Error preparing VXLAN in VCNS

[quintadtp]

Hi

I've been trying to configure VXLAN in a VCNS for a VDS on a vCenter Cluster, but it didn´t work.

I'll list the steps i made for help on troubleshooting:

1) I chose the target vCenter Data Center and clicked on "network preparation"

2) Clicked on "connectivity" and than on "edit".

3) So I chose the target cluster and the VDS, configured the transport VLAN and clicked on "next configure fail over" and "1603" for MTU.

4) Clicked on "finish"

Result: It ran for some time and the behavior that I expected (without a DHCP server in the transport VLAN) was seeing the hosts on VCNS, their "not ready" status and that odd IP address  on vmknic that are assigned to then when there is no DHCP server. With this cenario I might configure a static ip addresses on the vmk interfaces for vteps into the vCenter port group, and click on "resolve" at VCNS manager.

But it´s not what it's happening. The VCNS shows me the hosts, the "not ready" status and "N/A" filled as the vmknic IP Addresses. The "resolve" button is disabled  because the status button is green ("normal" condition), and that status doesn't change when I try to click on that.

The steps I made created the portgroup but no vmk interface for vteps are assigned to the hosts.

I´m already tryed to register vCenter Server again but the result was the same.

Someone has any tip about that problem ?

Thanks in advance for any help.

[Sreec]

What is the version of VCNS/ESXIhost/VC & DVS?

vCenter Server is operating on port 80 ?

Check the status of EAM  VC-Home- vCenter Solutions Manager ->  EAM -> Summary tab (Health should be green)

Few places that you could take a look esxupdate.log on the host,VSM logs and EAM logs in VC.

[nomre]

Hi, I also have this issue.

I tried to prepare new cluster (with 3 hosts), but:

- esx-vlan is not installed

- vmk is not created

- hosts is in "Not Ready" state

- Resolve button is grey out

But when I move hosts to diffrent old clusters with vxlan ready, everything seems to be working fine (vib is installed and vmk created)

My setup:

- 6 old clusters with vxlan enabled - working

- 1 new cluster - vxlan not working

- esxi 5.1, vcns 5.5.3, vcenter 5.1, vcloud director 5.5.2

I tried without success

- move hosts to working clusters and back to new

- install vib manually

- reconnect vshield manager to vcenter

Any ideas?

Best Regards,

Jacek

[Sreec]

Hi,

    Most of the informative errors would be there in the esxupdate,EAM and VSM logs(Should check those locations when the vibs are not getting installed and it shows unprepared). Mean while i'm also curious to know the version of DVS you are using in Older Cluster and New Cluster? Can you try creating one more test cluster and try to move just one host to that cluster and do a prepare and check the status?

[nomre]

Hi,

I did the test in another new cluster and did not work. I'm using the same DVS in all clusters.

eam.log - log is clean when I'm trying to prepare vxlan in new cluster, only this message is show

ERROR | 2015-11-15 18:28:36,559 | vlsi-3 | AgencyImpl.java | 990 | Invalid scope value when testing as ClusterComputeResource for MoRef: 'ClusterComputeResource:domain-c25392' Ex: vmodl.fault.ManagedObjectNotFound {

   obj = 'ClusterComputeResource:domain-c25392',

}

But I don't have this Cluster ID

PowerCLI C:\> Get-Cluster | select Id

Id

--

ClusterComputeResource-domain-c10107 -> OLD CLUSTER - vxlan working

...

ClusterComputeResource-domain-c45633 -> NEW CLUSTER - vxlan not working

esxupdate.log - log is clean when vxlan prepared in NEW CLUSTER

In the past I had problem with VIB but I made changes described in http://kb.vmware.com/kb/2053782

p.s When I move host from new cluster to old cluster VIB is installed and vmk created, vxlan is working. When I go back to NEW cluster VIB is removed.

One more thing, the NEW cluster did not show up in http://VCENTER_IP/eam/mob/?moid=agency-0&doPath=config.scope

[SRoland]

Hi Jacek,

That means that the Manager knows details about a now non-existing cluster which was prepared earlier but was not properly removed. What might have happened is:

- you created cluster

- prepared it in manager

- removed the hosts (if there was any)

- deleted the cluster from vCenter, but did not unprepare first from manager

Now what you may do is check the switch/cluster mappings with REST API  calls :

run :

GET https://<vsm-ip>/api/2.0/vdn/map/cluster

then you should get a list clusters and mapped distributed switches as a preparation.

Also you can execute (since you know the cluster MoRef(domain-c25392):

GET https://<vsm-ip>/api/2.0/vdn/map/cluster/<clusterID>

If you get switch details for this non-existing cluster then you can try to remove the mapping :

DELETE https://<vsm-ip>/api/2.0/vdn/cluster/<ClusterID>/switches/<SwitchID>

Note: replace the variables between <> accordingly.

For help on the above REST calls and how to use them see the vShield API Programming Guide

If that does not help then open ticket to remove the stale entries from the database manually.

HTH,

Roland

[nomre]

Hi Roland,

I thing you are right. I had cluster that I removed from vcenter but not from vshield (unprepared).

Unfortunately I can not delete the mapping

GET https://<vsm_ip>>/api/2.0/vdn/map/cluster

<?xml version="1.0" encoding="UTF-8"?>

<error>

<details>id to load is required for loading; nested exception is java.lang.IllegalArgumentException: id to load is required for loading</details>

<errorCode>100</errorCode>

</error>

GET https://<vsm_ip>/api/2.0/vdn/map/cluster/domain-c25392

    <?xml version="1.0" encoding="UTF-8"?>

    <clusterMapping>

    <clusterMappingSpec>

      <switch>

        <objectId>dvs-33</objectId>

        <type>

          <typeName>VmwareDistributedVirtualSwitch</typeName>

        </type>

        <name>dvVFA1sw</name>

        <revision>694</revision>

        <objectTypeName>VmwareDistributedVirtualSwitch</objectTypeName>

        <scope>

          <id>datacenter-21</id>

          <objectTypeName>Datacenter</objectTypeName>

          <name>clrdc01</name>

        </scope>

        <extendedAttributes/>

      </switch>

      <vlanId>0</vlanId>

    </clusterMappingSpec>

    </clusterMapping>

DELETE https://<vsm_ip>/api/2.0/vdn/cluster/domain-c25392/switches/dvs-33

404 Not Found

I will open ticket. Thanks for helping me!

Best Regards,

Jacek

[durrutia]

What was the resolution to your problem?