Recent comments

All of our internal development and testing is done using the official Sun Java JDKs. We have only used the OpenJDK with very basic java fabric applications on RHEL5 and OpenSuSE. Therefore, we do not ensure that it will work with more robust java fabric applications. You would want to use the Sun version.

Support

Thanks Mark, it's working now!

Hi,

We are trying to deploy an existing WAR we have to the Tomcat service running on the fabric. We create a FAR by using the folliwing xml :

<?xml version='1.0' encoding='utf-8'?>
<far xmlns='http://www.appistry.com/ns/far' 
     name='HPC' 
     version='1.04' 
     display-name='HPC' 
     description='HPC for Windows'>
   <support-files>
      <file name='HPC.war'/>
      <file name='install.bat'/>
   </support-files>
   <service-app service="Tomcat">
      <install>
         <exec executable='install.bat' 
               output='instout.txt' 
               error='insterr.txt' 
               timeout='3 minutes'>
         </exec>
         </install>
      <uninstall>
         <exec executable='echo Do Nothing' 
               output='uninout.txt' 
               error='uninerr.txt' 
               timeout='3 minutes'/>
      </uninstall>
   </service-app>
</far>

Our install.bat simply copies the WAR file to the Tomcat webapps folder :

@echo off
copy "HPC.war" "C:\Program Files\Apache Software Foundation\Tomcat 6.0\webapps\HPC.war"

The WAR file does get copied to the system/fars/ folder on each worker, but does not get copied from there to the Tomcat webapps folder. It is not clear if the install.bat is ever executed.

In the CloudIQ console, the FAR package shows the status : Installation Pending for all workers.

Please help! Thanks.

Great! Thanks for the update.

Ok, that clears up the first problem. You need to have an extension on the file that you are submitting in the uri. Here is the original line:

Uri address = new Uri(WorkerIP + "/files/Elastic-System-Config-File");

Once you resolve that, you will discover another more important issue. The file name provided in the uri must always match the name provided within the contents of the file. Currently, it looks like it is a mismatch of "Elastic-System-Config-File" and "MyFar". That is not supported and will cause issues.

I suggest that you repackage your far with a descriptive name instead of "MyFar.far". For example, you could name it Elastic_System_Config_File. After you package that xml with that name, it will have created an Elastic_System_Config_File.far file.

You can then modify your code to the following:

Uri address = new Uri(WorkerIP + "/files/Elastic_System_Config_File.far");

FileStream fs = new FileStream(@"C:\Elastic_System_Config_File.far", FileMode.Open, FileAccess.Read);

I look forward to your progress after these suggestions.

Thanks,
Mark

Hi wnordmann ,

Is it not building? Is it providing you with an error?

I will ask around to see if there are C# REST API samples, but my first guess is that you would want to include the port in your uri. The default REST port is 16080. Does that solve the issue?

Thanks,
Mark

Richa,

I just confirmed that you would want to make you admin-port within the range of 17000 to 30000 (30000 default). Otherwise, you risk trip over other ports.

Your request port would then be 1000 higher than your admin-port. (default 31000)

If you change your admin-port back to 30000, you should be able to submit transactions on 31000.

Thanks,
Mark

Hi Mark
First of all I am happy to find that the queries are answered promptly with required level of elaboration. Great work!!

I have few queries in line with what Rich had asked.

The tutorial says the following:

" The fabric API needs information listed in the table below to allow it to communicate with the fabric
fabric request port - The fabric MCAST port must match the request-port entry in the addr.cfg file deployed on to your fabric
workers. The default is 31000."

However the only entries in addr.cfg are mcast address and adminport. Is it that if there is no entry fabric_request_port in addr.cfg, the value 31000 be used. I could not find any other file containing 31000.

Also fabric. cfg has all the parameters commented. Do we need to uncomment/add entries?

Please let me know.

Cheers
Siva

Hi Siva,

I'm glad you're evaluating CloudIQ. Let me see if I can answer your questions....

1) Where exactly Cloud iq platform belongs? Is it PaaS, SaaS, fusion of both or an enablement service.

CloudIQ is PaaS. We sit above the infrastructure, and virtualize the cloud applications running above the CloudIQ platform. CloudIQ Manager enables you to deploy (as you saw) pretty much any service onto a cloud of servers. Rather than manage virtual machine or OS images, you need only a basic image with CloudIQ installed and configured. Then you deploy your services and applications through CloudIQ Manager, and it life-cycles them. CloudIQ Engine is a fully-distributed application container. It allows you to create applications from existing business logic (Java, .NET, C/C++, etc.) and run it in the cloud. These two components make up our PaaS offering. but we'll be adding more cloud services in the near future.

2) I suppose cloud iq can be used to move applications to any clouds (private, AWS etc). Is this (FARs) something similar to vApps in VMWare (using OVF)?

Yes, CloudIQ can run on virtual and physical infrastructure in private and public clouds.

Comparing FARs to vApps is a fair start, but they are quite different in implementation. FARs are basically packages of files, along with meta-data and, perhaps, scripts. The meta-data and scripts describe how the files should be installed, uninstalled, and in the case of services, started, stopped, killed, and health monitored. Using this information, CloudIQ Manager can care and feed the services and applications deployed on it. FARs can be used to describe shared libraries, services, configurations for services, and applications for services. Through the FAR meta-data, you describe what versions of these things are required, and what dependencies there are among them. For example, you can create FARs for Apache HTTP server, Ruby, Ruby on Rails, and a Ruby on Rails application. By setting the dependencies among these packages, you can assure that the Ruby on Rails application will not attempt to run unless the correct versions of RoR, Ruby and Apache are present and installed. You can then upgrade the individual components with a single command.

CloudIQ Engine applications are similar to FARs in that they package your application logic (Java, .NET, C/C++) along with meta-data that describes how you want these code libraries to run on the cloud. By doing this, your plain old java objects (POJOs), plain old .NET objects (PONOs), Spring Beans, etc. inherit extreme scale, reliability, and other cloud features without recoding. By having intimate knowledge of the application code (but not inserting our interfaces into that code), we can truly enable that code in a fully distributed environment like the Cloud.

3) Can we use cloud iq platform to provide custom package applications (say SAP etc) as a service? Is it possible for setting up a package application (with reqd configuration) in one worker and then deploy this as a service in all the available workers?

Taking a large architecture such as SAP and deploying it via CloudIQ may not be possible, simply because environments like that typically have static architectures that require static configuration, and do not adapt well to a dynamic environment like the Cloud where servers come and go. However, if a service (or services) can be started, stopped, and health monitored, then typically you could create FARs for them and deploy the environment via CloudIQ Manager. We'd have to look at specific examples of what you are looking to do. I will say that this will become more likely with future versions of CloudIQ Manager.

4) Consider a scenario where a private cloud has to be created and applications/packages installed in the same have to be exposed as Service (SaaS) for end users to use. In my understanding for creating the private cloud, we need to install cloud iq platform on all the machines/servers that are to be in the cloud.Are there any features in cloud iq that we can tap to manage SaaS implementation (along with user provisioning/subscription)

In general, I think my answers above touch on the fact that we give you the leverage to deploy and manage your own SaaS services using CloudIQ Manager, and CloudIQ Engine allows you to build scalable, reliable "cloud-native" applications to scale up business logic (SaaS logic specifically in your case). Currently, we do not offer interfaces for authentication, authorization, and self-service provisioning. However, we have partners in production with SaaS implementations. I'll point you to a case study.

5) Is there any case study/white paper available that says how cloud iq was used to address a specific situation/need. This will be very helpful for us to understand how to effectively use cloud iq.

Yes. Our resource library has a number of white papers and case studies. From an SaaS perspective, you will be particularly interested in the Presidio Health case study and webinar. Both of those, and the rest can be found here: http://www.appistry.com/resource-library

I look forward to talking with you more,

Thanks,

Guerry

Richa,

No problem on the long post. We are happy to help you learn about our software.

I think you are very close to getting your hello world example running. My guess is that you are pointing to the incorrect port. I noticed that you changed it in your steps. The default is 31000, and it is unlikely that you would change or need to change it to something else. However, I notice that a lot of people believe it should be 30000. The client should not point to 30000, and if you are using the defaults, it should point to 31000.

As far as answering your questions directly:
1.) Those are the correct steps. I just think that you just need to change the request port back to 31000. If that port number is an issue, you can change it on your fabric. I can walk you through that if necessary.
2.) Application configuration files are optional. One is not used in the hello world tutorial. An application configuration file is normally used to change parameters in your fabric application without re-packaging the entire application. This way your class files could read a parameter directly from that file and could be changed more dynamically. (i.e. company_name=appistry). An additional benefit is that you can override some of the general parameters in the fabric.cfg with a parameter for this specific application. (i.e. fabric-jvm-options=-Xrs -Djava.class.path=${FABRIC_HOME}/classes/fabric.jar). Again, this is not required, but may be beneficial when you want to change parameters specifically for this application and not for your other apps.
3.) Yes, another machine can access the fabric application even if you have not installed the appistry software on it. However, you would need to copy the API to that machine and of course the client binary. Here are the steps:
a.) Build the client on a machine just as you are doing now.
b.) Copy the client application that you built and the fabric.jar file to another machine. You should then be able to submit transactions from that machine to your fabric. This is also assuming that your network allows you to do so.

If you are wanting to install 3rd party services via CloudIQ, you could utilize our manager part of the product. This is somewhat different than a fabric application, but it is similar in the packaging and deployment process. Here is an example that would help you package up Tomcat and deploy it to all of your fabric workers:
http://www.appistry.com/community/wiki/display/latest/Deploy+Tomcat

Subsequently, you could then create other FARs that install web applications to that tomcat service. There are many ... many possibilities.

I hope this helps, and feel free to reply with status and any other questions.

Thanks,
Mark

Pradnya,

I will go through each of the steps mentioned. I hope that I can clear some things up for you.

Another thing to watch out for... If you have already deployed a version of apache to your fabric, you may need to update the version number in the apachewin.xml file and repeat step 9 and 10. Here is the line that includes the version number that you would change:
version="1.05"

I hope this helps. If you receive a specific error on one of these steps, feel free to post it and I would be happy to help.

Thanks,
Mark

I am trying to deploy apache on cloud iq platform- on windows.
I was able to deploy it using-'grab and go'; method.

but when i tried to follow the detailed steps,
i downloaded the apache setup.
But i am not getting the following steps- 3,4,6,7,8.
CAN YOU PLEASE HELP ME TO MAKE IT CLEAR?

1) Download Appistry's example Apache service files. Create a directory to contain the files you will be packaging: for example 'apachewin

2) Download the latest version of Apache (for example, 'apache_2.2.11-win32-x86-no_ssl.msi').

3) Review and update the Fabric Archive (FAR) Definition XML file ('apachewin.xml') that will direct the installation and operation of the
Apache service:

4) Update the Apache configuration file, 'httpd.conf':
Set the listening port, as appropriate:

6) Review and update the 'install.bat' (as appropriate). The example will install the service quietly, update the Apache configuration file
and copy the 'index.html':

7) Review the 'query.bat' (see Query Elements for more information), which will allow the Fabric to report on the state of the Apache
service:

8) Review and update the 'uninstall.bat', which will be invoked if the service is deleted or if a newer version is deployed:

9) Use the fabric_pkg command to package the FAR:

10) Deploy the newly created 'apachewin.far' to the fabric using the fabric_ctl deploy command:

THANK YOU.