Enthusiast
Enthusiast

Unable to Remove vDS from vCenter 5.5

Jump to solution

Hi All,

I'm trying to remove a distributed switch from the vcenter but it comes up with an error message box:

"vDS SW-DCN-C1_C2 port c-40 is used by entity <not-found.> Virtual NIC 4000, type:vmVnic"

I've delete all port groups, and remove all hosts and vm that were connected to vds.

I've move down, all vds and c-40 port id to "Unused Uplinks" the dvUplink in Teaming and Failover Policies.

What else can I do?

Thanks for the help everyone.

1 Solution

Accepted Solutions
Immortal
Immortal

you can remove the entry from vCenter Database its Under vpx_entity

run the below script in the management studio if your are using the MSQL.

select * from VPX_ENTITY order by name;

look for the object name which is DVswitch and get the ID of the object and delete it running the script.

involve you DB admin to run the script.

take a snapshot of the vCenter virtual machine if it is a VM and a backup of the DB  Smiley Happy

If the object is missing even in DB try to remove it from vCenter MOB.

https://vC.fqdn.com/mob

this will help you for sure.

I had a same issue once.

in my case the DvSwitch was removed but the UPLINK remained in the vCenter view

when i tried removing it (object can not be found)

View solution in original post

14 Replies
Enthusiast
Enthusiast

Any Help, please???

0 Kudos
VMware Employee
VMware Employee

Hello MinoDC,

Did you try these?

24 Hour Timeout Prevents Removing ESX vNetwork Distributed Switch | VMETC.com

VMware KB: Changing the default timeout for locked dvPorts

VMware KB: Disabling vNetwork Distributed Switches

I suppose either any of the VM's still have the portgroup attached to it. So you can check on each and every VMs network to see if that portgroup is used anywhere or you can try the above.

Hope this helps.

0 Kudos
Enthusiast
Enthusiast

Hello Sajal,

I had already read these articles, but I haven't blocked port and I haven't VMs / Hosts connected, as you can see in the image.

Error.JPG

I double checked the VMs (4) and none have the vNic on vDS but all on the standard.

Any other suggestions?

0 Kudos
VMware Employee
VMware Employee

Hi MinoDC,

Well I see. By the way the port C-40 is a conflicting port means it is not really an active port.

VMware KB: vNetwork Distributed Switch contains dvPort IDs with prefix c-

I suggest the following two actions:

1. Check each and every VM for the network configurations tab . Select VM --> Edit Settings and then networking. If it is still pointing to that vDS, change that.

     Do this to be double sure. vDS does not show and connected VM is ok, but what the VMs say? have you checked all the VMs?

If that does not help, then the other thing that comes to mind before going to the logs and analyzing for other clues is

A. First restart the network or management services of the ESXi hosts which were connected to the vDS and then check

B. Restart vCenter Service and then check.

Hope the above helps. If not then would need to have a look at the logs

0 Kudos
VMware Employee
VMware Employee

Another point, I still can see a Uplink NIC and a Port Group in the attached screen shot. To be able to remove the vDS you need to Unlink the Uplink NIC and then remove all the Portgroups and then try to remove the vDS. Since C-40 is not the real blocking port you need to check these as well.

0 Kudos
Enthusiast
Enthusiast

Smiley Sad

I've done these steps:

  • Check each and every VM for the network configurations tab, and all VMs have the Network in Standard Port Group
  • I tried to migrate VM network from vDS but without result, as you can see from attached images:
    • MigrateVM-1.JPG
    • MigrateVM-2.JPG
  • I connected vpshere to each Host (no through vCenter) and I checked the Network in Configuration ->Network , and each one has only Standard Switch
  • I restarted management services of the ESXi hosts and vCenter Service
  • I had already moved down, all vds and c-40 port id to "Unused Uplinks"  in the vDS and c-40 from Edit Settings ->Teaming and Failover Policies, as you can see from the images:
    • vDS-Uplink.JPG
    • c40-Uplink.JPG

Now unfortunately I can't migrate back hosts in vds because, among the various attempts, I also updated vDS to version 5.5 , having the host in 5.1.

There is a way for force the removal of the vDS from the vCenter?

0 Kudos
Enthusiast
Enthusiast

Others suggestions, help, please?


0 Kudos
VMware Employee
VMware Employee

HI MinoDC,

At this stage I would suggest to go for a full fledged support ticket from VMware, where a proper log analysis would be done to understand the reason behind this

0 Kudos
Immortal
Immortal

you can remove the entry from vCenter Database its Under vpx_entity

run the below script in the management studio if your are using the MSQL.

select * from VPX_ENTITY order by name;

look for the object name which is DVswitch and get the ID of the object and delete it running the script.

involve you DB admin to run the script.

take a snapshot of the vCenter virtual machine if it is a VM and a backup of the DB  Smiley Happy

If the object is missing even in DB try to remove it from vCenter MOB.

https://vC.fqdn.com/mob

this will help you for sure.

I had a same issue once.

in my case the DvSwitch was removed but the UPLINK remained in the vCenter view

when i tried removing it (object can not be found)

View solution in original post

Virtuoso
Virtuoso

Hi MinoDC,

What is the build number of your ESXi host?  You can check by performing 'vmware -vl' from an ssh session to the host.  I noticed an interesting fix in the Resolved Issues of the latest Release Notes for ESXi 5.5 Update1:

<snip>

Unused vSphere Distributed Switch (VDS) ports are not cleared from the .dvsData directory on the datastores

During vMotion of a virtual machine that has a vNIC connected to VDS, port files from the vMotion source host are not cleared from .dvsData directory even after a while.         

This issue is resolved in this release.

</snip>

It's probably not related, and likely wouldn't fix the problem retroactively without the help of support (speculation).  I just haven't observed this in the wild so I cannot say.  Just figured I would share in case it is relevant.  Also notable is that vCenter 5.5.0 supports the updated ESXi host level of 5.5 U1 so it would be safe to update to this revision of ESXi without worrying about vCenter compatibility.  Of course with any update, you may introduce additional new issues so study well.

vCenter 5.5.0 Supports ESXi 5.5 Update1:
vCenter-5.5.0-supports-ESXi5.5U1.png

0 Kudos
Enthusiast
Enthusiast

Hi grasshopper,

in this moment, I've ESXi 5.1.0 build 799733 and can't upgrade to 5.5 upd1 because I've others esxi with the same version in production.

Now I've disconnected two ESXi from the production environment for migrate all in another environment with vCenter 5.5.0 1312298.

If you think that the upgrade resolve my issue, I can try to fo it.

I keep you updated

Thanks for your reply.


0 Kudos
Virtuoso
Virtuoso

Thanks for clarifying the ESXi host version (799733).  That is likely not related to the release notes I linked.  Unfortunately VMware generally tells us what issue is fixed in the release notes, but they don't always specify when it started.  In your case though your current plan to move to a fresh vCenter is the right one.  Let us know how it goes.

BTW, that ESXi version 799733 is really old.  That's the very first GA release of 5.1.  This version presents other risks to stability but I don't recommend upgrading just to resolve your current issue.  Since you manually checked each host via the vSphere Client I would venture to say the problem is in the vCenter DB as others have mentioned above.  Moving to a new vCenter is the easiest way to get around it without engaging support or hacking the DB.

Not sure if you have ruled this out as well, but ensure you have checked the templates by converting them to VM and reviewing edit settings.  You might also choose to grep the vmware.log of all VMs for the UUID of the old dvSwitch, but that's probably overkill at this point.  Again your plan to move the hosts to a fresh VC should take care of this.

Enthusiast
Enthusiast

Hello.

Problem solved.

It 'was deleted the entry of the vds from dabase oracle.

Thanks to all

0 Kudos
Virtuoso
Virtuoso

For the benefit of future readers, I ran into a similar problem today (1000v vDS in my case) and blogged the fix (SQL query) at the link below:

http://vmkdaily.com/?q=node/45

0 Kudos