<html> <head> <meta content="text/html; charset=utf-8" http-equiv="Content-Type"> </head> <body bgcolor="#FFFFFF" text="#000000"> <div class="moz-cite-prefix">On 04/20/2015 10:20 AM, <a class="moz-txt-link-abbreviated" href="mailto:wh.h@foxmail.com">wh.h@foxmail.com</a> wrote: </div> <blockquote cite="mid:2015042022201347062321@foxmail.com" type="cite"> <meta http-equiv="content-type" content="text/html; charset=utf-8"> <style>body { line-height: 1.5; }blockquote { margin-top: 0px; margin-bottom: 0px; margin-left: 0.5em; }p { margin-top: 0px; margin-bottom: 0px; }div.foxdiv20150420220936035296 { }body { font-size: 10.5pt; font-family: 微软雅黑; color: rgb(0, 0, 0); line-height: 1.5; }</style> <div></div> <blockquote style="margin-top: 0px; margin-bottom: 0px; margin-left: 0.5em;"> <div class="FoxDiv20150420220936035296"> <div class="moz-cite-prefix"> </div> <blockquote cite="mid:201504191002476058644@foxmail.com" type="cite" style="margin-top: 0px;"> <div>   inquiry about differences between the tap and the vnet in the virt-manager Greetings! I encounter a problem about the network when using virt-manager to create and run a VM. when I want to establish a network bridge for the guest OS, I generally use two kinds of ways described as fellow : First method : create a bridge with the help of linux commad: brctl, or Second method : using virt-manager <o:p></o:p> (1) create a bridge with the help of linux commad: brctl<o:p></o:p> I use a script to establish a bridge, there are many script examples on the web to establish a bridge. when I restart the network service, I get the bridge : 1.1 create network bridge <o:p></o:p> [root@localhost ~]# ifconfig br0 Link encap:Ethernet HWaddr C2:CD:89:61:2C:F0 inet addr:192.168.1.189 Bcast:192.168.1.255 Mask:255.255.255.0 inet6 addr: fe80::c0cd:89ff:fe61:2cf0/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:16 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 b) TX bytes:1080 (1.0 KiB) eth2 Link encap:Ethernet HWaddr 00:25:90:8E:C0:F0 inet addr:172.16.1.189 Bcast:172.16.255.255 Mask:255.255.0.0 inet6 addr: fe80::225:90ff:fe8e:c0f0/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:918998 errors:0 dropped:0 overruns:0 frame:0 TX packets:2033 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:84486598 (80.5 MiB) TX bytes:148116 (144.6 KiB) Memory:fbe20000-fbe3ffff [root@localhost ~]# brctl show bridge name bridge id STP enabled interfaces br0 8000.0025908ec0f0 no virbr0 8000.525400b1de42 yes virbr0-nic 1.2 add bridge interface <o:p></o:p> the eth2 device is a physical ethernet device, [root@localhost ~]# brctl addif br0 eth2 [root@localhost ~]# brctl show bridge name bridge id STP enabled interfaces br0 8000.0025908ec0f0 no eth2 virbr0 8000.525400b1de42 yes virbr0-nic 1.3 create tap and add it to the bridge <o:p></o:p> create tap:<o:p></o:p> tunctl -t tap0 -u root brctl addif br0 tap0 ifconfig tap0 up chmod a+rw /dev/net/tun echo 1 > /proc/sys/net/ipv4/ip_forward iptables -t nat -A POSTROUTING -j MASQUERADE </div> </blockquote> The previous two lines do nothing for a bridge'd setup. Neither the setting of ip_forward nor a POSTROUTING rule will have any effect on packets that are bridged directly to a physical device (i.e. L2) - they only take effect on packets that are routed via the host's IP stack (i.e. L3).</div> <div class="FoxDiv20150420220936035296"> </div> <div class="FoxDiv20150420220936035296">///////////////////////////</div> <div class="FoxDiv20150420220936035296">weihua: </div> <div class="FoxDiv20150420220936035296">thanks for you suggestion !</div> <div class="FoxDiv20150420220936035296">I want to use "ping" command to test network state , so I make these setting .</div> <div class="FoxDiv20150420220936035296">/////////////////////////// </div> </blockquote> </blockquote> Those will also do nothing for the ability of ping to work. As I said, they only have an effect on packets that are L3 *routed* by the host, but you have the bridge directly connected to a physical device (and presumably the guest has an IP address that is on the physical network) so all the guest traffic will go directly across the bridge to the host's ethernet and out onto the physical network - everything is L2 forwarding, no IP routing involved. <blockquote cite="mid:2015042022201347062321@foxmail.com" type="cite"> <blockquote style="margin-top: 0px; margin-bottom: 0px; margin-left: 0.5em;"> <div class="FoxDiv20150420220936035296"> <blockquote cite="mid:201504191002476058644@foxmail.com" type="cite" style="margin-top: 0px;"> <div> add it to the bridge <o:p></o:p> [root@localhost ~]# brctl addif br0 tap0 [root@localhost ~]# brctl show bridge name bridge id STP enabled interfaces br0 8000.0025908ec0f0 no eth2 tap0 virbr0 8000.525400b1de42 yes virbr0-nic [root@localhost kvm_img]# ifconfig br0 Link encap:Ethernet HWaddr 00:25:90:8E:C0:F0 inet addr:172.16.1.189 Bcast:172.16.255.255 Mask:255.255.0.0 inet6 addr: fe80::c0cd:89ff:fe61:2cf0/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:37947 errors:0 dropped:0 overruns:0 frame:0 TX packets:10485 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:2581300 (2.4 MiB) TX bytes:40996927 (39.0 MiB) eth2 Link encap:Ethernet HWaddr 00:25:90:8E:C0:F0 inet6 addr: fe80::225:90ff:fe8e:c0f0/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:957962 errors:0 dropped:34 overruns:0 frame:0 TX packets:38466 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:87721507 (83.6 MiB) TX bytes:42551007 (40.5 MiB) Memory:fbe20000-fbe3ffff tap0 Link encap:Ethernet HWaddr F2:55:FC:F8:95:DC inet6 addr: fe80::f055:fcff:fef8:95dc/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:11 errors:0 dropped:0 overruns:0 frame:0 TX packets:1328 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:500 RX bytes:846 (846.0 b) TX bytes:173159 (169.1 KiB) <o:p> </o:p> 1.4 boot the guest (using the tap0)<o:p></o:p> kvm -boot c -localtime -m 4G -hda /home/kvm_img/centos6p4.img -net nic -net tap,ifname=tap0,script=no -enable-kvm </div> </blockquote> I haven't seen anything in this case where you attach the host's eth2 to the bridge. (I'm not sure why you would want to do all of this by hand rather than using libvirt and letting it take care of the details for you, but that's not what you asked...)</div> <div class="FoxDiv20150420220936035296"> </div> <div class="FoxDiv20150420220936035296"> <div class="FoxDiv20150420220936035296">///////////////////////////</div> <div class="FoxDiv20150420220936035296">weihua: </div> <div class="FoxDiv20150420220936035296">I use this command to add it to the bridge :</div> <div class="FoxDiv20150420220936035296"> <blockquote cite="mid:201504191002476058644@foxmail.com" type="cite"> brctl addif br0 eth2 [root@localhost ~]# brctl show </blockquote> </div> </div> </blockquote> </blockquote> Right. Derp. I scanned through the output a half dozen times looking for that and for some reason skipped over it *every* time :-/ <blockquote cite="mid:2015042022201347062321@foxmail.com" type="cite"> <blockquote style="margin-top: 0px; margin-bottom: 0px; margin-left: 0.5em;"> <div class="FoxDiv20150420220936035296"> <div class="FoxDiv20150420220936035296"> <blockquote cite="mid:201504191002476058644@foxmail.com" type="cite"> bridge name bridge id STP enabled interfaces br0 8000.0025908ec0f0 no eth2 tap0 virbr0 8000.525400b1de42 yes virbr0-nic what I want to do this by hand is that I want to manage the VM by script commands . But the virt-manager is a graphic tool. Would you please tell me something about the differences between the tap created by linux command and the vnet device created by libvirt? so I could write some scripts to make the tap the same as the vnet device . thanks a lot! </blockquote> </div> <div class="FoxDiv20150420220936035296">///////////////////////////</div> <blockquote cite="mid:201504191002476058644@foxmail.com" type="cite" style="margin-top: 0px;"> <div> <o:p></o:p> (2) using virt-manager<o:p></o:p> 2.1 create network bridge <o:p></o:p> [root@localhost ~]# ifconfig br0 Link encap:Ethernet HWaddr C2:CD:89:61:2C:F0 inet addr:192.168.1.189 Bcast:192.168.1.255 Mask:255.255.255.0 inet6 addr: fe80::c0cd:89ff:fe61:2cf0/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:16 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 b) TX bytes:1080 (1.0 KiB) eth2 Link encap:Ethernet HWaddr 00:25:90:8E:C0:F0 inet addr:172.16.1.189 Bcast:172.16.255.255 Mask:255.255.0.0 inet6 addr: fe80::225:90ff:fe8e:c0f0/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:918998 errors:0 dropped:0 overruns:0 frame:0 TX packets:2033 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:84486598 (80.5 MiB) TX bytes:148116 (144.6 KiB) Memory:fbe20000-fbe3ffff [root@localhost ~]# brctl show bridge name bridge id STP enabled interfaces br0 8000.0025908ec0f0 no virbr0 8000.525400b1de42 yes virbr0-nic 2.2 add bridge interface <o:p></o:p> the eth2 device is a physical ethernet device, [root@localhost ~]# brctl addif br0 eth2 [root@localhost ~]# brctl show bridge name bridge id STP enabled interfaces br0 8000.0025908ec0f0 no eth2 virbr0 8000.525400b1de42 yes virbr0-nic 2.3 run the virt-manager and select the NIC option<o:p></o:p> I find the bridge created above , configure it as fellow : in "Virtual Network Interface " setting: Source device: Host device tap0 (bridge br0) </div> </blockquote> </div> </blockquote> </blockquote> I would figure that "Host device tap0" would make a macvtap connection to tap0 (which would probably work, but be inefficient; anyway, since the tap device created by libvirt is called "vnet0" rather than "macvtap0", I know that's not the case, but *something* doesn't add up). What you want to select would look something like this on the version of virt-manager I'm using: "Bridge br0: Host device eth2". <blockquote cite="mid:2015042022201347062321@foxmail.com" type="cite"> <blockquote style="margin-top: 0px; margin-bottom: 0px; margin-left: 0.5em;"> <div class="FoxDiv20150420220936035296"> <blockquote cite="mid:201504191002476058644@foxmail.com" type="cite" style="margin-top: 0px;"> <div> Device model: e1000 Mac address: 52:54:00:84:E3:62 2.4 I start the VM using virt-manager<o:p></o:p> I get a vnet0 device created by virt-manager, [root@localhost ~]# ifconfig br0 Link encap:Ethernet HWaddr 00:25:90:8E:C0:F0 inet addr:172.16.1.189 Bcast:172.16.255.255 Mask:255.255.0.0 inet6 addr: fe80::c0cd:89ff:fe61:2cf0/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:62160 errors:0 dropped:1 overruns:0 frame:0 TX packets:22071 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:4007196 (3.8 MiB) TX bytes:59626981 (56.8 MiB) eth2 Link encap:Ethernet HWaddr 00:25:90:8E:C0:F0 inet6 addr: fe80::225:90ff:fe8e:c0f0/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:982209 errors:0 dropped:47 overruns:0 frame:0 TX packets:61101 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:89496644 (85.3 MiB) TX bytes:61778801 (58.9 MiB) Memory:fbe20000-fbe3ffff vnet0 Link encap:Ethernet HWaddr FE:54:00:84:E3:62 inet6 addr: fe80::fc54:ff:fe84:e362/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:5 errors:0 dropped:0 overruns:0 frame:0 TX packets:114 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:500 RX bytes:670 (670.0 b) TX bytes:13720 (13.3 KiB) <o:p> </o:p> both of the methods above seems that the network of guest OS is OK at most of environments, and I can get the attributes of the virtual network device . They are seemed to be the same. about the tap0：<o:p></o:p> [root@localhost ~]# ethtool -i tap0 driver: tun version: 1.6 firmware-version: bus-info: tap supports-statistics: no supports-test: no supports-eeprom-access: no supports-register-dump: no supports-priv-flags: no about the vnet0 created by virt-manager :<o:p></o:p> [root@localhost ~]# ethtool -i vnet0<o:p></o:p> driver: tun version: 1.6 firmware-version: bus-info: tap supports-statistics: no supports-test: no supports-eeprom-access: no supports-register-dump: no supports-priv-flags: no but the action between tap and vnet is not exactly the same！I will describe it in the following.<o:p></o:p> this is my project <a moz-do-not-send="true" href="http://www.iciba.com/application_environment">environment</a>:<o:p></o:p> hardware : intel xeon E5 2600 server from supermicro host OS: centos 6.4 (kernel version is changed to 3.15.10 by me) virtual software : KVM+qemu 2.1.3 guest OS: centos 6.4 (kernel is not changed : 2.6.32 64bit) <o:p> </o:p> when I establish the network by the first method: create a bridge with the help of Linux commad: brctl, tap0 device is created in the host and assigned to the guest OS. tap0 device gets a name “eth0” in the guest OS. I use “ping ” command to test network state . I get the following result : concept explanation : <o:p></o:p> local host: the OS where the guest OS is running remote host: another server which is connected to the local host using ethernet network . and this is my results: ping local host from guest OS is OK, I get reply from local host; ping remote host from local host is OK, I get reply from remote host; ping remote host from guest OS is not OK, I don’t get any reply from remote host; and this is my question. </div> </blockquote> </div> </blockquote> </blockquote> Usually this behavior indicates that the following sysctl is set to 1: net.bridge.bridge-nf-call-iptables You can check for this with: "sysctl net.bridge.bridge-nf-call-iptables". If it shows "1" set it to 0 like this: sysctl -w net.bridge.bridge-nf-call-iptable=1 But again I wonder why you even want to go to the trouble of creating your own tap devices and constructing a qemu commandline, when libvirt will do all of that for you. <blockquote cite="mid:2015042022201347062321@foxmail.com" type="cite"> <blockquote style="margin-top: 0px; margin-bottom: 0px; margin-left: 0.5em;"> <div class="FoxDiv20150420220936035296"> <blockquote cite="mid:201504191002476058644@foxmail.com" type="cite" style="margin-top: 0px;"> <div> <o:p></o:p> <o:p> </o:p> when I establish the network by the second method: using virt-manager, at this time , vnet0 device is created in the host and assigned to the guest OS. vnet0 device gets a name “eth0” in the guest OS. I also use “ping ” command to test network state . I get the following result : ping local host from guest OS is OK, I get reply from local host; ping remote host from local host is OK, I get reply from remote host; ping remote host from guest OS is OK, I get reply from remote host;<o:p></o:p> <o:p> </o:p> my question is : why I cannot get reply when I ping remote host from guest OS using the first method?<o:p></o:p> I use the same bridge: br0, so I think that the bridge: br0 is not responsible for the fault. and I have also stop the firewall . I guess the reason is that there are some differences between the tap device created by linux command and vnet device created by virt-manager. could you please kindly give my some advices to solve the problem ? <o:p> </o:p> </div> <div style="font-size: 13px;">I would be grateful if you could give me some advice . look forward to your reply !</div> <div> </div> <div>weihua</div> <hr style="width: 210px; height: 1px;" align="left" color="#b5c4df" size="1"> </blockquote> </div> </blockquote> <fieldset class="mimeAttachmentHeader"></fieldset> <pre wrap="">_______________________________________________ libvirt-users mailing list <a class="moz-txt-link-abbreviated" href="mailto:libvirt-users@redhat.com">libvirt-users@redhat.com</a> <a class="moz-txt-link-freetext" href="https://www.redhat.com/mailman/listinfo/libvirt-users">https://www.redhat.com/mailman/listinfo/libvirt-users</a></pre> </blockquote> </body> </html>