies; 2+ messages in thread
From: Daniel @ 2002-12-14 15:21 UTC (permalink / raw)
To: lartc
Ok, I found it out by myself, it is 1514b.
I set up two length matches in the PREROUTING chain in the mangle table, one
set to match on packets >\x1500b and one >\x1501b. The latter rule never
matched so I thought that the max packet really was 1500b. However I then
decided to put all packets matching the first rule in a separate class.
From there the answer was obvious, I simply divided the byte-count with the
packet-count and voila, I got 1514. So I guess the length match doesn't
count for the 14b's of ethernet encap.
This led me to change another thing in my setup, I no longer match small TCP
packets by matching them by <db but rather <Pb.
Well... this prolly seem real newbie like to most of you but hey we all are
at first.
// Daniel
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2002-12-14 15:21 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2002-12-13 18:54 [LARTC] Confusion about MTU Daniel
2002-12-14 15:21 ` Daniel
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.