DRS not functioning the way I would like it too..

[virtualnoob2]

Hi All,

I am currentley involved in a VMware View PoC.. I am running now 5 ESXi hosts with a toal of 24GHz of CPU and 20GB RAM. I currentely have 3 view servers (regular,security,transfer) and 4 Desktop VMs .. each of these machines have 1GB of RAM allocated. For some reason all these VM's are only going to two of my servers in the cluster. The other 3 servers have NOTHING on them .. what gives? .. I have turned off HA since I didn't feel it was nessecery at this time for my PoC but DRS is turned on and I've set it to automated (aggresive) ..I am getting errors on one of the hosts about host resource usage.. that wouldn't happen if DRS was working properly. Whenever I click 'runDRS' nothing occurs.. it just says "refresh recommendations" in the tasks pane but dosen't actually give any recommendations.. Also I should add that I get the load imbalanced error (0.367)

Any Help on This would be greatly Appreciated..

Thanks All

[Troy_Clavell]

why not set DRS to fully automated with the most aggressive threshold.  However, a single ESXi Host may be providing all the necessary resources for your guests and therefore DRS has done it's job.

You have confirmed you can manually vMotion between all hosts in your cluster?  having in imbalanced cluster of 0.367 tells me your not that far off, but it's balance enough where DRS, depending on your Migration threshold, has done it's job.

[opbz]

suggest you put more of a load on your esx servers.

either create more vms and have them do something or increase the memory of them and get them to do something.

DRS will not vmotion stuff off if 1 or servers are capable of handling the full load of your vms.

[bulletprooffool · Accepted solution]

Your VMs will not migrate to different ESX hosts, if the hosts they are running on are not under any sort of strain (even if you set the DRS policies to fully automated and 5 stars)

If you have the luxury of being over-resourced, you could consider using something like DPM to actually shut down ESX hosts that are not in use.

If in the mean time, you would like to verify that DRS is functioning properly, just drop some more VMs on the hosts and make sure to hammer the resources (you'll be wanting CPU / Memory usage of over 70% on each host to get DRS to do some 'balancing'

good luck.

[jayolsen]

You could also place a host into maintenance mode and tell it to move the guests off it, this would use DRS to complete if you have it fully auto.

[bilalhashmi]

Like Jayolsen pointed, put your host in maintenance mode and if ur drs is set to fully automated and your vmotion is working, it should move your VM. DRS will not come into play unless your host is under stress and your VM is gauranteed to perform better on the other host. I think the reason why your VM is not moving to another host yet beacuse, the host doesnt have enough load.

I will not recommend this, but if you have access to the host, you could perhaps take all the sockets out but 1 and see if that puts the host under stress.. this will really be a waste though.. if putting it in maintennce mode moves the VM, i wouldnt worry to much.. you just dont have enough load on your VM.

[virtualnoob2]

Hi All,

SORRY, for the late response.. Things have been a little hectic these past 2 weeks and have been crazy busy. While the View PoC has been on hold since my last post I went back at it today and began loading all the VMs I had.. once I had 3 windows 7 machines, an xp vm, a smoothwall vm, and 3 view servers.. I got an error on one of my ESXi hosts complaing about host resource usage.. When I try to push it and clone a VM, it just cancels after a while and the host remains with the resource usage error. I now have a 5 host cluster and all my VMs keep defaulting to only two of my host. I feel like the network settings mite be messing this up. Becuase when I right click on my cluster and go to add a new VM to the cluster >> at the end of the process I get an eror about every host on in cluster except the two hosts that keep occupying my VMs.. something about "network A can't communicate with Network B" or something like that.. will copy the error again tomorrow and post back... once again sorry for the lateness

[virtualnoob2]

BUMP

[bulletprooffool]

Looks pretty sure that you have networking issues on the 3 faulty hosts.

Please post your error message  - so we can attempt to help you.

[jayolsen]

Sounds like maybe your guest networks are not named the same across all of your hosts.

[virtualnoob2]

Hey All,

I fixed the issue! Was a network adapter problem with a few of my VMs.. all is well now. Thanks for the Support !

[bulletprooffool]

No prob virtualnoob - please mark the thread as 'answered'

Glad you resolved your problem.