Monitor the vSphere API with PERL using nagios or icinga

I spend all day every day doing development against the vSphere API. It is crucial that we know when the API is down. If this is something you need to know about you have many choices out there to do your monitoring. Some names that come to mind for me are Nagios and Icinga. In the past icinga was something I used to take care of this task. In this post I will discuss how you can do the same thing and I will provide the PERL code required to take care of this task.

First it is important to note I will not be discussing how to setup nagios or icinga, that is beyond the scope of this post, so make sure you already have it setup and are familiar with how to do configuration tasks with it. With that out of the way lets get started.

Monitoring the API is a pretty simple task. I have found that all it takes is something as basic as logging in and checking the time the vSphere API returns. If you are unable to preform this basic task there is no chance any of the advanced features in the API will work. To get this task done you really only need to do 3 steps.

  1. Login
  2. Obtain a ServiceInstance
  3. Call the CurrentTime method on the ServiceInstance

Thats it. If you can do this your API is up and usable.

VMWare provides the vi-perl tool kit which makes the code pretty simple to write, and the code I provide at the end of this article depends on it to function. Now you just need to configure Nagios or Icinga to run that code at what ever time interval you deem fit. For me I like to know right away so I always had it run every 3 mins and alert me on any failure. As promised here is a link to the code.

How to contribute to an opensource project like a boss

So you found a cool project online and its opensource and now you want to give some of your time and efforts back to the community, but you dont know where to start. This blog post should help get you started and help you get your code accepted on just about any project out there.

The first step you should take is to read the documentation that comes with the project. A great place to start is often the README in the project. Generally you will find all kinds of info in this file. Some of that information may be code standards, where to find help, TODO lists, or just general project guide lines.

Next get familiar with the tools used in the project. This could mean having to learn to use a new revision control system such as git with a focus on using it via github or bitbucket. Many projects have some special requirements when it comes to their source code. You do not have to be a master with these tools, but you should at least understand how to do some basic tasks with them. For example, with projects using git and github you should understand forking, branching, fetching, merging, pushing, and most importantly pull requests. Its OK to ask for help getting these things done, and you can generally find where to get that help in the README.

Next it may be helpful for you to look at the rest of the code in the project, or at least some of it. I know this may be a lot of work because some projects are so large. Maybe you don’t need to see every line of every piece of code but it will help to look at some of it to get an idea of how the rest of the code is written so that you do not camelCase where the rest of the project uses underscore_syntax.

Next you should be open to community feedback. Opensource development generally speaking works differently than what you may be used to doing at work. Most projects I have worked on have some kind of peer review system where community members and projects owners will look at your code and make sure it meets the code standards, and has logic in it that is easy to follow and makes sense. These reviewers may ask you to make changes, but don’t take it personally. This feedback should not be taken as a personal attack. Remember you are going to donate this code and you may never come back again, so it will be up to other community members and project owners to maintain it. You may be doing this on your free time, but odds are that so are some if not all of the other people reviewing your code. Generally speaking no one will ask you to make changes to your code that they wouldn’t expect of their own code.

On the topic of changes no one is forcing you to make these changes. Generally speaking it is just an ask. If you don’t want to make the changes kindly tell the reviewers something like “Im sorry I am not interested in making these changes, please feel free to take my code if you can”. Most projects would rather have working code that needs to be “reformatted” than have nothing while others may not be interested in your code if you don’t make the changes.

Finally and this may seem very obvious but try to write good, well documented, clean code that is easy to follow and understand. I have been doing this for almost 15 years now and I have never run into some code where I said “Damn this documentation is way too good, and there is just way too much of it”, and I am willing to bet I never will. Often it is the opposite of that and I am cussing as I read through line by line trying to figure out how or why something is the way it is.

If you follow these steps, or suggestions as they really are, it should help the process be very smooth and pain free for all parties involved. If you have any suggestions about things that I should add to this list please feel free to let me know. I would love your feedback. I hope this blog post will be helpful to someone out there who is thinking about getting into working on opensource projects.

Making the id field start at 1000 instead of 1 in a Grails domain class

Recently I wanted to make a domain class in a grails app and have the built in id field not start at 1. I wanted it to start at 1000. This was for a generic “Device” class I plan on extending with specific devices like Virtualmachine and Hostsystem devices. I wanted these devices to have numbers starting with 1000 for no reason other than ‘just because’. I did some searches but couldnt find many simple explanations so once I got a working solution I thought I would share what I used.

Ive been working on a tool I will be talking about in future posts, and will only talk about in this post for the code used to make our id generator work. If you look at my Device.groovy you will see exactly what to do.

static mapping = {
id generator: "", params: [initial_value:1000, increment_size:1]

Here I select the type of generator to use, then in the params I set an initial_value of 1000 You can set this to be anything, next I tell it to increment by 1. I found this documentation helpful in my quest: I hope you find this post helpful.

How to install Python 2.7 on CentOS 6.x

I needed to install python 2.7 on a CentOS server. I did some searching and found some very broken scripts on github. I picked one of them and started hacking til I got it working. You can now find it here: centos_python_env_setup To use it, you can simply grab the raw version using wget, set the script executable with chmod, then as root run it. I tested this script about 25 times using my Rackspace Cloud server and picking the CentOS 6.4 option. Please let me know if you have any issues running it.

Simple Tomcat Hosting using Tomcat 7, Java 7, and httpd with mod_proxy_ajp

Ive been spending a lot of time in the last year and a half dealing with Apache Tomcat. The DevOps team I am on focuses on Grails for almost all development needs. We use Tomcat 6 and 7 to host a variety of applications we have written. As such I have learned a lot about Tomcat, and Im going to use this opportunity to share some of the things I have learned with you.

The server setup

I like using CentOS on my personal stuff, so for this example Ill be using CentOS 6.3 and using my Rackspace Cloud Server. Im just using the default image provided by Rackspace. From the cloud server control panel I name my server tomcat-dev1. Next I selected CentOS 6.3 for the image and 512M of RAM for the server size. Finally in the networks section I leave it defaulted to a public and private adapter, and then press the create server button to begin building my new cloud server. Keep in mind this is just a development server for me so I dont need much RAM, for a production system you will need to calculate your needs accordingly and make your server as large as you need it. Once my server is done building I will log on and install:

  • httpd
  • jdk 7 from Oracle
  • Tomcat 7
  • git (optional, but I like to keep my config files in version control)
  • jsvc (this is bundled with tomcat)

First steps

Lets begin by logging into our new cloud server as root. The first thing I do to a new cloud server is change the password. Once that is done we can begin by updating the system.
yum -y update
This is done because the base image could be several months old so we want to get the latest updates from the vendor before we begin. Once that finishes reboot if you need, if you do not know if you need a reboot or not just do it anyway to be sure.

Gathering the goods

Some of what we will be installing is either not provided by yum, or is older and we want flashy and new. We need to fetch the jdk; I will download the x64 rpm to my local desktop then upload it to my cloud server using winscp. Next we need Tomcat, and jsvc. Since these two things do not require me agreeing to some terms I will download them directly from the server.
Now on my server I should have the jdk rpm, and the Tomcat 7, and jsvc packages.(jsvc is bundled with tomcat)

Installing the jdk, Tomcat, and jsvc

We need to install the jdk first. To do that lets install the rpm:
rpm -Uvh jdk-7u10-linux-x64.rpm
If you see output like this:

Error: Could not open input file: /usr/java/jdk1.7.0_10/jre/lib/rt.pack
Error: Could not open input file: /usr/java/jdk1.7.0_10/jre/lib/jsse.pack
Error: Could not open input file: /usr/java/jdk1.7.0_10/jre/lib/charsets.pack
Error: Could not open input file: /usr/java/jdk1.7.0_10/lib/tools.pack
Error: Could not open input file: /usr/java/jdk1.7.0_10/jre/lib/ext/localedata.pack

It is safe to ignore. I didnt dig into why this output happened but it did not seem to affect my install. Now that Java is installed we need to add the java install to the alternatives system, and make it so that the JAVA_HOME environment variable is set for all our users when they log in. Lets do that by creating a file in /etc/profile.d called Using your favorite text editor create the file /etc/profile.d/ and enter the following lines into the file and then save it:

export JAVA_HOME=/usr/java/default
export JRE_HOME=/usr/java/default/jre
export JAVA_OPTS=””
export JAVA_CLASSPATH=/usr/share/java/*


Now the file needs to be set executable so issue the following command:

chmod +x /etc/profile.d/

Now to add java to the alternatives system

alternatives --install /usr/bin/java java /usr/java/latest/jre/bin/java 20000
alternatives --install /usr/bin/jar jar /usr/java/latest/bin/jar 20000
alternatives --install /usr/bin/javac javac /usr/java/latest/bin/javac 20000

Lets test that this works by typing:


Right now we should get no output. Now try logging off our server, and logging back in. Once you do that type the command again. We should now see:


Now we need to unpack Tomcat, and the jsvc daemon.

tar xzvf apache-tomcat-7.0.34.tar.gz -C /opt/

This will unpack Tomcat into the /opt directory into a folder named apache-tomcat-7.0.34 Once that is done unpack the commons-daemon source code. That code is bundled with tomcat:

cd /opt/apache-tomcat-7.0.34/bin/
tar xzvf commons-daemon-native.tar.gz -C /usr/local/src/

The source code for the jsvc daemon is now in /usr/local/src/commons-daemon-1.0.10-native-src and needs to be built. Before we can build it we still need to install a couple of things on our server. According to the docs we need a compiler, and make. In an effort to make it simple I will be using yum and the groupinstall option to install “Development tools” This will provide git, make, gcc, cpp, and many other tools.

yum -y groupinstall "Development tools"

Once that is complete we have what we need to build jsvc, so lets do that now. Since we are using CentOS we will want to:

cd /usr/local/src/commons-daemon-1.0.10-native-src/unix

From here we can issue the configure command:

./configure --with-java=/usr/java/default

This should render some output, the last thing it says if you were successful is this:

*** All done ***
Now you can issue “make”

Lets do that now.


If you got no errors you should now have a binary file in the same directory as you called jsvc. Since this Makefile does not have a target for install the next natural step of make install would fail. You have to manually place the binary somewhere on your system. I like to just put it with the tomcat binaries since thats the only thing I use jsvc for, so thats what Ill do now:

cp jsvc /opt/apache-tomcat-7.0.34/bin/

You should now have Java 7, Tomcat 7, and jsvc 1.0.10 installed on your system. We are getting close to being done. Only a few more steps left!! :)

Doing some Tomcat configuration, and making Tomcat start at boot

Now that we have all these things installed we need to do some configuration, and also make it so Tomcat will start when the system boots up. Lets begin with making tomcat start at boot. Im going to provide you with the init script I wrote and use on my systems. You can get the latest version from here. Download it and save the file in /etc/init.d as tomcat

wget -O /etc/init.d/tomcat

This file needs to be executable so lets issue the following command:

chmod +x /etc/init.d/tomcat

Now we need to make sure tomcat will start at boot:

chkconfig --add tomcat && chkconfig tomcat on

This will add tomcat to the chkconfig system, and enable it for start up on boot. Lets take a look at what this file does for us. On line 4 we see this:

# chkconfig: – 85 15

This is needed for chkconfig so our command above would work. For more information see man chkconfig. Scroll down in the file until you see:


This is where the user that tomcat will run as is defined. By default it wants to run as the user named tomcat. We have not created that user yet, but we will shortly. The next two things to notice are right below the TOMCAT_USER line.


These locations on the file system are not there by default so we need to add them, and make sure our tomcat user has access to them, but lets keep looking through this file for now and we will fix all the things at once. Lets scroll down to lines 20 and 21

[ -f /etc/profile.d/ ] && . /etc/profile.d/
[ -f /etc/profile.d/ ] && . /etc/profile.d/

This is checking to see if these files exist and if so source them, this will set up our environment variables for us. The tomcat file is missing still, but we will add it soon. I think one of the last things to note here in this file is the JSVC_BIN variable. If you put jsvc in some other location than where I put it then you need to adjust this to your location. Finally on line 62 we have the command we use to start tomcat. You should not need to adjust this, but if you need to thats where to do it. Now that we know what this file will do for us, lets fix all the things we found while going through it.

Finializing the Tomcat bits

The first problem we found above was our script is looking for a “tomcat” user and we dont have one yet, so lets add one:

useradd -r tomcat -m

This added a new system user named tomcat, and created a group also named tomcat. A home dir was created in /home/tomcat The next problem we found was that a couple of files need to be created, and our tomcat user needs access to them. Lets make it happen:

mkdir -p /var/lock/subsys/tomcat/
mkdir -p /var/run/tomcat/
chown -R tomcat. /var/run/tomcat
chown -R tomcat. /var/lock/subsys/tomcat

Finally the file needs to be created and make executable. Using your favorite text editor create the file /etc/profile.d/ and add the following to it, then save it:

export CATALINA_HOME=/opt/tomcat
export CATALINA_TMPDIR=/opt/tomcat
export CATALINA_OPTS=”-XX:MaxPermSize=256m”

Now we need to make it executable:

chmod +x /etc/profile.d/

You may have noticed that the location of CATALINA_HOME and CATALINA_TMPDIR dont exist on our server, good catch. I use symlinks to this. Lets get them created.

cd /opt
ln -s /opt/apache-tomcat-7.0.34/ /opt/tomcat

I do this so if I update tomcat I only have to update this symlink and do not have to edit my init script, or my profile.d script. Now all we have left for tomcat is to make sure we have the server.xml file created correctly, and if we want to allow access to the manager we need to make a tomcat-users.xml file as well. I have provided the server.xml and the tomcat-users.xml file I used here. Those files need to be saved in /opt/tomcat/conf/ as server.xml and tomcat-users.xml Once you have done that you should be able to start tomcat, so lets try that now.

service tomcat start

Lets verify its running:

ss -nap

We should see some output like the following:

[root@tomcat-demo unix]# ss -nap
State Recv-Q Send-Q Local Address:Port Peer Address:Port
LISTEN 0 100 ::ffff: :::* users:((“jsvc”,20103,39))
LISTEN 0 100 ::ffff: :::* users:((“jsvc”,20103,38))

This shows us that jsvc has a process listening on port 8009 and 8080, on the local loop back interface only, just like we setup in our config file. This is wonderful now we are done with tomcat for now. Lets move on to httpd and setting up the ajp.

Apache httpd and mod_proxy_ajp

We need to install httpd now.

yum -y install httpd && chkconfig httpd on

We do not need to do anything to get mod_proxy_ajp it is part of the base httpd install as of 2.2. Now that our web server is installed lets configure our first vhost, and serve up some tomcat content. Using your favorite text editor open the httpd.conf file located at /etc/httpd/conf/httpd.conf The vhost section is at the bottom of the file by default. In my case I added the following:

NameVirtualHost *:80
<VirtualHost *:80>
ProxyRequests Off
<Proxy *>
Order deny,allow
Allow from all
ProxyPass / ajp://localhost:8009/

Next I need to open port 80 on my firewall, since the default firewall will be blocking traffic on port 80. Using my text editor I will edit /etc/sysconfig/iptables so it looks like so:

# Firewall configuration written by system-config-firewall
# Manual customization of this file is not recommended.
-A INPUT -p icmp -j ACCEPT
-A INPUT -i lo -j ACCEPT
-A INPUT -i eth0 -m tcp -p tcp –dport 80 -j ACCEPT
-A INPUT -m state –state NEW -m tcp -p tcp –dport 22 -j ACCEPT
-A INPUT -j REJECT –reject-with icmp-host-prohibited
-A FORWARD -j REJECT –reject-with icmp-host-prohibited

Next I need to load this new rule set like so:

iptables-restore < /etc/sysconfig/iptables

Now I am done, and Im ready to fire up httpd and visit my site.

service httpd start

You can now visit your site in a browser and see the default Tomcat landing page, as well as add a /manager and log in using "tomcat" as the user, and "s3cret" as your password. In the coming weeks I will be adding more info about how to do vhosting using this method, and eventually covering how to do this all in a few mins using chef

SYSPRP Parse Commands:No action flag was specified

Ive been doing a fair bit of work around the vmware vix api. One of the things I needed to do was run sysprep after a clone is taken, and beofre the vm is brought on to the network and rejoined to AD. I decided to use vix to do it. Getting the command to run using vix is simple enough with VMRunProgramInGuest(). The issue I ran into was with the parameters I was passing to sysprep.exe

Sysprep will log into:


In those files is where I found the error:

SYSPRP Parse Commands:No action flag was specified

I was leaving off the Action flag clearly, but what are valid Actions? It wasnt clear to me until I ran sysprep with out any command line args and saw the gui. Then you can see that oobe and audit are your 2 options. oobe best fit my needs. Once I added that my error was corrected and sysprep would successfully run

Install vmware tools on Debian

Just a quick post. If you are running Debian 6 as a virtual machine and want to get the vmware tools installed in the vm its pretty simple. I recommend using the open-vm-tools package (I recommend using these for ALL Linux distros). In Debian 6 this package is provided in the contrib repo. To enable access to this repo you simply edit (as root or sudo) your /etc/apt/sources.list file and add contrib to it like so:

deb squeeze main contrib
deb-src squeeze main contrib

deb squeeze/updates main contrib
deb-src squeeze/updates main contrib

# squeeze-updates, previously known as 'volatile'
deb squeeze-updates main
deb-src squeeze-updates main

Once you save, still as root or sudo simply run:
apt-get update
Once it finishes you can run:
apt-get install open-vm-tools.
Thats all there is to it.

Install yum on your vmware vma

DANGER!! Please note this will likely null and void any kind of warranty or support you may have on your vma, and it will also break vima-update depending on what you allow yum yo update, it might even cause the world to explode.. I dont know and Im not responsible for anything that goes wrong following this..

With that out of the way.. So you want to install yum on your vma so you can install some packages on your vma, and you dont want to track down rpms for it.. I had the same desire. Here is how I took care of the problem.. First I found that /etc/redhat-release stated I was using a Red Hat Enterprise Linux Server release 5.2 (Tikanga). Once I found this out I decided to use CentOS mirrors to get me a copy of yum and to update rpm and such so I could have yum.

First the base vma is missing some stuff in order to just grab yum and start using it. I found mine to be missing python-initparse, yum-fastestmirror, and yum-metadata-parser. For me I got a root shell using sudo -s, next I ran the following command from my root shell:
# rpm -Uvh

This installed yum and the missing libs I needed to use it. Next I had to create a yum.conf repo file that the system can use to pull packages from. I used the following:


# Note: yum-RHN-plugin doesn't honor this.

installonly_limit = 5

# PUT YOUR REPOS HERE OR IN separate files named file.repo
# in /etc/yum.repos.d

name=CentOS-5.2 - Base

#released updates
name=CentOS-5.2 - Updates

Please note the repo info like the file states can be added to a .repo file if you want, but to shorten this for me Im doing it all in 1 place. I would also suggest going though and adding things like kernel* to the excludes= section (that is really beyond what Im trying to cover here though). Once you have this minimal info in your file you can save it. Next you need to import the RPM-GPG-KEY so you can install files.. As root:

rpm --import

Now you can update things like rpm, or even install vim-common or emacs :) You could also go a step further and install the epel repo and then install puppet so if you are in a large scale environment you can keep the system updated and files in sync with all your other hosts.

Using psphere and Python to automate your vSphere tasks

For almost the last year I have been working heavily with vSphere automation. Recently I have found that there is a way to do a lot of the tasks with Python, taking advantage of a fairly newish python library called psphere. I have created a github project where I am keeping tools I make outside of work. These tools do common tasks that administrators and engineers who manage small and large scale environments might find useful.. Its a new project and I should be adding things to it as often as I can. There you will find tools written in Java and Groovy/Grails using vijava. In time Perl scripts using the VMWare provided Perl SDK will show up, and there are already some Python scripts in there using the psphere library. I hope that you will fork the project and add your own stuff to it and share it back with everyone. The github project I started can be found here:

In my first example I am going to show you how to use a few lines of simple python code along with psphere to get the total number of hosts in a vCenter. This is a handy report that you can send out to your management, or even put into some graphing program so you can watch for trends in your environment. If you run vCenter in linked mode you only need to hit the master vCenter to get ALL the hosts for that vCenter cluster.

Lets being with setting up our system. Im using an Ubuntu 11.10 server setup. I did a minimal install and selected OpenSSH Server for my server type. Once the system finished installing I logged in and and got the system up to date. I am of course running this Ubuntu server as a Virtual Machine, so the next thing I want to do after I install the base system and get it updated is to install the open-vm-tools package. Once that is done I reboot. Now I move on to setting up my python environment. Psphere requires Python Yaml, and Python Suds, and Python pip, once these packages are installed you can move on to the pip install -U psphere And now your environment should be all ready to go for coding. Not to much to it. This should also work on Windows. For a reference Ubuntu 11.10 comes with Python 2.7.2, Suds 0.4.1, and Yaml 3.10, and at the time of this I am using psphere 0.5.1 As a note I first tried this using Debian 6.04 with Python 2.6.6 and it did not work, so for best results I can say I got it working with the above Ubuntu configuration..

Now lets cover the code it takes to connect to vCenter, and print all the hosts and at the end give the total number of HostSystem objects stored in this vCenter. Please note this code was tested on both 4.1 and 5.0 vCenter systems.

Open your favorite text editor. and enter the following:

#!/usr/bin/env python

from psphere.client import Client
from psphere.managedobjects import HostSystem
from getpass import getpass

usr = raw_input("vCenter user name: ")
passwd = getpass(prompt="vCenter password: ")
server = raw_input("vCenter server hostname or ip: ")

client = Client(server,usr,passwd)
hs = HostSystem.all(client)
for host in hs:
print "There were: " + str(len(hs)) + " HostSystems found in vCenter"

(Please note in the above code the indent is wrong so dont copy and paste, for some reason my wordpress theme is messing with the indents)
The code is pretty simple and wouldnt take much to modify to output into a csv with some more properties. You could also make it so you did not need to be prompted for user names and passwords or vcenters and just pass them in as an arg, or use the config file ~/.psphere/config.yaml

Switching Falconstor IPStor to use 1 to 1 mapping for LUN assignment

Switching your Falconstor IPStor server to use 1 to 1 mapping from any of the other avaliable mapping options Like All to All or All to 1 can be an annoying task, but if you have VMWare ESX or ESXi and have VMotion you can do this with 0 down time and Ill show you how. These steps will assume you are NOT in a fail over cluster with Falconstor

In my example I am using 2 ESX 4.1 hosts. Monster01 and Monster02. Each one of my hosts is designed to support running all our virtual machines from so moving all the vms to a single host like this is not going to have a negative impact on any vm and users will never know this is going on.

First Im going to need to vmotion all the machines to a single host. I move all the virtual machines running on Monster01 to the host Monster02. Once this is done I put Monster01 in maintence mode, then shut it down. This step is only needed if you are SAN booting your ESX/i host and need to change the LUN the ESX/i OS is running on.

Once its shut down comes the hard part if no one ever documented what WWPN is used for what. If that is the case for you like it was for me then you can figure this out by looking at your physical adapters and finding all the adapters that are in target mode. Write down the WWPN of each one. You might have a bunch.. Thankfully I only had 2 adapters in terget mode that were online. Yay!! a 50/50 chance of getting it right on the first try! I took LUN 0 which has my ESX 4.1 install on it and did a 1 to 1 map. On the ESX host this was simple since we only had 1 adapter plugged in and configred. Next I had to take a shot in the dark because our cables are a mess and no one documented what WWPN was used for what. I got it wrong the first time because once I assigned the LUN and turned the host back on the box failed
to boot. I switched to my other choice and Bam it booted!

Next to quickly switch the other 40 LUNS I went to the Falconstor management console and went into the SAN Clients and selected Monster01 I right clicked on each LUN and selected properties. From here you can switch from your current mapping to 1 to 1 using a select box. You will get a warning about data transmission stopping when you do this. That is fine since there are no running virtual machines on Monster01 (remember its still in maint mode) Next you select the 2 WWPNs needed for the initiator (the ESX host) and the target (the Falconstor server). Once you have completed this for all LUNs you can bring the ESX host out of maint mode and VMotion the machines from Monster02 to Monster01 and then repeat this process on Monster02.

Be on the lookout for useless Information