Skip navigation

    This blog post explains how to share a VMDK between two virtual machines so they can be clustered or any application that needs two machines to share their disks.

  

    Assume there are two VMs that participate in the clustering

    Node-01 and Node-02


    Create a Virtual disk on Node-01 that has to be shared with Node-02

1.jpg



    Add a hard disk by clicking on Add a hardware and select Hard disk and click Next

2.jpg



 

   Select Create a new virtual disk

3.jpg



    Choose SCSI (1:0) which creates a fresh SCSI controller apart from the default one created during virtual machine creation and click Next

4.jpg




    Click on Finish

5.jpg




    Click on the newly created SCSI controller and change the SCSI Bus sharing option to “Physical

6.jpg

   Click on OK and power ON the machine

 

    Now the newly added disk shows up on the Node-01 machine as /dev/sdb.

7.jpg


     We need to add the Virtual disk created on Node-01 to Node-02


   

     Click on Edit Setting of Node-02 machine and select Add hardware and click on hard disk

     And click Next

8.jpg

 

    Click on “use an existing virtual disk” and click Next

9.jpg


 

    Browse to the location where Node-01’s VMDK file is present and select it

10.jpg


 

     Confirm the path and click Next

11.jpg

 

 

  

 

    Select SCSI (1:0) which creates a new SCSI controller on Node-02

12.jpg



     Review the changes and click Finish

13.jpg

 

 

    Select the newly created SCSI controller and select Physical under SCSI bus sharing option and click OK

14.jpg


    Power on Node-02 and confirm that Node-01’s shared disk is showing up on Node-02

15.jpg


   You might want to turn the Multi-writer flag in the advanced options of VM so you can let the machines write to the disk. Follow the below link for the same

    VMware KB: Disabling simultaneous write protection provided by VMFS using the multi-writer flag


   Note: This might cause some problems while doing vMotion of the machines involved in clustering. So do some research before you do it.

 

I ran into some issues in my lab while trying to delete the entities like Org vDC, Edge gateways and Org Networks on vCloud Director(happens with all the versions) . I couldn't delete the Org vDC as it was dependent on the Edge gateway. I couldn't delete Edge Gateway as it was dependent on a routed Org Network. And Org Network would not get deleted . Things like this happen when vCD losses connectivity to vShield or some activities that are performed on vCenter directly which affect the vCD environment. The best thing to do is not to mess around with the vCenter once it is added to the vCD. But sometimes things get out of hand and everything gets messed up and we need to look for a way out of it.

 

I tried deleting the entities on vCenter as they had lost connectivity to vCD and that made things worse. The problem with vCD is, it does not have an option to force-delete an entity. If it looses connectivity then the entities are there in the portal in an inconsistent state and you will not be able to delete them. The whole Environment looks messy. Sometimes it so happens that after deleting all the entities in an Org vDC it still does not allow you to delete the Org vDC, though all the resources that are related to it on vCenter too are deleted.

 

The only way to get out of this is to either get into the database level and delete the entries for these stranded/un-deletable entities or to reset the database and configure the vCD cell all over from scratch.

 

Obviously someone like me who does not have knowledge about SQL would not want to go to the level of deleting the entries of tables in a database. But configuring the cell all over again would require configuring all the things again. Which is again a tedious task. So i decided i will try deleting the entries and if i screw up the database i anyway have to configure the vCD cell again which i was thinking of doing initially.

 

After doing a lot of research online i found couple of blogs and links that spoke about deleting the entries manually on a vCloud DB level(for MS-SQL).

 

Lets jump straight into it. At first you need to connect to your database server and connect to your instance through SQL Management Studio(for SQL Server, not sure how this works on Oracle)

 

Run the below script to get the entries of all the entities like the Org vDCs, Network pools, Edge gateways and Org vDC

 

/****** To List the Table entries *******/

SELECT TOP 1000 [id]

      ,[vdc_id]

      ,[lr_type]

      ,[name]

FROM [vcloud].[dbo].[vdc_logical_resource]

 

Where [vcloud] is the name of the vCloud Director database instance.

 

Now to move the Org Networks to other Organization vDCs you will have to run the below query:

 

UPDATE vcloud.dbo.vdc_logical_resource

SET vdc_id = 0xabunchofUUIDnumbersfortheorganization

WHERE id = 0xabunchofUUIDnumbersforthenetwork;


Here (vdc_id) is the organization you want the network to move to and (id) is the parameter that pertains to a particular entity which you consider to move to another vDC.



If you need to delete the entry in the database table


DELETE FROM vcloud.dbo.vdc_logical_resource

WHERE id = 0xabunchofUUIDnumbersforthenetwork;


where (id) will be the id of the entity you want to delete.


And if you are someone who is familiar with SQL Database then you can figure out more commands to modify tables and makes things right.


Have Cisco UCS Servers? Use Different management tools(UCS manager for cisco and vCenter for VMware) to manage the blades and your virtual environment?

What if i say you can manage all your virtual and physical servers in a single management pane? In your vCenter?

Yes, Cisco UCS-vCenter integration plugin is the solution. It brings capabilities of UCS Manager into vCenter and you can use your vCenter to manage all your Cisco Blades.

 

This Plug-in works only with vSphere 5.1 and needs web client to work with. If you don't have web client in place then you will have to install it.

Believe me its worth it.

 

Lets jump in to the installation steps and see how this configuration is done.

 

Pre-requisites:

  1.     As already mentioned, the plug-in only runs on vSphere 5.1 within the Web Client.
  2.     VMware PowerCLI version 5.1 should be installed on the machine where we’ll be doing the actual install – Download it here 



We will have to Download the Plug-in from here.  Download the ZIP file that is available and extract RegisterVCPplugin.ZIP on a machine where we have installed the PowerCLI.

run the extracted file. If we choose to use Enter plugin URL, We will have to download the plugin package and put it on our web-server so its available over an URL.

1.jpg

Enter the IP/Hostname of the vCenter where the plugin has to be installed.

If Hostname is used then the machine where the installer is running should have DNS configured to resolve the Hostname.

 

Now login into your web-client if the previous step completed successfully.

 

Setup3.jpg

 

Next we need to register our UCS domain(s) with the Web Client.  Click on the Cisco UCS and then click on “Register” at the bottom of the screen.

 

Setup4.jpg

 

We'll have to enter the details once we click on Register.

 

Register.jpg

 

If the registration was successful we will be able to see the registered domain in the vSphere Web client. Repeat the steps if you have more than one UCS Domains.

 

after register.jpg

 

Now the registration is done. We will see how to navigate and see the components of UCS Equipment in next few steps.

 

In the Web Client, click on vCenter -> UCS Domains


Setup6.jpg


we should be able to see the domain registered in the web client screen


show.jpg

Now you can see all the servers in every chassis


Setup10.jpg


We will be able to take kvm console for individual blade servers. Launch UCS Manager and other functions.


kvm.jpg

 

Apart from modifying things related to service profiles we can do all the management tasks like taking KVM console and monitoring the errors.

Hopefully Cisco lets us do more tasks using this plugin in future releases.

 

Do install this plugin and explore.

Abhilashhb Virtuoso
VMware Employees

VDI Sizing Tools

Posted by Abhilashhb May 28, 2013

While Sizing a VMware VDI Infrastructure, we need to take a lot of things into consideration. What is the number of total desktops that are going to be provisioned, how many desktops will be used concurrently, what's the IOPS requirement, What's the Bootstorm IOPS blah blah blah blah.

 

It takes a lot of calculations to complete the sizing and its a tedious work. After all we are humans, we are known for committing mistakes. Take for example you made a mistake at the beginning of the calculation and you didn't realize it. Imagine if you'll have to redo the whole thing. Even if you got through the whole process without committing mistakes and it turned out to be a good solution, what if another requirement comes for more users next time? More users = complex calculations. What would you do then?

 

Here's a tool developed by Andre Leibovici

http://myvirtualcloud.net/?page_id=1076

 

It makes your life easy as it asks you every detail that's possible and should be considered while sizing a VDI Infrastructure.

 

Off course there's a Excel sheet that is available on the VMware Communities but i would recommend it only for initial sizing of the infrastructure as it "assumes" lot of things by default.

http://communities.vmware.com/docs/DOC-18924

 

If you are a storage guy and you are more concerned about the performance of VDI and want to drill down to the IOPS level and see what kind of workloads are generated out of our infrastructure, you can check the link below which lets you work on the IOPS calculation part

http://www.unidesk.com/blog/iops-calculator

 

 

Remember, the way a infrastructure works is not only determined by what products are being used but also the kind of designing that goes into building it.

I have been working on Integration of vCOPS with other devices to explore the capability and the large scope of monitoring vCOPS can perform.

Here i would like to share some steps to monitor Cisco Switches using SNMP Adapter provided for the vCOPS.


Pre-requisites for this configuration


1. vCOPS 

2. Download the SNMP Adapter from here.

3. IP Address,community and SNMP version details of the Switch to be monitored.



Now Login into vCOps Admin UI with a user with appropriate access rights.


Click on Update ---> Browse the location of the Adapter (vcops-SNMPAdapter-X-XXXXXXX.pak) that was downloaded.


Click on Update.


1.jpg

Accept the EULA and Proceed further.


Once the Installation completes. It shows Update is successfull.


Now Login into vCOPS Custom UI with appropriate access rights.


Click on Admin---->Support-----> Info


On the Adapters Info Tab Click on the small icon  screw.jpg

 

Once this is done the SNMP adapter should show  up in the list


2.jpg

Now that the adapter is installed. We need to Add an adapter instance and credentials to start discovering the devices.


Select Environment---->Configuration----->Adapter Instances


Select Collector (vCOPS)


Select SNMP Adapter from the Adapter drop down menu.


adapter.jpg


Now Add the credential for the switches


cred.jpg


Click OK.


Now Discover the resources. Goto Environment overview---->List-----> Auto Discover resources


discover1.jpg

 

Now enter the details like IP range and map the credentials to the Instance as shown below

 

 

discover 2.jpg

 

Once these details are given. vCOPS starts discovering all the devices that are present in this range.

Wait for the discovery results to populate the results of the discovered resource.

 

 

After addning.jpg

 

Click on Import to just discover and not collect any data and collect to start collecting the data.

Click OK

 

Now all the devices will show up on your Custom dashboard widgets.