Thank you for your answer.

Does OVS respect the queues issued with tc? I have not tested it yet, but pfifo_fast could be in effect if type= is omitted.

Fabian

On 03/23/2016 04:32 PM, Guru Shetty wrote:


On 23 March 2016 at 03:15, Fabian Schörghofer
<fabian.schoergho...@salzburgresearch.at
<mailto:fabian.schoergho...@salzburgresearch.at>> wrote:

    I want to create QoS flows.

    The examples on the Internet that I can find realize this by using
    the type=linux-htb in the creation, like the command below:

    ovs-vsctl set port p1p1 qos=@newqos -- --id=@newqos create qos
    type=linux-htb other-config:max-rate=1000000000
    queues:1=@eth04queue1 queues:2=@eth04queue2 -- --id=@eth04queue1
    create queue other-config:max-rate=10000000 -- --id=@eth04queue2
    create queue other-config:max-rate=20000000

    This will create a HTB scheduler as can be seen from this:

    5: p1p1: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1500 qdisc
    htb state UP mode DEFAULT group default qlen 1000
         link/ether 68:05:ca:39:da:bc brd ff:ff:ff:ff:ff:ff

    What other schedulers can I use for type? I want to use the
    pfifo_fast scheduler, but when I enter the first command with
    type=pfifo_fast or type=linux-pfifo_fast the result from ip link
    show will not change.

    This leads me to believe that type=pfifo_fast or linux-pfifo_fast
    has no effect.

    On some Linux versions pfifo_fast is the default queueing mechanism,
    some newer versions use mq, but that might not be relevant here.

    Is it documented somewhere what strings I can use for those mechanisms?


Documentation is at 'man ovs-vswitchd.conf.db'. Search for "QoS TABLE"


    Best Regards
    Fabian
    _______________________________________________
    discuss mailing list
    discuss@openvswitch.org <mailto:discuss@openvswitch.org>
    http://openvswitch.org/mailman/listinfo/discuss


_______________________________________________
discuss mailing list
discuss@openvswitch.org
http://openvswitch.org/mailman/listinfo/discuss

Reply via email to