Hi anyone
I try to use htb and got some strange situation
My hbt config is root htb with default to 999 class
Then I make 2 classes 999 and 998
And make many many (near 60) leafs attached to 998
root
! !
999 998 ---...---> many leafs
root attached to ethernet dotQ subinterface
999 make rate where high (50Mbit) but according to
tc -s class show dev <intf> traffic at this class extremally
small (near 9,6Kbit) so any other traffic going throw 998
class
On 998 class I make rate X Mbit''s (without.same ceil)
Than I make many many leafs with it''s own rate/ceil
Summary of rate (and ceil) is biggest then X
I think that I have 3 level of queue at my config
High level - root htb (unused because speed too high)
Second level - queue at classes 998 and 999
And next level - queue and any leaf
But I can''t see queuning at class 998 even if traffic more
then X Mbits
I think I got priority queue (I set prio at nex level
classes) but nothing I got and at interface I got high speed
that X Mbits
Why? Could I do something to get such config?
M-m-m and some another information. To any leaf I connect
sfq qdisc. To 998 I doesn''t connect anything. May be I need
to connect prio queuening discipline?
Also I want to ask another question
If I make filter like 1:1 to parent 1: and than make two
another filters 1:2 and 1:3 with parent 1:1 is I got logical
and between 1:1 and 1:2/1:3?
--
Best regard, Aleksander Trotsai aka MAGE-RIPE aka MAGE-UANIC
My PGP key at ftp://blackhole.adamant.ua/pgp/trotsai.key[.asc]
Big trouble - Someone else stole your IP address, call the Internet detectives!
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Hi All, I am runnnig a small network of Linux-routers.In this, I want to measurre some traffic performance. For this I have to make sure that clock of all the routers is synced upto millisecond or better. To achive this, I did "ntpdate <server >".(ntp syncing) After doing this I noticed that my nodes are no synced accurately. In my testing I found that receive time was less than transit time by some milli/micro seconds. Can any one plaese suggest me how can I achive micro(may be milli) second or better syncing. Thanks, Nitin __________________________________________________ Do you Yahoo!? U2 on LAUNCH - Exclusive greatest hits videos http://launch.yahoo.com/u2 _______________________________________________ LARTC mailing list / LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Download xntpd. It also has documents attached which explains what precition CAN be reached by regylar PC. devik On Tue, 12 Nov 2002, nitin panjwani wrote:> Hi All, > > I am runnnig a small network of Linux-routers.In this, > I want to measurre some traffic performance. For this > I have to make sure that clock of all the routers is > synced upto millisecond or better. > > To achive this, I did "ntpdate <server >".(ntp > syncing) > > After doing this I noticed that my nodes are no synced > accurately. In my testing I found that receive time > was less than transit time by some milli/micro > seconds. > > Can any one plaese suggest me how can I achive > micro(may be milli) second or better syncing. > > Thanks, > Nitin > > > > > __________________________________________________ > Do you Yahoo!? > U2 on LAUNCH - Exclusive greatest hits videos > http://launch.yahoo.com/u2 > _______________________________________________ > LARTC mailing list / LARTC@mailman.ds9a.nl > http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/ > >_______________________________________________ LARTC mailing list / LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/