vSphere: Create a Distributed Switch

In VMware ESX 3.x you had to make sure that is you have a cluster every host has the same network configuration with the same port group names. Whit a couple of host this isn’t that hard, but what if you have a 16 node cluster, or more? Making a change to the network configuration could take some time.

With vSphere you have the ability to maken a switch that is automaticly distributed acorse all attached hosts. This is called a Distributed Switchs (dvSwitch).

So, how do we created a Distributed Switch?

• In vCenter goto Home and under Inventory select Networking
• Right click the Datacenter object en select New Distributed Switch. The following screen is displayed

no images were found

• Give the dvSwitch a name and select the number of dvUplinks . A dvUplink is later connected to one or more physical nic’s in the host.

no images were found

• Here you can select the hosts you want the vdSwitch on. When you select a host, you can also select the physical nic for the dvUplink. Just put a mark before it.
  You can also do this later bij clicking on Add Later.

no images were found

• If you click finish a default port group is created. I like to create my own port group so I deselect Automatically create a default port group

no images were found

• Wait a few seconds before the dvSwitch is created and is distributed to all the hosts. You can check this to look at your recent tasks
• When the dvSwitch is ready click on the name of the dvSwitch you just created. In the command section you can select New Port Group.

no images were found

• Give the name for this dvPortGroup (In my case dvPortGroup-Demo) and the numbers of ports for this portgroup and if you want VLAN’s.

no images were found

• Select finish to let the dvPortGroup be created

The only thing we have to do is to migrated the excisting Virtual Machines that are connected to a ‘normal’  port group to the newly dvPortGroup.

• In vCenter select the dvSwitch and in the commands field select Migrate Virtual Machine Networking

no images were found

• Select the source Network (the ‘normal’ port group) and destination Network (the dvPortGroup) and click on Show Virtual Machines.
• The Virtual Machines connected to the source Network are displayed in de bottom field. Here you can select the Virtual Machines who you wanne migrated to the new dvSwitch

So what if I want to add for example the VMkernel or Service Console portgroup to it? Ok, here we go.

• First, you can only do this at host level. Every host will have it’s own ip adres for a VMKernel or a Service Console.  Go to the VMware host | Configuration | Distributed Virtual Switches and click on Manage Virtual Adapters

no images were found

• Click Add to launch the Add Virtual Adapter

no images were found

• Now we have to choose if we want to migrate existing virtual adapters or if we want to add a new virtual adapter. In my case, I choose to migrate.

no images were found

• Select the port group you want to migrate and to witch dvPortGroup you want to migrate it to.

no images were found

• You will a summary and when you click finish the port group will be migrated

no images were found

no images were found

Of course you will have to repeat this step on all the other hosts.
As you can see, it isn’t hard to create a dvSwitch with dvPortGroups.  I find it a great step forwards and a great new feature of vSphere.

vSphere: Migrate from a Thick to a Thin VMDK

Inflating a Thin VMDK to a thick VMDK is easy, just browse your Datastore, select the VDMK file you want to inflate, right mouse click on it and select inflate.
Note: Your Virtual Machines needs to be powered off and inflating a disk can take-up some time. In my test I inflated a 5 GB Thin disk on witch 3,6 GB was occupied took 8 minutes.This was on a NetApp with Fiber Channel (1GB).

But what if we want to do the opposite? There are a few way to create a Thin VMDK file:

• Create a Virtual Machine
• Clone to a template
• Clone a Virtual Machine
• Use VMware Storage VMotion to migrate the virtual machine storage

So, the most easy way to migrate a Thick VMDK file to a Thin VMDK file is to use Storage VMotion.

1. Right click the Virtual Machine and select Migrate
2. Select change datastore
3. Select the Lun you want to move the Virtual Machine to
4. Select the format. This will be Thin provisioned format of course
5. Select finish

That’s all

Extend VMFS Datastore in VMware vShpere

When your VMFS Datastore in VMware ESX 3.x was running out of disk space, you had 2 options. One was to move all of the VM’s on that Datastore to a new VMFS Datastore with more space avaliable or increase the Lun on your storage and add a extend to the excisting Datastore. There are many blog writting if it’s wise to add a extend to a Datastore.

Now with vSphere it’s easy to increase the size of your VMFS Datastore if you have the space avaliable on your lun.

Here my setup. 2 vSphere servers with vCenter installed. I have one Lun on a NetApp. The lun is 15GB big with I created a 10GB VMFS datastore on it.

no images were found

Select the properties of the Datastore and click on Increase.

no images were found

Notice that on the same Lun the Datastore excits Expandable is marked as: Yes

no images were found

You will see the free space avaliable. In my case 5GB.

no images were found

Select the size you want to add to the Datastore.

no images were found

A summary it displayed. If you agree select Finish.

no images were found

When the action is finished you will see that the Capacity is increased. In my case 15GB.

no images were found

To prove there is no second partition. Here a screenshot from the Service Console with fdisk (fdisk -l).

no images were found

As you can see, now it’s pretty easy to extend a excisting VMFS Datastore.

reinitialize ingres database for Arcserve for Linux

A couple of customers of mine have Arcserve for Linux to backup there OES2 machine.
With Arcserve for Netware, if you had a problem with the database. Arcserve provided a empty database on the installation CD (You can question your self that if a vendor provides a empty database for recovery, how stable is that database?). But with Arcserver for Linux you cannot delete some files and copy the empty database any more. You have to uninstall Arcserver en reinitialize the database.

Here is how:

• Go to opt/CA/ingresii  and execute  ./uninstall_ingres -a
• Verify that all the ingres RPM’s are uninstalled by doing   rpm –qa | grep CA-ing*
• Remove /opt/CA/ingresii folder
• cd to the $BAB_HOME/bin and start ./cadbase_setup
• Run the setup for a new database

If you have a 64bit installation for Arcserve you can find the unistall_ingres on the installation CD.