Error: "Failed to attach all ports" seen when a vMotion task fails

Products

• VMware NSX

Issue/Introduction

• Customer is unable to vMotion any VM to a particular host\transport node or subset of hosts in the cluster. The VM vMotion\migration task fails with the following error:
  •  ERROR: "Failed to attach all ports"
• The host or hosts with the issue recently had NSX uninstalled\unprepared and installed\prepared again.
• VM's connected to NSX-T segments.
• Manually trying to connect a powered-off VM vNIC to the vDS\virtual switch on a host with the issue fails as well.
• In the /var/run/log/hostd.log log of the vMotion destination host a log message like the following will be observed:

2024-04-02T07:59:24.998Z info hostd[2101935] [Originator@6876 sub=Solo.Vmomi opID=lj3y4hja-21051468-auto-cj7f1-h5:72192407-e0-01-ff-3c13 user=vpxuser:user\domain] Result:
--> (vim.fault.PlatformConfigFault) {
-->    text = "Failed to attach all ports",
-->    msg = "",
--> }
2024-04-02T07:59:24.998Z error hostd[2101935] [Originator@6876 sub=Req@internalvim25... opID=lj3y4hja-21051468-auto-cj7f1-h5:72192407-e0-01-ff-3c13 user=vpxuser:user\domain] Method >vim.host.VMotionManager.initiateDestination threw undeclared fault of type vim.fault.PlatformConfigFault

• In the /var/run/log/nsx-syslog.log an error will be seen indicating there is a problem in creating a port because it cannot be found:

2024-04-02T07:59:24.797Z nsx-opsagent[2103639]: NSX 2103639 - [nsx@6876 comp="nsx-esx" subcomp="opsagent" s2comp="nsxa" tid="2104234" level="ERROR" errorCode="MPA44205"] [PortOp] Host switch info with uuid >for tzId > [<UUID>] was not found
[..]
2024-04-02T07:59:24.797Z nsx-opsagent[2103639]: NSX 2103639 - [nsx@6876 comp="nsx-esx" subcomp="opsagent" s2comp="nsxa" tid="2104234" level="ERROR" errorCode="MPA42001"] [CreateLocalDvPort] >createPort(uuid=<UUID>, zone=<UUID>) failed: Failed to create port <UUID> with VIF <UUID> >because DVS id is >not found by transport-zone id <UUID>

 

Environment

NSX 3.1.X.

 

Cause

• The VM cannot be connected to the destination host, since the vDS cannot be found due to it being in a disabled state.
• The vDS is in this state due to the following settings being configured on it. Please note the configuration of the vDS can be retrieved by running the 'net-dvs -l' command:

- com.vmware.common.opaqueDvs.status.component.vswitch = up , propType = RUNTIME

- com.vmware.vswitch.disabled = false , propType = CONFIG

• The vSwitch parameters are set that way when the host has NSX uninstalled\unprepared, but fail to be configured correctly when the host is prepared again as a transport node.

Resolution

• This issue is resolved in 3.2.0 due to changes made on host sync. If a variation of this issue is seen on a later version, then export the NSX Manager and host logs as soon as possible. Open a NSX support case with the logs referencing this KB.
  
  
• The workaround to enable the vDS on the host is to do the following:

• • com.vmware.common.opaqueDvs.status.component.vswitch = up , propType = RUNTIME ---------------> propType needs to be set to CONFIG, instead of runtime
  • com.vmware.vswitch.disabled = false , propType = CONFIG ---------------> this needs to be removed
    

• The steps for the workaround are the following:

1. 1. Verify that the host\transport node is reporting state 'success' in the NSX Manager UI. If it is,
   2. Then manually remove “com.vmware.vswitch.disabled” property using the following command:
      
      - net-dvs -u com.vmware.vswitch.disabled -p hostPropList <vds-Name>
   3. Set “com.vmware.common.opaqueDvs.status.component.vswitch” property to 'CONFIG' using the command:
      
      - net-dvs -s com.vmware.common.opaqueDvs.status.component.vswitch=up -p hostPropList <vds-name>
      
      
   4. Restart opsagent —> /etc/init.d/nsx-opsagent restart