VMM Integration on ACI

[kthned]

Hi 

I am just doing a VMM integration to my ACI. vCenter is not directly connected to ACI. I just wanted to test installation of vDS on the remote vCenter but no luck. see below the ACI can nicely talk and reterive ESX host + VMs but the state is "unknown" and there is no vSwitch created on my vCenter, Any Idea ?

[Robert Burns · Accepted solution]

If you can ping the vCenter from the APIC, it should at least create the DVS. Are you using the root/Administrator account to connect APIC to the VCenter?

Robert

[kthned]

Hi Robert,

Thanks for reply.

Yes first I was missing the administrator password. But than I see "failed to retrieve service content ... " fault. Looking google so I found to change vcenter hostname as IP address. and viola I got my vDS on the vCenter. Strange enough I can ping DNS name of vCenter from APIC CLI.  

is APIC hit by

*

DNS provider updates ignored by VMM

CSCuy94595

*

I am using version version: 3.1(1i)

//Umair

[cesar panaijo]

Hi syedumairali,

I have the same issue and I don't understand very well what you did in order to get connection to the vCenter. I'd really appreciate if you may explain a little more the steps you followed. Thanks

[Robert Burns]

One other thing to note, is that each APIC needs to have connectivity to the VC.  I've seen the instance where there was a FW between the APIC & VC, and the user only allowed APIC-1 through the FW via ACL.   When creating a VMM, one APIC will be elected as the "Master" controller for that VMM and will be solely used to push configuration data to VC.  In this case, APIC2 was elected the VMM shard owner and was unable to push configuration to the VC. 

Additionally all APICs will establish a listener socket to the VC for receiving updates (Endpoint moves etc).  You can't force which APIC becomes the owner for a VMM, so just ensure all APICs (if this is a cluster) have reachability.  Testing connectivity (ping/DNS) between the VC and one APIC within a cluster is not adequate.   If this is a standalone APIC, then obviously this wouldn't apply.

Robert

[Claudia de Luna]

Hi @cesar panaijo,

The fix is to use the vCenter IP address rather than the host name, when you are defining your vCenter Controller object.

The Bug alert mentioned above also indicates an APIC reboot could resolve the issue.

[cesar panaijo]

Thank you guys,

My problem was that the connection to the vcenter is in-band, So when I created the access policies, I forgot to associate the VMM domain with the EPG related to my vCenter, and after that, insted of putting the hostname, I did what you recommend, putting the IP address, for now it seems well. :)

[Claudia de Luna]

Thanks for circling back with an update. Always good to hear!