We've experienced the same problem during our testing. Applying the following patch to DPDK seemed to solve the issue:
http://dpdk.org/ml/archives/dev/2015-May/017919.html On 26/05/2015 01:32, "钢锁0310" <l...@dtdream.com> wrote: >i find a mail >http://lists.freebsd.org/pipermail/freebsd-net/2015-January/040812.html > > >In that mail,virio drive add 12byte, > >For instance, your >guest kernel can offload the TCP checksum to the virtio-net device, >which in turn uses the virtio-net header (that requires TAP driver >support) to postpone the checksum to the host kernel. If packets >arrive to a physical NIC that supports checksum offloading (e.g. a >r8169 NIC attached to the same bridge to which the TAP is attached), >you have effectively offloaded the checksum computation from the guest >kernel straight to the physical NIC in the physical host. >In my experiment the tcp packet from vm to host has be added 12bytes, >so host can't interpret itit looks like the same Does the vhost-cuse >consider the 12bytes?How should i do for skip the 12bytes?udp id ok only >tcp is bad >********************* >RTFSC >********************* > > > >------------------------------------------------------------------ >发件人：钢锁0310 <l...@dtdream.com> >发送时间：2015年5月21日(星期四) 14:47 >收件人：gowrishankar <gowrishanka...@linux.vnet.ibm.com>，discuss ><discuss@openvswitch.org>，通天晓0280 <do...@dtdream.com> >主 题：答复：答复：答复：[ovs-discuss] 答复： For help about ssh between vms through >ovs-dkdp > > >sorry my fault > > >tap0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500 > inet6 fe80::c469:e6ff:fe78:d30f prefixlen 64 scopeid 0x20<link> > ether c6:69:e6:78:d3:0f txqueuelen 500 (Ethernet) > RX packets 0 bytes 0 (0.0 B) > RX errors 0 dropped 0 overruns 0 frame 0 > TX packets 0 bytes 0 (0.0 B) > TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0 > >tap1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500 > inet6 fe80::3c73:6cff:fe7a:49c3 prefixlen 64 scopeid 0x20<link> > ether 3e:73:6c:7a:49:c3 txqueuelen 500 (Ethernet) > RX packets 0 bytes 0 (0.0 B) > RX errors 0 dropped 0 overruns 0 frame 0 > TX packets 0 bytes 0 (0.0 B) > TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0 > > > >no packet stat in tap0 and tap1 neither to tcpdump -vvv -ni tap0 > > >but ping is work > > >[root@25 ~]# ovs-ofctl dump-flows br0 >NXST_FLOW reply (xid=0x4): > cookie=0x0, duration=536.513s, table=0, n_packets=207, n_bytes=18734, >idle_age=0, in_port=1 actions=output:2 > cookie=0x0, duration=536.512s, table=0, n_packets=200, n_bytes=18440, >idle_age=0, in_port=2 actions=output:1 > cookie=0x0, duration=551.217s, table=0, n_packets=8, n_bytes=648, >idle_age=434, priority=0 actions=NORMAL >[root@25 ~]# ovs-ofctl dump-ports br0 >OFPST_PORT reply (xid=0x2): 3 ports > port LOCAL: rx pkts=8, bytes=648, drop=0, errs=0, frame=0, over=0, crc=0 > tx pkts=0, bytes=0, drop=0, errs=0, coll=0 > port 1: rx pkts=213, bytes=?, drop=?, errs=?, frame=?, over=?, crc=? > tx pkts=214, bytes=?, drop=0, errs=?, coll=? > port 2: rx pkts=206, bytes=?, drop=?, errs=?, frame=?, over=?, crc=? > tx pkts=199, bytes=?, drop=22, errs=?, coll=? > > > >the stat of ovs is ok > > >********************* >RTFSC >********************* > > >------------------------------------------------------------------ >发件人：gowrishankar <gowrishanka...@linux.vnet.ibm.com> >发送时间：2015年5月21日(星期四) 14:26 >收件人：钢锁0310 <l...@dtdream.com> >主 题：Re: 答复：答复：[ovs-discuss] 答复： For help about ssh between vms through >ovs-dkdp > > >I think this is same as what you shared earlier in ML i.e captured from >inside VM2. >Or correct me if I am wrong. > > >I wanted to see running tcpdump -vvv -ni vnet0 from host (for eg, vnet0 >is vnic of guest VM1 >and connected to your ovs-dpdk switch as I hope). > > >Could that be possible. You may please include ML in CC as well, if >others also could help if some more >traces there this time. > > >Regards, >Gowrishankar > > >On Thursday 21 May 2015 11:43 AM, 钢锁0310 wrote: > > > > >the screen picture above > > >********************* >RTFSC >********************* > > > >------------------------------------------------------------------ >发件人：gowrishankar ><gowrishanka...@linux.vnet.ibm.com> ><mailto:gowrishanka...@linux.vnet.ibm.com> >发送时间：2015年5月21日(星期四) 14:08 >收件人：钢锁0310 ><l...@dtdream.com> <mailto:l...@dtdream.com> >主 题：Re: 答复：[ovs-discuss] 答复： For help about ssh between vms through >ovs-dkdp > > >May be, you could share tcpdump -vvv for us to infer anything else (from >host). > > >Regards, >Gowrishankar > > >On Thursday 21 May 2015 11:31 AM, 钢锁0310 wrote: >the tcp state on vm which run ssh is SYNSEND > > >VMs`csum is off > > >the target vm catch wrong packet still > > >********************* >RTFSC >********************* > > > >------------------------------------------------------------------ >发件人：gowrishankar ><gowrishanka...@linux.vnet.ibm.com> ><mailto:gowrishanka...@linux.vnet.ibm.com> >发送时间：2015年5月21日(星期 四) 13:55 >收件人：钢锁0310 ><l...@dtdream.com> <mailto:l...@dtdream.com> >抄 送：discuss ><discuss@openvswitch.org> <mailto:discuss@openvswitch.org> >主 题：Re: [ovs-discuss] 答复： For help about ssh between vms through ovs-dkdp > > >tcpdump from host on guest's vnic could help ? See if 3way handshake >completes. > > >Also, check if csum correct on packets from guest (tcpdump -vvv). > > >Regards, >Gowrishankar > > >On Thursday 21 May 2015 10:37 AM, 钢锁0310 wrote: >ovs log: >2015-05-21T04:46:37.941Z|00023|connmgr|INFO|br0<->unix: 1 flow_mods in >the last 0 s (1 adds) >2015-05-21T04:46:38.604Z|00024|connmgr|INFO|br0<->unix: 1 flow_mods in >the last 0 s (1 adds) >2015-05-21T04:48:42.692Z|00001|dpdk(cuse_thread3)|INFO|vHost Device >'tap0' (0) has been added >2015-05-21T04:48:51.750Z|00002|dpdk(cuse_thread3)|INFO|vHost Device >'tap1' (1) has been added > > > >qemu create tap with default name 'tap[id]' adding port to ovs is >ok ping is work > > >and i add options >'csum=off,gso=off,guest_csum=off,guest_tso4=off,guest_tso6=off,guest_ecn=o >ff' > > >in guset >ethtool -k eth0 > > >all things is off but > > >ssh is still not work > > >qemu 2.3.0 >dpdk 2.0.0 >ovs compiled by git master > > >the OS of host and guest both are 'Linux 25.dpdk 3.10.0-229.el7.x86_64 >#1 SMP Fri Mar 6 11:36:42 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux' > > >********************* >RTFSC >********************* > > > >------------------------------------------------------------------ >发件人：Traynor, Kevin ><kevin.tray...@intel.com> <mailto:kevin.tray...@intel.com> >发送时间：2015年5月 20日(星期 三) 22:21 >收件人：Daniele Di Proietto ><diproiet...@vmware.com> <mailto:diproiet...@vmware.com>， 钢锁0310 ><l...@dtdream.com> <mailto:l...@dtdream.com> >抄 送：通天晓0280 ><do...@dtdream.com> <mailto:do...@dtdream.com>，discuss ><discuss@openvswitch.org> <mailto:discuss@openvswitch.org> >主 题：RE: [ovs-discuss] For help about ssh between vms through ovs-dkdp > > > > >> -----Original Message----- >> From: Daniele Di Proietto [mailto:diproiet...@vmware.com] >> Sent: Wednesday, May 20, 2015 1:47 PM >> To: 钢锁 0310 >> Cc: 通天晓 0280; discuss; Traynor, Kevin >> Subject: Re: [ovs-discuss] For help about ssh between vms through >>ovs-dkdp >> >> This might be related to offloading features. >> >> Could you try again with this qemu "-device" option and let us know? >> -device >> >>virtio-net-pci,netdev=net1,csum=off,gso=off,guest_tso4=off,guest_tso6=off >>,g >> uest_ecn=off > > >I checked this previously and the DPDK vhost lib will report those >features >as not available during negotiation so you should be ok with not >specifying >them explicitly in the qemu cmd line. I haven't tested with qemu 2.3. > > >> >> >> Kevin, do you this we should mention this in INSTALL.DPDk.md? > > >I had changed the original patch to make it optional in 3a based on >feedback >from Michael Tsirkin and testing. So unless we find an issue, I'd prefer >to >leave optional. >https://github.com/openvswitch/ovs/blob/master/INSTALL.DPDK.md#dpdk-vhost- >vm-configuration ><https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_openvswit >ch_ovs_blob_master_INSTALL.DPDK.md-23dpdk-2Dvhost-2Dvm-2Dconfiguration&d=B >QMFaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=SmB5nZacmXNq0gKCC1s_ >Cw5yUNjxgD4v5kJqZ2uWLlE&m=uyMUTwlAFqpkT9D3pmPQi1LFOOG-ByD1htMGkp5eXLc&s=Ob >AhVG4R4gABevsLZVIKFuJ1snNlK84wH5JIFHmMX0o&e=> > > >> >> >> On 20/05 /2015 10:39, "钢锁0310" <l...@dtdream.com> >><mailto:l...@dtdream.com> wrote: >> >> >i build the vm-ovsdpdk-vm environment by using vhost-cuse >> > >> > >> >the cmd is following >> > >> > >> >ovs-vsctl add-br br0 -- set bridge br0 datapath_type=netdev >> >ovs-vsctl add-port br0 tap0 -- set Interface tap0 type=dpdkvhost >> >ovs-vsctl add-port br0 tap1 -- set Interface tap1 type=dpdkvhost >> > >> >qemu-system-x86_64 cirros.img -m 2G -vnc :1 -object >> >memory-backend-file,id=mem,size=2G,mem-path=/dev/hugepages,share=on >>-numa >> >node,memdev=mem -mem-prealloc -netdev >>type=tap,id=net1,script=no,vhost=on >> >-device virtio-net-pci,netdev=net1 >> >qemu-system-x86_64 cirros.img -m 2G -vnc :2 -object >> >memory-backend-file,id=mem,size=2G,mem-path=/dev/hugepages,share=on >>-numa >> >node,memdev=mem -mem-prealloc -netdev >>type=tap,id=net1,script=no,vhost=on >> >-device virtio-net-pci,netdev=net1 > > >You are not specifying the ifname on the command line. What does the >vswitchd.log say regarding the adding/removing of vhost ports? >e.g. |INFO|vHost Device 'dpdkvhost1' (1) has been added > > >> > >> ># add ovs flow >> >ovs-ofctl add-flow br0 in_port=1,action:output=2 >> >ovs-ofctl add-flow br0 in_port=2,action:output=1 >> > >> > >> >mac of vm1 is 00:00:00:00:00:01 vm2 is 00:00:00:00:00:02 >> > >> > >> >ping form vm1 to vm2 is ok, but ssh is not ok >> > >> > >> >On vm1 run ssh to vm2, Run tcpdump on vm2, the catching packet as >> >follow >> > >> > >> > >> >it seems the packets is all wrong >> > >> > >> >Anyone meet this situation? > > >What kernel version are you using in the guest? There is a known DPDK >issue >with 3.19.5 in the guest when using the kernel to fwd pkts. The fix for it >is here: http://dpdk.org/ml/archives/dev/2015-May/017251.html ><https://urldefense.proofpoint.com/v2/url?u=http-3A__dpdk.org_ml_archives_ >dev_2015-2DMay_017251.html&d=BQMFaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMN >tXt-uEs&r=SmB5nZacmXNq0gKCC1s_Cw5yUNjxgD4v5kJqZ2uWLlE&m=uyMUTwlAFqpkT9D3pm >PQi1LFOOG-ByD1htMGkp5eXLc&s=arX_2ZgEhNr8lPRr4LSZ2xlbzjvNSgIjq3k_e14OiDI&e= >> > > >> > >> > >> >********************* >> >RTFSC >> >********************* >> > >> > >> > > > > > > > > > > >_______________________________________________ >discuss mailing list >discuss@openvswitch.orghttp://openvswitch.org/mailman/listinfo/discuss ><https://urldefense.proofpoint.com/v2/url?u=http-3A__openvswitch.org_mailm >an_listinfo_discuss&d=BQMFaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs >&r=SmB5nZacmXNq0gKCC1s_Cw5yUNjxgD4v5kJqZ2uWLlE&m=uyMUTwlAFqpkT9D3pmPQi1LFO >OG-ByD1htMGkp5eXLc&s=4nNAuidLaEB358dGzxHZdMhXNbvM9Gi-C5JHG9Y3ye0&e=> > > > > > > > > > > > > > > >-- >Regards, >Gowrishankar M >Linux Networking > > > > > > > > >-- >Regards, >Gowrishankar M >Linux Networking > > > > _______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss
