Replication Failure - Data Centres with Distributed Switches

[richardhindley]

Since upgrading to Veeam 6, our backups are now faster than ever. Great!

However, we are now unable to replicate VMs between our two data centres which run distributed switches. We can replicate fine if the VMs do not contain any NICs. However, if we try to replicate VMs containing NICs (with or without "enable network remapping" ticked) the replication job fails immediately after the replica VM is created in destination data centre.

The error we are seeing in Veeam is "Processing configuration Error: A specified parameter was not correct.spec.deviceChange.device.port.portgroupKey".

In Vsphere client, we are seeing the error "Spec for device 4000 on entity MYVMNAME_replica has portgroupKey dvportgroup-41912 that is different from the portgrup key dvportgroup-41921 of the port 321"

Within Vsphere client, I can see that Veeam has created MYVMNAME_replica but if I "Edit Properties", Network Adapter 1 summary says "Invalid backing". Under "Network Connection", the Network Label box is blank and the port number shows"Port: 321"

Port 321 is the network port which is listed on the VM in the source data centre. However, the destination data centre has a completely different port numbering scheme, and port 321 is not valid in the destination data centre.

Any ideas what is meant to be happening at the point the error occurs? Presumably the replication process cannot rely on port numbers being identical on the two distributed switches, as this would probably only be the case in a very limited test scenario.

We have open an support case [ID#5162823] which has just been escalated to next level, but I wondered if anyone else is seeing this error or this behaviour, or can explain the architecture?

Richard

[Gostev]

Hi Richard, could this be Issue #16 from the Known Issues sticky topic?

[tsightler]

I think that's definitely it. I've seen this at a couple of user sites over the last few weeks and Hotfix 2 has been the answer. If you haven't applied Hotfix 2 then I would strongly suggest asking support to provide it and see if this corrects your issue.

[richardhindley]

Thanks - I will request hotfix 2 and try it.

[Jamesatnuss]

Hey,

I'm trying to set up replication between two sites. Hq and colo. both sites use dvswitches with same names and vlan mappings

When replication runs it fails at configuring the replicas network. When I look at the replicas network setup it has a different vlan etc. it seemed like it was trying to pick a similar port ID but after a little more digging it seems to randomly pick the vlan it's assigned too

Is this possible ? Am I missing something? Has anyone been able to make this work using dvswitches?

Hardware is dl360s and dl380s. Hq San is a dell equilogic 6000p and colo is an Emc vnx 5300. We tried to make this work with VMware site recovery manager but were told that we need to have replication set up between the sans as a separate elimate and that it isn't possible with different sans

Thanks everyone!

[JWester]

We also have this problem with a few VMs. We use B&R 6.5, so the patch should be included.

When I try to replicate a specific VM I get the error "A specified parameter was not correct. spec.deviceChange.device.port.switchUuid" in vSphere client. The (not successfully) replicated VM also has an "Invalid backend" entry in its network configuration.
Some VM replicate without problems, some don't.

Any ideas? Case (00137560) is open.

[dellock6]

"invalid backing" is a specific problem with distributed switches, and it usually happens when there is some change in the connections of the VMs while vCenter is offline and so cannot coordinate the configuration change on the distributed switches.
This behaviour is a real achille's heel of DVS, and has been documented in several blog posts, take a look here

http://longwhiteclouds.com/2012/02/27/t ... ed-switch/

Inside you will find kinks to other posts. Hope you find them useful.

Luca.

[ErikSi]

Wanted to restore a vm connected to a DVS. Tried to restore to a different esx server with no DVS and set network to "Not Connected" in restore job. Got same error as Jwester A specified parameter was not correct. spec.deviceChange.device.port.switchUuid". Picked a network from the target esx server and the restore went OK.