Created attachment 1192456 [details] bond_configuration.png Description of problem: Add the bond with static ip failed to register to RHEVM. Version-Release number of selected component (if applicable): 1.rhvh-4.0-0.20160817.0+1 2.imgbased-0.8.4-1.el7ev.noarch 3.cockpit-ovirt-dashboard-0.10.6-1.3.6.el7ev.noarch 4.cockpit-ws-0.114-2.el7.x86_64 5.rhevm-4.0.2.7-0.1.el7ev.noarch How reproducible: 100% Steps to Reproduce: 1.Install the rhvh4.0.0.20160817.0.x86_64 2.use the cmd "# nmtui" to enter NM TUI,delete the NICs enp2s0 and enp1s7 via NM TUI 3.Setup the bond0 over two NICs, and get ip with static successfully (please refer to attachment for detailed bond0_configuration.png info) 4.Add rhvh to engine Actual results: 1.After step2,the NICs enp2s0 and enp1s7 were deleted by NM TUI.And the files "ifcfg-enp2s0" , "ifcfg-enp1s7" were deleted under "/etc/sysconfig/network-scripts" 2.After step3,the files "ifcfg-bond0" , "ifcfg-enp2s0" , "ifcfg-enp1s7" automatically generated under "/etc/sysconfig/network-scripts" 3.After step4, add rhvh to engine failed.(please refer to attachment for detailed failed_to_rhevm_ip.png and rhevm_information info) Expected results: 1.After step2,the NICs enp2s0 and enp1s7 were deleted by NM TUI.And the files "ifcfg-enp2s0" , "ifcfg-enp1s7" were deleted under "/etc/sysconfig/network-scripts" 2.After step3,the files "ifcfg-bond0" , "ifcfg-enp2s0" , "ifcfg-enp1s7" automatically generated under "/etc/sysconfig/network-scripts" 3.After step4, add rhvh to engine successfully Additional info: 1.If don't delete the NICs enp2s0 and enp1s7 by NM TUI,continue to setup bond0,and the files "ifcfg-bond0" , "ifcfg-enp2s0-1" , "ifcfg-enp1s7-1" were appeared under "/etc/sysconfig/network-scripts". 2.If use the cmd "rm ifcfg-enp1s7" and "rm ifcfg-enp2s0",also,you should enter NM TUI (#nmtui) ,then delete the NICs enp2s0 and enp1s7 before setup bond0,or will appear some abnormal errors.For example,the attachment "failed_to_rhevm.png". 3. If Manually modify the configuration file “ifcfg-enp*",we don't encounter these problems,and add the RHVH to engine successfully.
Target release should be placed once a package build is known to fix a issue. Since this bug is not modified, the target version has been reset. Please use target milestone to plan a fix for a oVirt release.
Created attachment 1192457 [details] failed_to_rhevm.png
Created attachment 1192458 [details] failed_to_rhevm_ip.png
Created attachment 1192459 [details] rhevm_information.png
Created attachment 1192460 [details] engine.log
Created attachment 1192461 [details] rhvh.log.gz
Failed to add rhvh to engine over static bond+vlan either Test steps: 1.Install the rhvh4.0.0.20160817.0.x86_64 2.Use the cmd "# nmtui" to enter NMTUI,delete the NICs p3p1 and p4p1 via NMTUI 3.Setup the bond0 over two NICs p3p1 and p4p1 via NMTUI 4.Setup the static vlan bond0.20 over bond0 via NMTUI, ensure the vlan ip can be pingable from engine. 5.Add the rhvh to engine using the vlan ip created in step#4 Actual result: After step#5, add rhvh failed
Created attachment 1192464 [details] The ifcfg file under network-scripts before register to engine
Created attachment 1192465 [details] The ifcfg file under network-scripts after register to engine
The network setup eventually fails after loosing connectivity with Engine. The only anomaly I could see is with the slaves definition prior to the ovirt network setup: MainProcess|jsonrpc.Executor/4::DEBUG::2016-08-20 18:02:59,297::commands::68::root::(execCmd) /usr/bin/taskset --cpu-list 0-3 /usr/sbin/ifdown enp2s0 (cwd None) MainProcess|jsonrpc.Executor/4::DEBUG::2016-08-20 18:03:00,256::commands::86::root::(execCmd) SUCCESS: <err> = '/etc/sysconfig/network-scripts/ifdown-eth: line 121: /sys/class/net/a319c76b-8471-41ba-97e3-93b77c7fca76/bonding/slaves: No such file or directory\n'; <rc> = 0 MainProcess|jsonrpc.Executor/4::DEBUG::2016-08-20 18:03:00,256::commands::68::root::(execCmd) /usr/bin/taskset --cpu-list 0-3 /usr/sbin/ifdown enp1s7 (cwd None) MainProcess|jsonrpc.Executor/4::DEBUG::2016-08-20 18:03:01,323::commands::86::root::(execCmd) SUCCESS: <err> = '/etc/sysconfig/network-scripts/ifdown-eth: line 121: /sys/class/net/a319c76b-8471-41ba-97e3-93b77c7fca76/bonding/slaves: No such file or directory\n'; <rc> = 0 But I am not sure this is a real problem, but we should check this out. We will need your help to debug this. Please check the following: - After configuring the bond using NM and before deploying the host: Change the slaves "MASTER" value from the bond UUID to its name (bond0). - Issue a continuous ping to the host (from outside the host) and check if and when the connection is lost. - Continue as before and check results. (please provide supervdsm.log if still failing)
Also, please provide the output of `nmcli con show` before adding the host. It should show a single bond connection, and two bond-slave connections pointing to it.
Bug tickets must have version flags set prior to targeting them to a release. Please ask maintainer to set the correct version flags and only then set the target milestone.
It is seems that there is no way to modify an existing interface as slave-bond and if we will add new connection we will end up with : ifcfg-enp6s0-1 ifcfg-enp6s0
Hi,all Thanks for Edward's help about details.When setup static bond0 successfully,but we should add other steps to ensure add rhvh to engine successfully. Other steps: 1.Use "nmtui" to setup static bond0 successfully 2.Manually modify the ifcfg files about ifcfg-enp2s0 , ifcfg-enp1s7,Change the slaves "MASTER" value from the bond UUID to its name (bond0). ifcfg-enp2s0:(i manually Change the slaves "MASTER" value from the bond UUID to its name (bond0)). TYPE=Ethernet NAME=enp2s0 UUID=99af055c-0a46-4525-b682-d7fb175bd721 DEVICE=enp2s0 ONBOOT=yes #MASTER=f7079956-fcba-464e-9ea5-b59672b0e9db MASTER=bond0 SLAVE=yes ifcfg-enp1s7:(i manually Change the slaves "MASTER" value from the bond UUID to its name (bond0)). TYPE=Ethernet NAME=enp1s7 UUID=b5ae73f7-c1ac-40bf-81aa-0ce9e76063de DEVICE=enp1s7 ONBOOT=yes #MASTER=f7079956-fcba-464e-9ea5-b59672b0e9db MASTER=bond0 SLAVE=yes 3.Restart the NetworkManager,and ping to the host (from outside the host) is OK. #service NetworkManager restart 4.After above steps,i can add rhvh to engine successfully. #Some Specific information are listed on attachment: process.txt Thank you. Yihui Zhao
Created attachment 1192739 [details] process.txt
I also tried on rhel7.2 as a host: 1.Just configured the static bond via nmtui, and added to engine, failed either. 2.But creating the bond via nmtui and then modify the ifcfg files based on Edward's suggestion, which is "change slaves MASTER value from the bond UUID to its name", finally adding successfully.
According to comment 16, the original bug issue still happened on RHEL 7.2 as a host, so moving this bug component to vdsm for further investigation on vdsm + NetworkManager enhancement.
For clarification: Unless nmtui is able to define bonds in a manner that can be used with initscripts, there is not much we can do, except using NM as a driver to configure the system. Another approach could be to create an script that converts the NM ifcfg files to a format that initscripts can work with.
We will support nm-cli or cockpit. We will not support nm-tui.