I'm limiting the network bandwidth to a lan (on eth1) with the cbq.init
(v0.3alpha2 script. I'm using linux 2.2.10 on a Pentium 133 system,
iproute2-ss19990630 and a 4 port DEC Tulip based network card. The
clocksource is defined to PSCHED_JIFFIES. I want to give the workstations
on eth1 full speed to the eth0 lan and 256Kbit to the rest of the world.
eth1 is configured to have a private IP adress and eth1:0 is configured
with some real IP adresses.
tc -d qdisc ls dev eth1 gives:
qdisc tbf 8039: rate 256Kbit burst 10Kb/8 mpu 0b lat 190.7ms
qdisc tbf 8038: rate 10Mbit burst 10Kb/8 mpu 0b lat 4.8ms
qdisc cbq 11: rate 10Mbit cell 8b (bounded,isolated) prio no-transmit/8 weight 10Mbit
allot 1514b
level 2 ewma 5 avpkt 1000b maxidle 23us
tc -d class show dev eth1:
class cbq 11: root rate 10Mbit cell 8b (bounded,isolated) prio no-transmit/8 weight
10Mbit allot 1514b
level 2 ewma 5 avpkt 1000b maxidle 23us
class cbq 11:1 parent 11: rate 10Mbit cell 8b prio no-transmit/8 weight 1Mbit allot
1514b
level 1 ewma 5 avpkt 1000b maxidle 23us
class cbq 11:1270 parent 11:1 leaf 803e: rate 10Mbit cell 8b (bounded) prio 4/4 weight
1Mbit allot 1514b
level 0 ewma 5 avpkt 1000b maxidle 23us
class cbq 11:1271 parent 11:1 leaf 803f: rate 256Kbit cell 8b (bounded) prio 5/5
weight 26Kbit allot 1514b
level 0 ewma 5 avpkt 1000b maxidle 26390us
tc -d filter show dev eth1:
filter parent 11: protocol ip pref 100 u32
filter parent 11: protocol ip pref 100 u32 fh 800: ht divisor 1
filter parent 11: protocol ip pref 100 u32 fh 800::800 order 2048 key ht 800 bkt 0
flowid 11:1270
match XXXXXX00/ffffff00 at 12
filter parent 11: protocol ip pref 100 u32 fh 800::801 order 2049 key ht 800 bkt 0
flowid 11:1271
match 00000000/00000000 at 16
flowid 11:1270 is matching the source IP numbers from the machines on
the eth0 lan.
It all seem to work ok, but after a while (1-3 hours), the backlog starts
to fill up and all trafic stops (not a single packet comes out on the
interface). Taking down the interface and up again or running
cbq stop; cbq start makes everything work correct again.
Is this a bug or have I simply missconfigured something?
If the eth1 interface was running in 100Mbps mode, could that be the
reason for this?
/Sebastian
See http://www.hogia.net/keys/sa-pgp.asc for public pgp key.
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
