Notes/Troubleshooting tips:

• UI has changed since screen shots if you are using development or testing versions. Image Factory no longer available in UI.

  • When switching from stable to development versions you should remove previous repository and start fresh.

• Clean start

  • Reset Aeolus

    sudo yum remove aeolus-configure;
    sudo yum remove aeolus-all;
    sudo yum install aeolus-all;

  • aeolus-cleanup : use cleanup to reset database contents and service details.

  • aeolus-configure: reinitialize components and restart services.

    • Without cleanup and configure steps above, the UI display may display mangled contents between updates.

    • If you get errors in the configure/cleanup phases, there may be hung processes preventing correct cleanup. Use 'ps aux | grep aeolus' to locate duplicate/problematic processes for elimination.

'Step 4. Create a Fedora 14 Template' is no longer valid.

• You will need to drop back down to the command line terminal to complete these steps:

  • You will be building the cloud image with Aeolus which will later be pushed to Cloud provider.

    • 1) Use 'aeolus-image list --images' to show existing images.

    • 2) Use 'aeolus-image build --target ec2 --template (full path to template .tdl)' to build new images (if necessary).

      • See example template here: https://github.com/clalancette/oz/blob/master/examples/f14x86_64.tdl

    • 3) Create a deployable xml definition and host on an accessible web server.

    • See http://aeolusproject.org/page/Launching_a_Deployment_with_CLI_Tools for sample deployable.xml document.

  • 1) The 'hwp' attribute in the <assembly...> definition must match one of 'Administer>Hardware Profiles' available. A profile like 'hwp1' should already be available.

  • 2) The 'id' attribute of the <image..> element must match an existing image.

  • 3) Copy the deployable.xml document into /var/www/html/deployables directory and verify that http://localhost/deployables/deployable.xml is accessible and displays expected contents. (assumes http server started and accessible.)

  • 4) With conductor ui, use Administration>Suggested Deployables to add the complete http url to the deployable.xml.

  • 5) Push the image(sample ids: insert valid image ids):

    • NOTE: If you experience issues pushing an image with your amazon credentials, use the following script from your conductor box to verify that your credentials are correct.

      • Verifying credentials work outside of Aeolus

        #!/usr/bin/python
        import boto.ec2
        # Connect to EC2 region we want
        ec2region = boto.ec2.get_region("us-east-1", 
        aws_access_key_id='<insert access key only between quotes>', 
        aws_secret_access_key='<insert secret key only between quotes>')
        print ec2region.name
        ------- Should print the following without errors on successful run -----------------
        us-east-1 

    • NOTE: If you still experience credentials difficulties, after verifying with previous step and double checking contents of your provider.xml content, then it's possible that your current user does not have correct permissions to view the ssh certificate used during automated authentication with the cloud servers. Use 'ls -al /etc/oz/id_rsa-icicle-gen' to confirm that user has correct visibility. The permissions should be 400 for current user. Ssh will also not accept visibility that is too open.

      2011-07-08 11:32:15,125 DEBUG imagefactory.builders.BaseBuilder.FedoraBuilder pid(2108) Message: Waiting for EC2 ssh access: 270/300
      2011-07-08 11:32:25,470 DEBUG imagefactory.builders.BaseBuilder.FedoraBuilder pid(2108) Message: Waiting for EC2 ssh access: 280/300
      2011-07-08 11:32:35,860 DEBUG imagefactory.builders.BaseBuilder.FedoraBuilder pid(2108) Message: Waiting for EC2 ssh access: 290/300
      2011-07-08 11:32:46,235 DEBUG imagefactory.builders.BaseBuilder.FedoraBuilder pid(2108) Message: Stopping EC2 instance and deleting temp security group
      2011-07-08 11:32:48,014 DEBUG imagefactory.builders.BaseBuilder.FedoraBuilder pid(2108) Message: Exception caught in ImageFactory
      2011-07-08 11:32:48,017 DEBUG imagefactory.builders.BaseBuilder.FedoraBuilder pid(2108) Message: Traceback (most recent call last):
        File "/usr/lib/python2.7/site-packages/imagefactory/builders/FedoraBuilder.py", line 485, in push_image
          self.push_image_snapshot(target_image_id, provider, credentials)
        File "/usr/lib/python2.7/site-packages/imagefactory/builders/FedoraBuilder.py", line 497, in push_image_snapshot
          self.push_image_snapshot_ec2(target_image_id, provider, credentials)
        File "/usr/lib/python2.7/site-packages/imagefactory/builders/FedoraBuilder.py", line 729, in push_image_snapshot_ec2
          raise ImageFactoryException("Unable to gain ssh access after 300 seconds - aborting")
      ImageFactoryException: Unable to gain ssh access after 300 seconds - aborting

• aeolus-image push --provider ec2-us-east-1 --id 1ecc08d8-0fff-466b-88cb-17fd81a318ad

• Use 'tail /var/log/imagefactory.log' to monitor success of this process.

  • I prefer to use the following command to get immediate logging without having to tail the log as in previous step: 'imgfac.py' --debug --image 1ecc08d8-0fff-466b-88cb-17fd81a318ad --provider ec2-us-east-1 --credentials (insert full path to xml provider details)/cloud/provider_accounts.xml

5)Start the image. You will either use

• Conductor (currently broken) OR

• Use the Cloud provider i.e. Amazon AWS UI at this point to start up your instance.

Credential headaches again:

• NOTE: to verify that the image is correct you need to setup your SSH client to use the certificates. See instructions here: http://docs.amazonwebservices.com/AWSEC2/latest/GettingStartedGuide/.

  • Pay special attention to these instructions if you plan on using putty to ssh into started cloud instances.

  • As you are attempting to connect in via ssh, if you are challenged for a password at login then you do NOT have the certificates setup correctly. With correct client certification credentials the only thing you will be asked for is the user account that you would like to be be logged in as (ex. root).

Useful links:

• Debug dev setup suggestions | deployable

  • http://aeolusproject.org/page/Launching_a_Deployment_with_CLI_Tools

• From machine where conductor is installed:

  • Shows images created and stored in Image Warehouse Factory: http://localhost:9090/images/

    • Example of specific image details(insert valid id): http://localhost:9090/images/1ecc08d8-0fff-466b-88cb-17fd81a318ad

  • Shows images successfully pushed out to cloud providers: http://localhost:9090/provider_images/

  • Conductor UI: https://localhost/conductor

  • Deployable details: [http://localhost/deployables/deployable.xml]

• Some sample data

  • Succesful aeolus-configure run

  • Succesful aeolus-all install pkg list