WilliamLam.com

How to Automate the Deployment & Configuration of vShield Manager 5

by // 8 Comments

If you have ever worked with VMware vShield Manager, you know that deployment and configuration of the virtual appliance is pretty much a manual process. You can automate the deployment of the vShield Manager OVA using the various vSphere SDK's or the ovftool, but the initial IP address configuration for vSM still needs to be configured manually using the remote console for the very first time.

An easy solution to this problem would be for VMware to create the vSM OVA to support IP address configuration out of the box as part of the deployment options (but why make things easy). In any case, I will demonstrate how you can easily automate both the deployment and the initial configuration of vShield Manager to your vSphere environment.

Before I begin, I can not take credit for coming up with the idea of automating vShield Manager deployment, the credit goes to Alan Renouf. Alan recently contacted me and ask if it was possible to automate the IP configuration. The answer is yes and here is a solution.

One of the main challenges in figuring out how to automate the IP address configuration of vShield Manager was due to the vtysh integrated shell daemon for Zebra that launches by default as part of the "admin" user account. This interface is used to manage the kernel routing and management table and made it very difficult to interface with for any type of automation. I decided manually go through a vSM configuration and then using Knoppix LIVE-CD, I was able to mount up the vSM filesystem and look around to get a better understanding of what was going on. After some investigating, it looks like IP address configuration is stored in /common/configs/cli/zebra.conf, here is an example of what the configuration looks like:

Armed with this knowledge, it was pretty straight forward in developing an automated way of deploying and configuring vShield Manager. I created a script called deployvShieldManager.sh which utilizes guestOpsManagement.pl, vCLI and ovftool. It's recommended that you use vMA and install ovftool to quickly get started. At a high level, the script is doing the following:

  1. Deploy vShield Manager OVA using ovftool
  2. PowerOn vSM and wait 2 minutes for VMware Tools to be ready on the system
  3. Create new zebra.conf, backup the old zebra.conf and upload new zebra.conf using the new vSphere 5 VIX integration
  4. PowerOff vSM to force the configurations to be read in upon next bootup
  5. PowerOn vSM and it is now ready for use

At the top of the script, there are several configuration variables that need to be edited by the user to specify the vSM configuration, including vCenter and ESX(i) host to deploy vSM.

Here is a list of variables that need to be configured at the top of the deployvShieldManager.sh script:

Once you are done updating the variables, you are now ready to execute the script. Before the script performs any changes, it will first prompt the summary of configurations you have specified in the script. Once you are satisfiy, you may than proceed by typing "y" or "yes" to start, or if you would like to cancel, type "n" or "no".

Note: The script will perform some basic validation such as existence of the vShield Manager OVA, ovftool, etc. else you will get an error message and the script will exit.

Next, the script will perform the deployment of vSM using ovftool and proceed with the configuration of vSM once it has been deployed.

Note: If it takes longer to poweron vSM in your environment to get it into a ready state, you may want to tweak the sleep period from 120 seconds (2minutes) to something longer.

At this point, you now should see a new vShield Manager VM deployed and if you take a look at the summary page, you should see the new IP address and hostname configurations.

Now all that is left is to point your browser to the vSM address and you should be prompted to login to vShield Manager management interface.

Instead of manually deploying vShield Manager in your environment, you can now automate the initial deployment and configuration for general use or with VMware vCloud Director. For further automation and configuration of vShield manager, once vSM is online and accessible, you can leverage the vShield REST API.

How to Send vCenter Alarm Notification to Growl

by // 2 Comments

This tweet from Jason Nash and @PunchingClouds says it all and here it is!

I did some research this afternoon and stumbled upon this article Nagios: Notifications via Growl and leveraging the Net::Growl Perl module, I was able to forward alarms generated from a vCenter server to a system that was running Growl.

Software Requirements:

Step 1 -  Install Grow and configure a password under the "Security" tab and ensure you "Allow network notification"

Step 2 - Install vSphere SDK for Perl on your vCenter server. You may also need to update the PATH variable with Perl bin directory (e.g. C:\Program Files\VMware\VMware vSphere CLI\Perl\bin)

Step 3 - Install Net::Growl Perl module using ppm (Perl Package Manager) which is part of ActiveState Perl with the installation of vSphere SDK for Perl. This will require your vCenter server have internet access to ActiveState Perl site, if you can not get this access, you can install this locally on your system and extract the Growl.pm and copy it to your vCenter server C:\Program Files\VMware\VMware vSphere CLI\Perl\site\lib\Net

Step 4 - Copy the Perl script from here and store it somewhere on your vCenter server, make sure it has the .pl extension. In this example, I named it growl.pl

Step 5 - To verify that Growl Perl script works and can communicate to the system with Growl install, you can manually test it by running the following command:

growl.pl -H william.primp-industries.com -p vmware -a custom -t Alert -m "hello william" -s 1

You will need to change -H to the hostname or IP Address of the system with Growl installed and of course the password you had setup. You should see a notification of the message you had just sent.

Step 6 - Create a batch script which will call the growl.pl script and store it somewhere on your vCenter server. Here is what the script (sendGrowl.bat) looks like, you can modify it to fit your requirements.

:: http://www.virtuallyghetto.com/
:: Custom vCenter Alarm script to generate growl notifications

set GROWL_SERVER=william.primp-industries.com
set GROWL_PASSWORD=vmware
set GROWL_SCRIPT_PATH="C:\Documents and Settings\primp.PRIMP-IND\Desktop\growl.pl"
set PATH="%PATH%;C:\Program Files (x86)\VMware\VMware vSphere CLI\Perl\site\bin"

%GROWL_SCRIPT_PATH% -H %GROWL_SERVER% -p %GROWL_PASSWORD% -a %COMPUTERNAME% -t Alert -m "%VMWARE_ALARM_EVENTDESCRIPTION%" -s 1

Note: If you would like to get a list of other default VMware alarm variables, run the "SET" command and output it to a file to get more details on various variables that can be accessed.

Step 7 - Create a new or update an existing vCenter alarm and under "Actions", specify "Run a command" option and provide the full path to the sendGrowl.bat

Step 8 - For testing purposes, I created a new alarm that would trigger upon an ESX(i) host going in/out of maintenance mode and you can see from the "Tasks and Events", our script is triggered on the vCenter server

and now for the finale, you should see a notification from Growl on your system and since we enable the "sticky" parameter, the notification will stay on your screen until you click on it. You can see that in the script example, I set the message to the event and application is registered as the name of the vCenter server, which allows you to have multiple vCenter forward you notifications.

So there you have it, forwarding vCenter alarms to Growl.

Note: Once a vCenter alarm has been triggered, the script will not fire off again until the original alarm has been reset to green. This behavior probably is okay for majority of the events one would want to monitor, but if you want it to continuously alert you, you will need to fiddle with a way to reset the alarm on the vCenter server.

UPDATE:  Thanks to Richard Cardona for reminding me, but this can also be implemented on the new VCVA (vCenter Server Virtual Appliance) in vSphere 5. Here are the instructions on setting it up

Step 1 - Install Grow and configure a password under the "Security" tab and ensure you "Allow network notification" on the system that is receiving the Growl notifications

Step 2 - To install Net::Growl, we'll be using cpan which requires 2 modules that are not installed by default on the SLES VCVA. Using the Tips and Tricks for vMA 5 (running SLES as well), we'll go ahead and setup zypper package manager for VCVA to install the two required packages: make and yaml

zypper --gpg-auto-import-keys ar http://download.opensuse.org/distribution/11.1/repo/oss/ 11.1
zypper --gpg-auto-import-keys ar http://download.opensuse.org/update/11.1/ Update-11.1
zypper refresh
zypper in make
zypper in perl-YAML

Step 3 - You will use cpan to install Net::Growl

perl -MCPAN -e shell

Step 4 - Once you are inside the cpan shell, type the following to install Net::Growl

install Net::Growl

Step 5 - Copy the Perl script from here and store it somewhere on your vCenter server (e.g. /root), make sure it has the .pl extension and has execute permission. In this example, I named it growl.pl

Step 6 - To verify that Growl Perl script works and can communicate to the system with Growl install, you can manually test it by running the following command:

vcenter50-2:~ # ./growl.pl -H william.primp-industries.com -p vmware -a custom -t Alert -m "hello william" -s 1

Step 7 - Create a shell script which will call the growl.pl script and store it somewhere on your vCenter server (e.g. /root). Here is what the script (sendGrowl.sh) looks like, you can modify it to fit your requirements.

Step 8 - Create a new or update an existing vCenter alarm and under "Actions", specify "Run a command" option and provide the full path to the sendGrowl.sh

ESXi Migration Worksheet Script

by // 7 Comments

Over the weekend Duncan Epping released a new whitepaper called VMware ESXi 4.1 Operations Guide which documents some of the operational tasks for managing/configuring an ESXi environment. Not only is this a really cool document, but I also found that I was referenced in the whitepaper. This was really neat as it is probably my first reference ever in a whitepaper 🙂

While looking through the other links, I noticed there was another whitepaper written by Duncan's colleague, Kyle and it referenced a host configuration worksheet. When I opened up the document, it was basically a plain PDF form requiring users to manually input some of the important configurations of an ESX host prior to upgrading to ESXi. I thought this would be fine if you only had a few hosts to upgrade, but what if you had several dozens or hundreds of hosts? Manually editing the non-editable PDF document meant you had to use a PDF tool or worse printing it out and writing in the information by hand.

I thought, why not automate it? I decided create a very simple vSphere SDK for Perl script called generateHostConfigurationWorksheet.pl that would allow a user to connect to either a standalone ESX host or to a vCenter server. The script also allows you to specify a list of ESX hosts to extract the information when connecting to vCenter, else by default it will collect information for all your ESX hosts. You also have the option of selecting either an html or csv output.

To run this script, you will need a system that has the vCLI installed or you can use VMware vMA. You will also need to download the script from here. Make sure you set the script to be executable

The script accepts two paraemters:
--output [html|csv] (required)

--hostlist [file-containing-esx-hosts] (optional)

Here is an example of connecting directly to an ESX host:

[vi-admin@tancredi ~]$ ./generateHostConfigurationWorksheet.pl --server vesx41-1.primp-industries.com --username root --output html
Enter password:
Processing vesx41-1.primp-industries.com
Generating vesx41-1.primp-industries.com.html ...

Here is an example of connecting to vCenter and specifying a list of ESX hosts to collect information on:

[vi-admin@tancredi ~]$ ./generateHostConfigurationWorksheet.pl --server reflex.primp-industries.com --username primp --output csv --hostlist hostlist
Enter password:
Processing vesx41-1.primp-industries.com
Generating vesx41-1.primp-industries.com.csv ...

Note: If you like to select the ESX hosts you want to report on, just create a file that has the displayname of your ESX host seperated by a newline

Here is sample report that you can view

You will notice that syslog section is left blank, this is done intentionally as with ESX, syslog configuration is not exposed through the vSphere API like ESXi. This will be something you will need to fill in manually.

I think this worksheet will be useful, but if you are interested in a more extensive report prior to upgrading, I would highly recommend you check out the popular vSphere Health Check script and here is what one of those reports can look like.