[ovs-discuss] OvS flow setup time.

We use the ovs_lock/ovs_unlock when we new a flow(ovs_flow_cmd_new) in the kernel datapath. May this affect the flow setup time ? because there may be multiple upcall handler threads. Another question, how can I test the flow setup time. The ovs-benchmark has been removed. Thanks. Nick

Re: [ovs-discuss] in mininet

Hi, You can install mininet without OvS and install OvS from git or tar.gz 1. install mininet without OvS mininet/util/install.sh -nf 2. install OvS http://docs.openvswitch.org/en/latest/intro/install/ You should remove the old OvS from yo

Re: [ovs-discuss] [openvswitch 2.7.90] testsuite: 2180 failed

Yes, thanks for your report. I try to find what might cause this. Thanks. Nick > On Mar 21, 2017, at 7:12 AM, Joe Stringer wrote: > > Hi Nick, > > Looks like there's another report of trouble with this test below. > Would you be able to take a look? ___

Re: [ovs-discuss] Bonding slave disabled

> On Feb 15, 2017, at 1:37 PM, Neelakantam Gaddam wrote: > > Hi Nick, > > When I do "ifconfig br0 up", interestingly the bond slave ports are getting > enabled including may_enable. > > This means that, port_run is not running before bond_run. port_run function > runs when there is a change

Re: [ovs-discuss] Bonding switches using Mininet

100mbps, 1 per host --> switch_1 <--- 4 > links of 100mbps > switch_2 <- 4 links of 100mbps, 1 per host > --> 4 hosts > > > What is the right way to configure a bond switch_1 and switch_2? > > On 13 February 2017 at 06:24, nickcooper-zhangt

Re: [ovs-discuss] Bonding switches using Mininet

> On Feb 12, 2017, at 11:26 PM, Ronaldo Resende wrote: > > ronaldo@localhost:~$ sudo ovs-vsctl add-bond s1 bond0 s1-eth9 s1-eth10 > ovs-vsctl: cannot create an interface named s1-eth9 because a port named > s1-eth9 already exists on bridge s1 > > Any other thoughts? The s1-eth9 and s1-eth10 h

Re: [ovs-discuss] Bonding slave disabled

> On Feb 12, 2017, at 1:05 PM, nickcooper-zhangtonghao > wrote: > > If you use the virtual ethernet driver, you should set the type of net > device. If not, the “system” will be used. > > For example, tap device > > ovs-vsctl add-br br0 > ifconfig tap1 up >

Re: [ovs-discuss] Bonding slave disabled

If you use the virtual ethernet driver, you should set the type of net device. If not, the “system” will be used. For example, tap device ovs-vsctl add-br br0 ifconfig tap1 up ifconfig tap2 up ovs-vsctl add-bond br0 bond0 tap1 tap2 bond_mode=active-backup ovs-vsctl set interface tap1 type=tap ov

Re: [ovs-discuss] Bonding switches using Mininet

> On Feb 12, 2017, at 8:20 AM, Ronaldo Resende wrote: > > ronaldo@localhost:~$ sudo ovs-vsctl add-bond s1 bond0 s1-eth9 s1-eth10 > ovs-vsctl: Error detected while setting up 's1-eth9', 's1-eth10'. See > ovs-vswitchd log for details. > > In both cases I tried doing this creating and without cr

[ovs-discuss] How to get started on OpenFlow implementation.

How can I get started on OpenFlow implementation in OvS. I reviewed the dpcls and EMC codes but I don’t know how to get started on ofproto table and OpenFlow implementation. Can you help me? Thanks. Nick ___ discuss mailing list disc...@openvswitch.org

Re: [ovs-discuss] cycles_count_xxx CPU overhead

Thanks very much. Got it. Thanks. Nick > On Jan 16, 2017, at 12:31 PM, Daniele Di Proietto wrote: > > 2017-01-12 23:47 GMT-08:00 nickcooper-zhangtonghao <mailto:n...@opencloud.tech>>: >> The dp_netdev_process_rxq_port() will use the “cycles_count_start” and >> “

[ovs-discuss] cycles_count_xxx CPU overhead

The dp_netdev_process_rxq_port() will use the “cycles_count_start” and “cycles_count_end” for counting the cycles. But via perf tool, I found that cycles_count_xxx uses a lot of CPU. One question to ask: there is one thread to use the pmd struct to processing the packets, right? If so, pmd shoul

Re: [ovs-discuss] how many queues in openvswitch?

No, the queues of netdev(e.g tap, internal, system) are part of linux, but you can only set the queues num for dpdk. More info, see manpage. Thanks. Nick > On Jan 13, 2017, at 9:27 AM, 杨润垲 <16125...@bjtu.edu.cn> wrote: > > yes ,i know, that is an example. Is there any queues in ovs like rx dma

[ovs-discuss] internal dev mtu

The upstream net-next has removed the “internal_dev_change_mtu”. It will be ok for ovs(e.g. mtu_request) ? I think that code should be keep. Thanks. Nick ___ discuss mailing list disc...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ov

Re: [ovs-discuss] MTU on ovs bridge

Hi Vijay, The section “How can I configure the bridge internal interface MTU? “ and the ovs-vswitchd manpage may help you. https://github.com/openvswitch/ovs/blob/master/Documentation/faq/issues.rst Thanks. Nick > On