Solved

VMWare HA with limitated ressources

Posted on 2010-11-27
14
553 Views
Last Modified: 2012-05-10
Hi Experts,

I am running a VMWare 4.1 cluster with 2 ESX servers.

When both nodes are up, I am running critical VM on a dedicated ressource pool, and some lab VM without interest in another one. Both servers ressources are needed to run everything.

Is it possible, in case of a node failure, to stop running lab VM on remaining node, to move critical VM with enough ressources to operate properly on it?

Best regards,
0
Comment
Question by:acharreau
  • 6
  • 5
  • 2
  • +1
14 Comments
 
LVL 11

Assisted Solution

by:louisreeves
louisreeves earned 50 total points
ID: 34223329
The VM is running on shared storage? I am assuming it is, and without Storage VMOTION. In this case, you are only speaking of Vmotion 1 mission critical machine? THis can be handled many ways. However, to answer in absolutes, VMWARE will not shut the (test) machine down for you dynamically. However, You can either

1. Dynamically move the machine if the needed resources are not available
2. USE DRS and and HA to keep machines together or keep them apart.
3. Configure the Test box resourses to "very low" to accomodate the needed resources.

THe formation of Distributed resource scheduling and high availability work with resources you already have. VMware works with Mins and Maxs. You have to be aware of the maximum machines you can place on a node.

http://www.vmware.com/pdf/vsphere4/r41/vsp_41_config_max.pdf

This game is about controling the resources so you know how much you have. To answer your question, you have to know you have resources on the second node - "all the Time" to use DRS and HA. How much memory and CPU are available? Do you have enough so  the mission critical machine can be failed over in the middle of the night so you dont have to come into work?  If you dont have the extra resources, then plan it on the in your next budget. HA, DRS and FT all depend on it.




0
 
LVL 40

Expert Comment

by:coolsport00
ID: 34223343
The Availability Guide explains what you need in addition to what is shared above:
http://www.vmware.com/pdf/vsphere4/r41/vsp_41_availability.pdf

Good Guide to get familiar with and not too difficult to understand. The key, as what was shared above, is you need to have your hosts connected to SHARED storage. Once you have a HA Cluster created/config'd, if 1 host goes down, a VM will be moved to the remaining working host.

Let us know if you have further questions.

Regards,
~coolsport00
0
 
LVL 1

Author Comment

by:acharreau
ID: 34223467
Thanks for your first replies.

Storage is shared between both ESX server and HA / FT cluster is made.
Ressources of lab ressource pool are also set to low and mission critical ressource pool to high.

No problems regarding maximums, we are on a very small infrastructure : 2 ESX servers, 16Gb RAM and 8 cores per ESX server for around 15 VM at all. Storage is 1.5 Tb shared.

Mission critical VM can operate with 13Gb and 8 cores, so on a single node, but only if other uncritical VM are not consuming ressources. I made an HA test disconnect network of a node : vCenter detected the host isolation and initiate the failover. But due to insuffisant ressources, I got an error during the process.

So there is probably a misconfiguration somewhere. But where?
Thanks for your ideas.
0
 
LVL 40

Expert Comment

by:coolsport00
ID: 34223570
You will get an error with only 1 node left in the HA Cluster (if it's what I think it is). Can you please provide a screenshot?

~coolsport00
0
 
LVL 40

Expert Comment

by:coolsport00
ID: 34223958
Also...may I ask why you need 8 vCPUs on 1 VM? Is it an app that requires 8 CPUs?

In re-reading through your post, you say you're getting 'insufficient resource' error. Well, that's because your other host cannot support your mission critical VM on the other host with all your other  VMs, resource-wise. So the error is valid. If your 1 VM takes up 13GB and all those cores (vCPUs), well, your other VMs need some resources as well. If 1 host goes down, you're out the resources on that host, so you're left with only the resources of the working host. You would need to double the RAM & cores for your HA to be able to host all VMs. Let me know if my explanation is unclear.

~coolsport00
0
 
LVL 1

Author Comment

by:acharreau
ID: 34225057
Thanks for your reply coolsport00.

I was not clear enough in my post. We have got 8 mission critical VMs and some test VMs.
So this is not a big VM but many small. I set HA priorities but this is not enough.

As I said, the goal fo me is to stop lab VM with the failover of mission critical VMs.
Or bypass process warning and manage this with high / low ressources allocation in ressource pools.
But at this time, failover process is not working.

I tried to manually stop all the test VM and disconnect an host network. Failover process works fine and quickly.
0
 
LVL 40

Expert Comment

by:coolsport00
ID: 34225435
It's still a bit confusing, but let me see if I have this correct. You're saying you want to separate your test VMs from critical ones so that if a host dies, only your cirtical VMs migrate to the operational host? There is no way to do what you're wanting...again, if my assumption is correct. HA doesn't 'discriminate' VMs on a failed host. It migrates all or none. You can't configure VMs to be powered down on a host if it fails such that only critical VMs will migrate and your test ones just power down so as not to take up needed resources for your critical VMs and thus HA failover fails. Hmm...actually, I may be wrong in my previous statement. In your cluster settings, there should be an area to configure what your VMs will do (power down, etc.) in the even of host failure. I'm not at my office to verify, but I believe this is accurate. Configure that functionality for your VMs and see if that works for you. I'm not sure you can do this configuration with individual VMs, but you may. (i.e. configure some VMs to power down and others to migrate)
Test and let me know.

~coolsport00
0
VMware Disaster Recovery and Data Protection

In this expert guide, you’ll learn about the components of a Modern Data Center. You will use cases for the value-added capabilities of Veeam®, including combining backup and replication for VMware disaster recovery and using replication for data center migration.

 
LVL 40

Expert Comment

by:coolsport00
ID: 34225444
"in the EVENT of a host failure", not in the EVEN of a host failure. Sorry about that. :)

~coolsport00
0
 
LVL 28

Expert Comment

by:bgoering
ID: 34227560
A couple of things to look at in your HA configuration. First it likely the admission control setting that is not allowing vms to power up on the failover host. If you set that to disabled then the system will allow hosts you deem to be mission critical to be restarted on the other host.

The other settings are under Virtual Machine settings - set you lab/non mission critical hosts to be powered off in response to host isolation. After you do that only your mission critical workloads will continue to run.

Good Luck
0
 
LVL 1

Author Comment

by:acharreau
ID: 34233951
Hello Guys,

You are right, I have a couple of settings to manage how an active VM on a failing host will migrate on remaining host. I can set priority and state in case of a failure.

But : these settings only impact VM that has to be migrated to remaining host.

My goal is, on remaining host, to control which VM has to remain active and which VM has to be powered off (because not critical) to release normally used ressources.
So I am not looking for management on failing host or migrated VM. I am trying to manage VM that are working on a clean host, but where I need to find available ressources to transfert critical VM on.

Any idea?
Best regards,
0
 
LVL 40

Accepted Solution

by:
coolsport00 earned 300 total points
ID: 34234252
There is no setting for that "acharreau". The startup/shutdown of VM setting is only if a HOST reboots/shuts down. There's nothing configurable for a normal running host on how to power down certain, non-critical VMs. I'm not really a scripter person, but there "may" be a way to do so via script.

~coolsport00
0
 
LVL 28

Assisted Solution

by:bgoering
bgoering earned 150 total points
ID: 34238622
Yes, that would likely be something you would have to script. Might be better to set up you resource pools such that your test workloads would get few resources during times of contention.
0
 
LVL 1

Author Comment

by:acharreau
ID: 34243181
All right, I will look for a script to achieve this goal.
Thanks for your time
0
 
LVL 1

Author Closing Comment

by:acharreau
ID: 34243194
Part of the solution provided
0

Featured Post

Complete VMware vSphere® ESX(i) & Hyper-V Backup

Capture your entire system, including the host, with patented disk imaging integrated with VMware VADP / Microsoft VSS and RCT. RTOs is as low as 15 seconds with Acronis Active Restore™. You can enjoy unlimited P2V/V2V migrations from any source (even from a different hypervisor)

Join & Write a Comment

Using a software based iSCSI solution, there is no requirement to purchase a hardware iSCSI initiator or TOE (TCP/IP Offload Engine) interface card. A standard network interface card can be used to connect an existing ESXi server to a remote iSCSI t…
It Is not possible to enable LLDP in vSwitch(at least is not supported by VMware), so in this article we will enable this, and also go trough how to enabled CDP and how to get this information in vSwitches and also in vDS.
Teach the user how to delpoy the vCenter Server Appliance and how to configure its network settings Deploy OVF: Open VM console and configure networking:
Teach the user how to use vSphere Update Manager to update the VMware Tools and virtual machine hardware version Open vSphere Client: Review manual processes for updating VMware Tools and virtual hardware versions: Create a new baseline group in vSp…

707 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question

Need Help in Real-Time?

Connect with top rated Experts

14 Experts available now in Live!

Get 1:1 Help Now