Wednesday, August 31, 2011

DRS/vMotion sudently stops working

Hi there,

It’s easy to forget about a few things when you have a lot of things we need to take care on our daily basis activities.

DRS is one of those things you set it up, watch it working a couple of times, get comfortable on how it works, forget about it and never revisit this topic. (until it goes bad).

Despite the fact my hosts were well balanced, I decide to take a look at the vMotion activities on my cluster and figured out that vMotion metric was really low.
My next step was to check the migration tasks history. It turned out that only 2 hosts were vMotionning VMs between them, there were no activity from the others hosts.

So, I tried to make a manual migration, to my surprise I got an error message during the validation test:

vMotion interface “VMotion” on destination host “hostA” does not have any operational physical network connections.




And another message said that my vMotion network label does not match destination/source.

That’s odd ! We all know the vMotion portgroup label MUST be the same among all hosts within a cluster.

I checked the label on them and they were the same on all hosts.
That’s when I remembered they were first configured with a mistyped label and to fix that I just rename it to match the others.

Well, this time I wanted to make it right, so I deleted the vMotion portgroup and created a new one with the right label.
After that the migrations are back in business.




Remember: does not rename the vMotion portgroup label, delete and create it instead

See you ; )

Tuesday, August 23, 2011

New vSphere Licensing 5 Model

Well, despite all the amazing new features VMWARE is launching with vSphere 5 the hottest topic seems to be the new licensing model.

At first I thought they were going into the wrong direction but as I started learning the “why” and motivator to this change I realized it has value.

If you consider in the future you will have hosts with less physical CPUs (sockets) and more cores, it definitely means less licenses.

But what about the vRAM pool concept ?

I believe everybody likes to have some room for growing or even temporary high utilization. So, it’s note unusual that you have 20% of your total amount memory reserved for future utilization.
Also as memory might be the less expensive component these days, you tend to buy as much as possible.

So instead of licensing all the physical RAM you have available, you will have to license just what has been utilizes for your VMs. (vRAM pool takes into consideration the amount of allocated vRAM on each powered on VMs)

Yeah, just powered on VMs, that means you will pay what you use.

It’s better to pay attention now on how much of memory you allocate to any given VM and monitor it’s utilization to right size it accordingly, or you will be just wasting your money on licenses.

I’m not here to debate if it’s right or wrong, it’s just different, but to be honest it seems a lot fairer to me.

I encourage all of you to read the VMware vSphere License White Paper.

You now might be wondering, how do I know how many licenses do I need when upgrading to vSphere 5 ?

Good news !! VMWARE released a tool to help you identify your vRAM pool needs and amount of licenses you will need.
It’s called vSphere Licensing Advisor and can be downloaded here.

Here’s a video from http://www.virtu-al.net/ that shows you how it works

VMware License Advisor from Alan Renouf on Vimeo.

Friday, August 12, 2011

VPXD.EXE at 100%


After a relaxing 20 days vacation I’m back.

And right on my first day back a weird issue had pop up.

Virtual Center could not be opened, the O.S. was very unresponsible, when checking the processor utilization it was at 100% all the time and the most offender process was vpxd.exe

Getting a little of background of what have happened during my absence I realized the VC has been moved from one cluster to a new one. (it was supposed to happen since we are migrating our environment).

I started wondering if the issues could be because of that change and then I decided to revert it.

After shutdown the VC and register it back to it’s previous host/cluster the O.S just started normally, Virtual Center was working fine again.

But several v-motions tasks (started by DRS) started.
I wait all v-motion tasks to finish and the work load to establish on the new cluster.

At this point I can only guess
When VC started on the new cluster and tried to load balance it’s load through DRS it got kind of crazy with the situation of a new guest has been added to the cluster while it’s off, not just any guest, in fact the problem was because it’s new guest is the virtual center itself, so it went in a kind of loop.

That’s what I did to successfully migrate the VC to the target cluster.

I disabled DRS for the new cluster.
Shutdown VC again
Register it on the new host/cluster
Start the VC again

This time the VC has started properly and I just needed to enable DRS back.

Lesson learned:
When moving your VC to another cluster, first disabled DRS on the target cluster and re-enable it back after the migration.

See you.

Who am I

My photo
I’m an IT specialist with over 15 years of experience, working from IT infrastructure to management products, troubleshooting and project management skills from medium to large environments. Nowadays I'm working for VMware as a Consulting Architect, helping customers to embrace the Cloud Era and make them successfully on their journey. Despite the fact I'm a VMware employee these postings reflect my own opinion and do not represents VMware's position, strategies or opinions.

Most Viewed Posts

Blog Archive