Problem Creating Provider vDC in vCloud 5.1 Evaluation Guide's path

[chadmichael]

I'm folling the vCloud 5.1 Evaluation Guide.  Everything seems to be going well, but when I try to create my first provider vDC, I have problems.  At the vCD level, I'm just told the operation has timed out while trying to install the agent on one of my two esx hosts.  Interestingly, this is the host that I installed the vShield manager on.  When I dig down into the vCenter level, I see more specific complaints n the event logs for the hosts. 

Apparently, the host has to be powered down ( for the installation of the vCloud agent I guess )  and this requires a DRS migration of the vShield Manager vm to my ohter host.  This migration fails.  And here is the most relevant information from the event.

DRS failed to generate a vMotion recommendation for a virtual machine on a host entering Maintenance Mode. This condition typically occurs because no other host in the DRS cluster is compatible with the virtual machine. Unless you manually migrate or power off this virtual machine, the host will be unable to enter Maintenance Mode.

That seems pretty specific, but I have no idea what could cause "compatibility" issues with the other host.  As far as I know, they are identical.  I certainly meant for them to be.  The Datastore that the vm is on is a shared datastore between the two hosts, as far as I can tell. 

Any trouble shooting tips would be appreciated.

[IamTHEvilONE · Accepted solution]

the ESXi host is not powered down ... the host is put into maintenance mode.

In an ideal world, you would have vCenter/vShield in a different cluster than your vCloud deployed VMs.

For this to happen, you need to ensure DRS is at full auto mode, and that vMotion does actually work.

Aside from that, if you create a provider vDC and prepare ALL hosts, they may all try to enter maintenance mode on their own.

This is a bit of an awkward scenario ... since you can't move VMs onto a host entering maintenance mode.

so even though if failed, if one prepared fine ... see if you can prepare the failed host ... this time it may work.

[chadmichael]

THanks.  Couple of questions.

In an ideal world, you would have vCenter/vShield in a different cluster than your vCloud deployed VMs.

I've heard this before, but note that the documentation for the evaluation set up clearly shows the vShield Manager in the resource cluster, as opposed to the management cluster.  Is this a legacy deployment pattern?

For this to happen, you need to ensure DRS is at full auto mode, and that vMotion does actually work.

I'm suprized that I am allowed to create DRS, as I did, on my cluster if vMotion is working . . .  shouldn't there be some sort of a warning here?

Aside from that, if you create a provider vDC and prepare ALL hosts, they may all try to enter maintenance mode on their own.

Are you saying that even with vMotion functioning, there could be a problem anyway?  I only have two hosts in my cluster, perhaps they are going to have a race condition on who can get into maintenance mode first?  This would suggest that you couldn't really deploy any vDC onto clusters that currently have vm's on them, right?

[IamTHEvilONE]

"i've heard this before, but note that the documentation for the evaluation set up clearly shows the vShield Manager in the resource cluster, as opposed to the management cluster.  Is this a legacy deployment pattern?"

No ... it was a matter of trying to use the fewest number of ESXi hosts.

"I'm surprised that I am allowed to create DRS, as I did, on my cluster if vMotion is working . . .  shouldn't there be some sort of a warning here?"

You mean a resource pool?  Don't confuse the automation mode (full vs manual) with having DRS enabled/disabled.  If you create a cluster of exactly 1 ESXi host, you have to set DRS to manual mode to allow resource pools to exist.

There is no warning if DRS is in Manual Automation mode, since it's doing exactly what it what it thinks is normal ... not moving VMs around.

"Are you saying that even with vMotion functioning, there could be a problem anyway?  I only have two hosts in my cluster, perhaps they are going to have a race condition on who can get into maintenance mode first?  This would suggest that you couldn't really deploy any vDC onto clusters that currently have vm's on them, right?"

vCloud Director is the one putting all the hosts into maintenance mode at the same time.  If you only have 2 hosts, it'll do all hosts.  There are some considerations to be had with the "proof of concept" that you are doing since the vShield manager is in the deployment cluster.  This issue would be avoided with a non-trial setup.

all that being said ... there are probably still some improvements to be made.

[chadmichael]

I found out that I didn't have vmkernel port with vmotion enabled  

[IamTHEvilONE]

LOOOOOOOOOOOL, and I did mention making sure that vMotion worked in the first place :smileysilly:

Anyways ... if you got it fixed up, please mark a post as the correct answer.  This way others can know which post resolved the problems.

In the new forum, it associates the correct answer to the original post, so you don't have to search the thread anymore.