View Updating Issue

[caryers]

I have a 100 user Pilot View 4.6 environment. I am experiencing a new issue that I have not come across before.

I have my two internal VCS servers behind a NLB VIP. I just created 3 new VMs that start with LIB to my existing media pool of that has 15 existing such VMs. When I access the View admin GUI, and place a filter that start with LIB, I get just the incumbent list of 15 not the expected 18. I have logged into the each VM manually and have confirmed that are 100% fully operational in the domain. I then happen to login to the other VCS server and place a filter on LIB, and can now view all 18 VMs. So it appears the the VCS servers are not insync properly. I don't think this a DNS thing as both VCS servers are pointing to the same two DNS servers and in the same order. I believe the View Connection servers are using the SQL DB that my vCenter is using which is local to the vCenter server itself and is SQL2k8 Express.

What should I do to try to troubleshoot this issue and resolve? Any advise will be greatly appreciated...

Thanks in advance...Scott

[mittim12]

This KB, http://kb.vmware.com/kb/1021805, should show you how to view replication status and how to force replication if you need to. 

[caryers]

This KB article is dead on. The ADAM database is out of sync. However, the command in the article to force replication does not work. Something happened to Node 1 on 1/18/2012. The ADAM database area within EventViewer shows a good ADAM sync on 1/18/2012, and then the very next entry is on 02/02/2012.

If this KB article you sent me doesn't work, do I need to rebuild my ADAM database? If so, is there an KB article for that? Looking for any direction.

Thanks in advance...

Scott

[Linjo]

What about removing the "out-of-sync" one and reinstall it? Should clear up the problem with minimal interruption since you have 2.

// Linjo

[caryers]

That was a thought.

I do have two security servers (SS) behind a NLB VIP in my DMZ for external access. However, at this time, my FW NAT is pointing directly to node 1 instead of the NLB VIP. So my SS Node 2 is basically doing nothing at this time. So with that said, isn't each SS paired with one particular VCS?

Last week, I actually removed both VCS servers from the domain and re-added them. VCS node 1 indicates the last successful sync was 01/18 and VCS node 2 indicates the last successful sync was from 02/02/12. So I am not sure what action should be taken....

[mittim12]

I agree with Linjo.   Just remove  VCS1 node all together from View, delete it is a VM, and then build a new connection broker server.  You can then add this to the replication group with VCS node 2

[Linjo]

Remove and readd to domain would not make any difference here, the AD LDS is totally separated from the Domain AD.

// Linjo

[caryers]

Since my 1 functional SS is paired to VCS node 1, my VDI environment would be down until for whatever amount of time it would take to reinstall VCS, etc.

Each SS can only be paired to 1 VCS server correct? 

I am also concerned that VCS node 2 indicates that the last time the ADAM was sync'ed was last 02/02/2012 which is some 4 days ago. Should I have any concerns there? Will it matter if VCS node 1 or node 2 was the primary ADAM server?

Thanks for any feedback...

Scott

[mittim12]

yes, I believe you can only have one VSS server associated with a connection broker.   You  could always associate the functioning VSS with the 2 broker to minimize the downtime.

[caryers]

Regarding the ADAM database in my scenario here, I believe one VCS is the primary and all other VCS are replica servers. Is this correct?

If so, prior to me starting this thread, I reverted to a 2-week old snapshot of my primary VCS server(node 1). Now my two VCS server are out of sync when it comes to View. My VDI environment is still working fine, however, I have slightly different information when I access the View Admin GUI on each server. Since I can not force replication per the KB article that was provided me earlier in this thread, how do I go about getting them sync'ed up?

I see your suggestions of removing a VCS node from View. Which node should that be? If node 1, will then node 2 transparently become the primary ADAM database? Which ever node I do remove per your suggestion, should I really have to delete the VM? I though abuot simply removing it from View and uninstalling View altogether, then simply do a fresh re-install. Basically I am asking is a View uninstall a truly clean uninstall for my purposes as I will be using the same Netbios name and IP address, etc?

Any direction would be greatly appreciated....

Thanks, Scott

[markbenson]

Sounds like your replication between View Connection Servers has failed.

Whether a Connection Server is a standard or replica is just an install-time concept. Once they are all up, there is really no concept of standard/replica they are all equals. When you install a new replica, you can point it to any existing server. Even if you loose the original standard instance, the replicas all continue.

The replication probably failed in your environment because of the snapshot revert operation.

If you are able to remove all but one (by uninstalling the Connection Server package and VDMDS package), then the easiest thing to do is simply reinstall the other servers as replicas. This will set up a new repliaction agreement and will copy the LDAP config data from the existing server. You will then have all servers operating in synch.

If this is a production environment, you'll experience outage on the replicas while you do this. Snapshot your main Connection Server as a precation before doing this.

Mark.

[caryers]

Great Feedback Mark! Much appreciated. Just curious, where did you acquire that info on the ADAM database? I don't recall seeing it in ny View Admin guide.

Although the VCS servers are out of sync from an ADAM db  perspective, my VDI environment is still working for the most part. All internal users that go directly through 1 of my 2 VCS servers works great! (VCS node 1 & 2  are behind a NLB VIP) However, I have just a few external users that are getting a time-out when the desktop is trying to spawn only externally. However, these same VDI users access on the internal network, have NO issues. I basically have 1 Security Server (SS) that is paired to VCS node 1. Consquently, all external remote traffic is routed to VCS node 1 server. (Remember node 1 was the VCS server that I reverted to a 2-week old snapshot). Keep in mind, the majority of my VDI users are external and work fine.

So when I create new VMs, they are seen in the View Admin GUI on VCS node 2 but not in the View Admin GUI on VCS node 1. I am not sure of the relationship with the correlation above. However, I am simply chalking up the timeouts to the ADAM database being out of sync on VCS node 1.

So given this info, which VCS server would you suggest I uninstall the Connection Server package and VDMDS package from, node 1 or node 2? If you indicate node 1(which I believe you will suggest), I will need to pair my SS to node 2 so external traffic can continue to flow during my uninstall/reinstall of View on node 1. With that said, are there any reboots with pairing my SS to a different node? Basically, is this a quick and dirty change over? Any gotchas, etc.

Thanks in advance...

Scott