VMware Cloud Community
cypherx
Hot Shot
Hot Shot

Difficulty testing SRM upgrade 5.0.3 / vsphere 5 U3 to SRM 6.1 / vsphere 6 U2

Hello,

I am testing a vSphere upgrade in a test network.  Basically I restored, using Veeam a DC, Internal CA and our vCenter windows VM's into their own test network, internet access provided by proxy when needed via Veeam SureBackup Lab Proxy.

Anyway I learned through the process that vSphere 6.1 will require a SQL upgrade.  No problem, Upgraded SQL express 2005 to 2012.  Sucessful so I performed the same upgrade in the live environment, as well as remove and recreate the ODBC drivers using the updated SQL Server 11.0 client.

So now in my lab vCenter 6.0U2 is up and running, esxi hosts obviously "not responding" since this is just a lab for learning... but the next thing I need to figure out is how to get SRM and vSphere replication upgraded.  I don't think I can test vSphere replication as this is in a test network with no actual ESXi hosts, but I should be able to get the SRM components installed correct?

Coming from vCenter serve 5.0 update 3, we were on the vSphere replication that involved 2 VM's, a management server and a replication server, version number 1.x.  SRM is version 5.0.3.

For no good reason it doesn't seem like VMWare built in upgrade routines to take you from 5.0.3 directly to 6.1, unlike vCenter server which has no qualms about such a big jump (must be two different teams with different motivational levels).  So I take it you have to take SRM 5.0.3 to 5.5.1-3572851 and then take that 5.5.1 to 6.0 and then take 6.0 to 6.1, according to the upgrade path.

However I go to install it and first it warns me that I need to create a 64-bit DSN.  I believe that is done by running C:\windows\syswow64\odbcad32.exe, so I do so and I create the same DSN name and pointer that I have under C:\windows\system32\odbcad32.exe.  That should take care of the requirement I would think, anyway in both cases doing a "TEST" indicates a successful connection to the database.  The next step asks if I want to include VSphere Replication, yes I do, then finally credentials to the vCenter server. I enter the brand new Administrator@vsphere.local credentials that I created during the 6.0U2 upgrade (which work great by the way either with the C# client or the web interface), but when I click next I get an "Internal error: unexpected error code: -1.". First thing I do with such a message is google it, along with Site Recovery Manager.  Since I'm not coming up with anything clear from others I decided to post here.

Any idea the best upgrade path?  Should I just start SRM over?  What ramifications would I have?  I'm replicating 28 VM's, and if its just a days work great... however if ALL the data has to be deleted and then rehydrated over the 50mbps wan, it could take a month. 

Tags (1)
2 Replies
vbrowncoat
Expert
Expert

If you decide to upgrade I would highly recommend following the supported path. That said, for an environment your size rebuilding wouldn't be that significant. For VR and your replicated VMs you can retain the data at the target site and use that as a seed for VR after you've installed the new version. This would save you from having to copy everything again. VR would just do a checksum process to see what was different and copy over only those changes. Details on all of this are in the VR FAQ vSphere Replication Frequently Asked Questions (FAQ)

What is the difference between pausing replication and stopping replication?

If an administrator pauses replication, the replication configuration remains in place but data will not be replicated until replication is resumed. If replication is stopped, the replication configuration is removed and replicated data at the target location will be removed. If an administrator wants to keep a copy of the replicated virtual disks at the target location, these steps can be followed:

Pause replication.

Copy the files that are to be kept at the target location to a different folder or rename the existing folder.

Stop replication, to remove the replication configuration.

For SRM it will require recreating your mappings, protection groups and recovery plans. There are some things you can do with the new vRO workflows or directly thru the APIs which may help speed up the process a little.

cypherx
Hot Shot
Hot Shot

Ok so suppose I do start over.  I'm trying this in a virtual lab, note this is not production so I will not be able to 100% complete the recovery site pairing as the lab is on an isolated network.  This exercise is just to get familiar with the process prior to moving forward with production.

I removed SRM 5.0.3 from the new vCenter 6.0u2 windows server.

I manually removed the vSphere replication extension from the new vCenter 6.0u2 windows server.

I ran the SRM 6.1 installation and choose a new internal database option since we are starting over.

Install suceeded and I logged into vSphere web client and the SRM icon is visible.  I go into this option and click on sites.  When I do that a pop up in the lower right appears that says "Failed to connect to Lookup Service at https://vcenter.domain.com:443/lookupservice/sdk. com.vmware.vim.vmomi.core.exception.CertificateValidationException: Server certificate chain not verified.  It fades away and the I can still navigate around the UI, however I do not see any way under "Sites" to walk me through adding a "protected site".

I think this is needed first because all the other options like Protection Groups, Recovery Plans allows me to walk through creating them, but it says Empty Inventory and there is no way to proceed.  To get "inventory" I think the sites have to be linked up first, correct?

Does the error about the lookupservice have anything to do with no UI exposed to enter a URL / path / credentials to "protected site"?

I did restart this server just to make sure all the services restarted properly after installing SRM but the same thing occurs.

Reply
0 Kudos