[Click] Packet loss with PollDevice
rchertov
rchertov at cs.ucsb.edu
Thu Jan 19 14:09:43 EST 2012
On 01.19.2012 11:06, Yu Chen wrote:
> 1.8.0 on Linux 2.6.24.7. Yes, I played with it(from 1 to 8). However,
Have you tried higher values like 16 and 32? Also, since you are using
Intel Pro NICs, you can use ethtool to increase the RX ring buffer.
ethtool -G ethX rx <SomeNumber>
Roman
> it does not seem to help much. Yes, polldevice::count is smaller than
> the packets I sent to that interface. Also, the packet size 64B. I
> forgot
> to mention it in the previous emails.
>
> Best Regards and Thanks,
> Yu
>
> On Thu, Jan 19, 2012 at 10:59:14AM -0800, rchertov wrote:
>> What version of click are you running? Also, you can play around
>> with
>> BURST parameter to PollDevice. So are you observing that
>> polldevice::count is smaller than the number of packets you sent to
>> that
>> interface?
>>
>> Roman
>>
>> On 01.19.2012 10:45, Yu Chen wrote:
>> > Hi all,
>> >
>> > I am having an experiment with the following topology running on
>> > Deterlab. I
>> > am having a 4-core Xeon and modified e1000 driver running ore Xeon
>> > and modified
>> > e1000 driver running. The memory is 4GB.
>> >
>> > User1
>> > |
>> > User3-----Router------User4
>> > |
>> > User2
>> >
>> > User 1 is sending packet to User 2; User 3 is sending packets to
>> User
>> > 4.
>> > The router only routes stuff with best effort. However, what I
>> > observed is
>> > very few losses, distributed on a small random set of rounds. What
>> is
>> > surprising is that there are losses on one particular netcard when
>> > the sending
>> > rate is only 125,000 packets/second. The loss happens at the
>> > receiving phase.
>> >
>> > Below is what I have tried:
>> > 1' I am using 4 threads, each corresponding to one pair of
>> PollDevice
>> > and
>> > ToDevice.
>> > 2' I have changed the queue length at the interface cards. I set
>> > txqueuelen=1
>> > for the senders to decrease the possible burstiness. By increasing
>> > the
>> > receiving queue length from 64 to 4096
>> > (the maximum queue length), the loss rate decreases.
>> > 3' I tried to discard the packets on their arrival at the router.
>> In
>> > this case,
>> > there is no observed loss.
>> >
>> > I am kind of lost at the point. Any suggestions on what can be the
>> > possible
>> > reason and what I can try next?
>> >
>> > Attached is my code.
>> >
>> > Best Regards,
>> > Yu
>> >
>> > Router:
>> >
>> >
>> -----------------------------------------------------------------------
>> > ddressInfo(router1-w1 10.0.1.1 00:15:17:5d:16:4a,
>> > //eth2
>> > user1-w1 10.0.1.2 00:15:17:5d:29:28, //eth0
>> > router1-w2 10.0.2.1 00:15:17:5d:16:48, //eth0
>> > user2-w2 10.0.2.2 00:15:17:57:ce:14, //eth0
>> > router1-w3 10.0.3.1 00:15:17:5d:16:4b, //eth3
>> > user3-w3 10.0.3.2 00:15:17:5d:27:e8, //eth0
>> > router1-w4 10.0.4.1 00:15:17:5d:16:49, //eth1
>> > user4-w4 10.0.4.2 00:15:17:5d:2a:3c, //eth0
>> > );
>> >
>> > c1 :: Classifier(12/0806 20/0001,
>> > 12/0806 20/0002,
>> > 12/0800,
>> > -);
>> > c2 :: Classifier(12/0806 20/0001,
>> > 12/0806 20/0002,
>> > 12/0800,
>> > -);
>> > c3 :: Classifier(12/0806 20/0001,
>> > 12/0806 20/0002,
>> > 12/0800,
>> > -);
>> > c4 :: Classifier(12/0806 20/0001,
>> > 12/0806 20/0002,
>> > 12/0800,
>> > -);
>> >
>> > q0 :: Discard;
>> > // Idle -> ToDevice(eth1);
>> > // Idle -> ToDevice(eth3);
>> > // Idle -> ToDevice(eth2);
>> > // Idle -> ToDevice(eth0);
>> > q1 :: CPUQueue(1000000) -> EtherEncap(0x0800, router1-w1,
>> user1-w1)
>> > -> td1 :: ToDevice(eth2);
>> > q2 :: CPUQueue(1000000) -> EtherEncap(0x0800, router1-w2,
>> user2-w2)
>> > -> td3 :: ToDevice(eth0);
>> > q3 :: CPUQueue(1000000) -> EtherEncap(0x0800, router1-w3,
>> user3-w3)
>> > -> td2 :: ToDevice(eth3);
>> > q4 :: CPUQueue(1000000) -> EtherEncap(0x0800, router1-w4,
>> user4-w4)
>> > -> td0 :: ToDevice(eth1);
>> >
>> > rt :: LookupIPRouteMP(10.0.1.0/32 0, 10.0.1.1/32 0, 10.0.1.255/32
>> 0,
>> > 10.0.2.0/32 0, 10.0.2.1/32 0, 10.0.2.255/32
>> 0,
>> > 10.0.3.0/32 0, 10.0.3.1/32 0, 10.0.3.255/32
>> 0,
>> > 10.0.4.0/32 0, 10.0.4.1/32 0, 10.0.4.255/32
>> 0,
>> > 10.0.1.0/24 1, 10.0.2.0/24 2, 10.0.3.0/24 3,
>> > 10.0.4.0/24 4, 0.0.0.0/0 0);
>> > rt[0] -> Discard;
>> > rt[1] -> q1;
>> > rt[2] -> q2;
>> > rt[3] -> q3;
>> > rt[4] -> q4;
>> >
>> > pd1 :: PollDevice(eth2) -> c1;
>> > c1[0] -> q0;
>> > c1[1] -> q0;
>> > c1[2] -> DropBroadcasts -> Strip(14) -> GetIPAddress(16) -> rt;
>> > c1[3] -> Discard;
>> >
>> > pd3 :: PollDevice(eth0) -> c2;
>> > c2[0] -> q0;
>> > c2[1] -> q0;
>> > c2[2] -> DropBroadcasts -> Strip(14) -> GetIPAddress(16) -> rt;
>> > c2[3] -> Discard;
>> >
>> > pd2 :: PollDevice(eth3) -> c3;
>> > c3[0] -> q0;
>> > c3[1] -> q0;
>> > c3[2] -> DropBroadcasts -> Strip(14) -> GetIPAddress(16) -> rt;
>> > c3[3] -> Discard;
>> >
>> > pd0 :: PollDevice(eth1) -> c4;
>> > c4[0] -> q0;
>> > c4[1] -> q0;
>> > c4[2] -> DropBroadcasts -> Strip(14) -> GetIPAddress(16) -> rt;
>> > c4[3] -> Discard;
>> >
>> > StaticThreadSched(pd2 0, td0 0, pd0 1, td2 1, pd3 2, td1 2, pd1 3,
>> > td3 3);
>> >
>> > User:
>> > //
>> > AddressInfo(router1-w1 10.0.1.1 00:15:17:5d:16:4a,
>> > //eth2
>> > user1-w1 10.0.1.2 00:15:17:5d:29:28, //eth0
>> > router1-w2 10.0.2.1 00:15:17:5d:16:48, //eth0
>> > user2-w2 10.0.2.2 00:15:17:57:ce:14, //eth0
>> > router1-w3 10.0.3.1 00:15:17:5d:16:4b, //eth3
>> > user3-w3 10.0.3.2 00:15:17:5d:27:e8, //eth0
>> > router1-w4 10.0.4.1 00:15:17:5d:16:49, //eth1
>> > user4-w4 10.0.4.2 00:15:17:5d:2a:3c, //eth0
>> > );
>> > //Send
>> > udp :: FastUDPSource(RATE 400000, LIMIT 20000000, LENGTH 78,
>> > SRCETH user1-w1, SRCIP user1-w1, SPORT 1,
>> > DSTETH router1-w1, DSTIP user2-w2, DPORT 1,
>> > CHECKSUM false, INTERVAL 0, ACTIVE false)
>> > -> ToDevice(eth0);
>> >
>> > s :: Script(TYPE ACTIVE, write ctr.reset, write udp.active false,
>> > wait 1, wait 9, write udp.active true);
>> >
>> >
>> > _______________________________________________
>> > click mailing list
>> > click at amsterdam.lcs.mit.edu
>> > https://amsterdam.lcs.mit.edu/mailman/listinfo/click
>>
>> _______________________________________________
>> click mailing list
>> click at amsterdam.lcs.mit.edu
>> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
More information about the click
mailing list