Description of problem: There are few problems with the network update flow on Setup Host Networks dialogue. Probably they should be addressed in separate bugs. Will do so, after opening this bug. In this bug I would like to concentrate in the Sync Network option, which behavior is not clear and appears to be buggy (or maybe just requires more documentation). Here is my current scenario on my local setup (running 3.2.5) I will try adding customer cases later. 1. After failing to update eth1 with new logical network on a host (for a reason), RHEV-M decided that rhevm logical network of the host, which is on eth0, is not in sync. Why? How is that related? I was trying to edit eth1, no changes to eth0 to which rhevm is assigned. 2. Trying to synchronize the network - no luck, giving me this error: "Error while executing action Setup Networks: Network is currently being used" Note: network ifcfg scripts on the host for rhevm logical network matches exactly the configuration on the portal. What is the meaning of networks not in sync? What should I do at this point? How bad is that the network is not in sync? All appears to be up and running correctly. Digging in the logs, I see this: engine.log: ~~~ 2014-01-29 18:10:14,718 INFO [org.ovirt.engine.core.vdsbroker.vdsbroker.SetupNetworksVDSCommand] (ajp-/127.0.0.1:8702-1) [229912a6] START, SetupNetworksVDSCommand(HostName = rhevh-1.gsslab.rdu2.redhat.com, HostId = f3cb3db6-ac31-48c5-85bc-27959e091567, force=false, checkConnectivity=true, conectivityTimeout=120, networks=[rhevm {id=d57f5d9c-13df-4038-b5f9-3fba5f402709, description=Management Network, subnet=null, gateway=null, type=null, vlanId=null, stp=false, dataCenterId=3493fa12-7133-4c02-ba77-a60b505178cf, mtu=9000, vmNetwork=true, cluster=NetworkCluster {id={clusterId=null, networkId=null}, status=OPERATIONAL, display=false, required=true}}], bonds=[], interfaces=[eth0 {id=8829c473-7a1e-481a-8d5e-005aa1e58db3, vdsId=f3cb3db6-ac31-48c5-85bc-27959e091567, name=eth0, macAddress=00:1a:64:ca:42:58, networkName=rhevm, bondName=null, bootProtocol=DHCP, address=10.10.177.178, subnet=255.255.248.0, gateway=10.10.183.254, mtu=1500, bridged=true, speed=1000, type=2, networkImplementationDetails={inSync=false, managed=true}}, eth1 {id=ee26add3-0c00-4332-858c-4c98a8e40be6, vdsId=f3cb3db6-ac31-48c5-85bc-27959e091567, name=eth1, macAddress=00:1a:64:ca:42:5a, networkName=null, bondName=null, bootProtocol=NONE, address=, subnet=, gateway=null, mtu=1500, bridged=false, speed=0, type=0, networkImplementationDetails=null}], removedNetworks=[], removedBonds=[]), log id: 10174699 2014-01-29 18:10:14,729 INFO [org.ovirt.engine.core.vdsbroker.vdsbroker.SetupNetworksVDSCommand] (ajp-/127.0.0.1:8702-1) [229912a6] FINISH, SetupNetworksVDSCommand, log id: 10174699 2014-01-29 18:10:15,249 ERROR [org.ovirt.engine.core.vdsbroker.vdsbroker.BrokerCommandBase] (ajp-/127.0.0.1:8702-1) [229912a6] Failed in SetupNetworksVDS method 2014-01-29 18:10:15,250 ERROR [org.ovirt.engine.core.vdsbroker.vdsbroker.BrokerCommandBase] (ajp-/127.0.0.1:8702-1) [229912a6] Error code ERR_USED_BRIDGE and error message VDSGenericException: VDSErrorException: Failed to SetupNetworksVDS, error = bridge rhevm has interfaces ['vnet0'] connected ~~~ Indeed, running on the host: ~~~ # brctl show bridge name bridge id STP enabled interfaces ;vdsmdummy; 8000.000000000000 no rhevm 8000.001a64ca4258 no eth0 vnet0 ~~~ No idea where it came from, I didn't add it manually for sure. This is probably another bug here, but I would like to understand the questions I asked earlier: - What is the meaning of networks not in sync? What is the impact on the system? All appears to be up and running correctly. - Why are we giving this error to the end user. The error is completely wrong.
Hi Marina, There is a brief mentioning of the 'sync' network in the admin guide [1] and also a feature page for it [2]. The 'sync network' feature was introduced as part of the setup networks, and basically it allows an easy method for the admin to sync the network configured on the host with the logical network definition on the data-center level. The network on the host might be out-of-sync with the logical network definition when one or more of the following properties is different: vlan-id, mtu or vm network (network implemented as vm network on host and defined as non-vm network on the dc and the opposite). Please file a doc bug if you find the admin guide is missing info. Specifically in the case above there is an attempt to sync the 'rhevm' network. The 'rhevm' network is defined as a vlan vm network. However #1, in comment 1 rhevm seems to be configured as a vlan (on top bond0.2). This will probably cause a connectivity issue. However #2, at the end of the log, the failure reported from vdsm is: "ERR_USED_BRIDGE and error message VDSGenericException: VDSErrorException: Failed to SetupNetworksVDS, error = bridge rhevm has interfaces ['vnet0'] connected" Which indicates there is a running vm which uses the bridge that is about to be modified. There is an open bug to enable it (Bug 999701) and this is the cause of the failure. I assume that once you move the host to maintenance and retry (or at least turn-off any running vms that use 'rhevm'), the action will succeed on vdsm side, but a potential loose of connectivity might happen due to the vlan-->no vlan change. [1] https://access.redhat.com/site/documentation/en-US/Red_Hat_Enterprise_Virtualization/3.3/html-single/Administration_Guide/index.html#Editing_host_network_interfaces [2] http://wiki.ovirt.org/SetupNetworks_SyncNetworks
First of all: I opened the bug to documentation. It is indeed very vague and should be fixed: https://bugzilla.redhat.com/show_bug.cgi?id=1066161
Honestly, after opening the documentation bug, which was my main complain, and after fixing my latest 3.3 environment per recommendation on comment#6, I think we should close this bug. If there would be new problems with the sync behaviour, we will open a new specific bug for it, having now a better understanding of how it should work. *** This bug has been marked as a duplicate of bug 1066161 ***