- July 19, 2012
- Posted by: Tamir Korem
- Category: Uncategorized
h3. Your Application is Installed on a Cloud. Now What ?
Ok, so you installed an application on a cloud and now you want to update it, replace your .war file, modify the DB schema etc.
I guess, some of the following thoughts may run in your head : “Now I need to login to the VM on the cloud (via ssh)”, “Where’s my Key-Pair file? Was is pem or ppk?”, “Do I have permissions to change files on the cloud?”, “I wish there was a way I could automate this process”.
In this post, I will walk you through Cloudify’s Custom Commands Mechanism.
“(newwindow)Cloudify Custom Commands”:/guide/2.1/developing/custom_commands enable admin users or third-party tools to remotely manage service instances, which is especially useful when remote service APIs aren’t available.
They can also come in handy in cross-service manipulations, where one service manipulates another service in the application, thus enabling static applications to become much more dynamic without replacing the stack.
All you need to do is add a few lines to your recipe file and you’re good to go.
h3. How to Create Custom Commands
Cloudify provides this mechanism by allowing you to describe Custom Commands using Groovy closures, external Groovy scripts, or external shell/batch scripts.
The commands can be parameterized, and they can be invoked by using the Cloudify shell or automatically by the lifecycle event handlers when certain conditions are met.
You need to add a @customCommands@ section in the main section of the recipe’s service file.
The @customCommands@ section should contain an array of custom commands and their descriptions.
Each custom command must be in the following format (syntax):
*@”customCommandName”@ : groovy closure code or a script to run*
The following code snippet demonstrates several general examples:
h3. Custom Commands Use Cases
In general, custom commands can be very useful in any of the following:
# Continuous deployment (e.g. updating a war file)
# Reconfiguration of services (e.g. changing a db schema, or apache load balancer configuration or re-running a chef-client)
# Triggering an arbitrary operation on the service (e.g. invoke a load command on the apacheLB)
h3. How to Use It
In order to invoke a custom command from the Cloudify shell, run the following command:
*@invoke serviceName customCommandName [param1] [param2] … [paramN]@*
Custom commands can also be invoked by another service as described in the following code snippet and in the examples below.
Here are some helpful custom commands that we have already implemented in our sample recipes:
h4. Example 1: Dynamically Updating Your Apache Load Balancer
We added “(newwindow)two custom commands:”:https://github.com/CloudifySource/cloudify-recipes/blob/master/services/apacheLB/README.md#registering-a-service-instance-to-the-apache-load-balancer to the “(newwindow)apacheLB”:https://github.com/CloudifySource/cloudify-recipes/tree/master/services/apacheLB recipe: addNode and removeNode, which update the configuration of the apache load balancer module with new/removed nodes members.
Here’s the code snippet in the apacheLB-service.groovy :
As you can see, in apacheLB-service.groovy, there are two custom commands: addNode and removeNode. You need to add a postStart lifecycle event to each service whose instances you want add to the load balancer configuration. You need to add a postStop lifecycle event to each service whose instances you want removed from the load balancer configuration.
For example, these commands are invoked during the @postStart@ and @postStop@ “(newwindow)lifecycle event handlers”:/guide/2.1/developing/lifecycle_events
,of the tomcat service in the “(newwindow)PetClinic Recipe”:https://github.com/CloudifySource/cloudify-recipes/tree/master/apps/petclinic.
The following presentation illustrates how the Custom Commands Machanism works with Apache and tomcat, from the installation of the services to modifying the @apacheLB@ configuration by invoking custom commands:
h4. Example 2: Testing Your Load Balancer Under Heavy Loads
On every machine on which apache is installed, you will also find the “(newwindow)ab command”:http://httpd.apache.org/docs/2.0/programs/ab.html which is a server benchmarking tool provided with the Apache HTTP distribution.
In the @apacheLB@ recipe, you can find a @”load”@ custom command that uses @ab@, thus enabling you to test your application under load, by using the recipe’s “(newwindow)load”:https://github.com/CloudifySource/cloudify-recipes/blob/master/services/apacheLB/README.md#load-testing custom command.
Here’s the @”load”@ code snippet in apacheLB-service.groovy :
h4. Example 3: Updating Your Application
Let’s say, you installed a tomcat web server and deployed your war file on it (with Cloudify).
Now you have a newer version of the war file.
All you need to do is to run the following custom command:
*@invoke tomcat updateWar NEW_WARFILE_FULL_URL_PATH@*
Here’s the @”update_war”@ code snippet in @tomcat-service.groovy@ :
h3. Oh, And One More Thing
As always with Cloudify, it works on every cloud and every OS: EC2, OpenStack, Rackspace you name it, Windows and Linux (Centos or Ubuntu).
So don’t hesitate to give a test drive and let us know what you think.