vCD 1.5 Resource Pools gone after DRS disable/enable

[Paule83]

Hi,

we just reconfigured the Cluster, where we host our vDC and the vOrgs (a test environment at the moment).

I disabled DRS and enabled it, which caused all resource pools to be deleted.

This prevents vCD to create any new VMs for the vOrgs, so basically, we can't work with vCD any more.

Is there any way to restore the resource pools or to make vCD start a recreation of them?

I don't want to reconfigure alle vOrgs, vDCs, networks, etc.

[admin]

Well there is no real good news here I think.  This is one of the things we have said in the past about disabling DRS once vCD is in play.  I am not aware of any way to have vCD automatically re-create them.  I will however use this in a test case for restoring a vCenter DB backup to see if that brings them back.  I am curious now, but AFAIK there is no way to have vCD recreate them, unless there is something I am not aware of.

[Paule83]

Thanks for the fast reply, even if the news aren't really that good 🙂

I think this is a "feature" vCD should have a look at for the future, as disabling/enabling DRS or HA clusters is a not so uncommon task, if you are trying to solve problems with these features, isn't it?

[admin]

Oh I agree 100%.  I will ask around and see if there is anything in the works.  Of course I will not be able to say but I will ask the question.

[_morpheus_ · Accepted solution]

You need to recreate the resource pools in VC, put the VMs back into the resource pools. Then you need to go into the VCD database and manually remap the VCD inventory to the new resource pools. This isn't something that anyone should attempt on their own. You should file an SR and request assistance from GSS to do that.

[admin]

Indeed goes without saying messing with the DB without a GSS ticket is not advised

[Paule83]

Ok, so I will do 🙂

Thanks a lot for your help.

I hope, you will still address this as an idea for a new feature to make such things easier, as I would believe, not everyone would like to file a support case, just he disabled HA for a second.

Let us know, what the outcome was.

Thanks again,

Paule

[admin]

Already having internal discussions about this yes.  Nothing in the near term but use GSS for now.  We are tracking this in a couple places though.  I will be doing some of my own lab testing to see if there is any other ways out of it.

For the record disabling HA does not do the damage of disabling DRS.  That would not really affect vCloud as much other than the admission control part.  This is definately something we are looking at to help assit in fixing, but right now it is what it is.

[Paule83]

As a wrapup for all users, who might stumble upon this thread, Chris wrote a nice article about the problem and I wrote one too, taking his suggestion to use the Inventory Snapshot script for a test drive.

Chris Colloti's Article

My Article

[admin]

Thanks.  It seems the inventory snapshot does not help.  Still changes the MoREF ID's.  This is just one use case in a list of many I am tracking for recovery modeling.

[Erik_Bussink]

First thing when I read this post & the follow on Chris's audi osegment in the VMware Communities Roundtable was to create a nice Message of the Day for any vCenter servers that host a vDC resource.

"vCloud Director is now being implemented in the XXX Infrastructure.

Do NOT ever turn off the DRS fonctionality in vCenter on the Production Cluster.

[mzzt]

So what would happen if you just restored vcenter from backup?  I mistakenly did exactly what I shouldn't have (disabled DRS) and killed all my resource pools.  Our environment is relatively small and no changes were made since the last full backup of the vcenter server.    What would happen to vcloud director if I just restored the entire vcenter server?

[Paule83]

As you are saying your environment is a small one and you have a backup, you could test this for the rest of us 🙂

Just create a snapshot of your VC, do the restore, let us know about the outcome.

Would be very much appreciated.

[mzzt]

So I was able to restore the Resource Pools by restoring the SQL database to the backup before I disabled DRS.  Luckily we have a small environment right now, so no damage done.   Good lesson though.

[CalicoJack]

ok - in the interest of those that have to do things the most difficult way possible... I don't have a back up to failback to.

Any suggestions of any other ways to get the Organizational vDC to have a resource pool?
So we can add more vApps and VMs and they can be modified?

I am talking to GSS and they are telling me that modifiying the database isn't the fastest way to do this, and I don't have an available DBA.

~thanks.

[admin]

Yes I would expect the Pools to come back from Backup with the same MoREF ID's.  Now had you had VM's and other objects deployed after the backup.....those would have most likely been un-associated since their MoREF ID's on restore would be re-created.

[admin]

There is no quick way.  The ONLY way is DB manipulation with GSS......or recreate them all from scratch and re-import the vApps to the new vDC's from vCenter since the VM's are still there.

[CalicoJack]

I have figured out a work around for this.  I only had about 40-50 vApps and these were mostly development and QA.

What I needed was the ability for vDCs to be able to add and make modifications to the vAPPS inside.  I went to the Organizational vDC page and added a new vDC using the organization that I wanted to replace.  I then went into the old vDC and stopped the vapp.  From there it gives you the option to move it to the newly created vDC.  When I had them all moved over, I removed the old vDC and renamed the new one.

[admin]

That will always work, but requries downtime and in a production setup might not be ideal, but is a way around it.