Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

I guess your problem is not about creating Trove instances but building a Trove guest database image. Once we have Trove guest database image ready for use, we can launch a Trove instance with that.

My experiences tell me that it may be faster to build a customized database guest image manually instead of using dib. A customized database guest image can be a snapshot of a Nova instance which has trove-guestagent and database software package installed and configured on it. With that idea in mind, we can build a customized database guest image and use it with following steps roughly. Note that there are more detailed issues to take care of in the process.

  1. Launch a Nova instance.
  2. Prepare environments of the above Nova instance for trove-guestagent, specifically, appropriate ubuntu repository for OpenStack version you have deployed.
  3. Install and configure trove-guestagent on the above Nova instance.
  4. Install database software package on the above Nova instance.
  5. Shutdown the above Nova instance.
  6. Take a snapshot of the above Nova instance. Note this snapshot is an image on OpenStack. This snapshot is the Trove database guest image.
  7. Create a Trove datastore for your database.
  8. Add the above snapshot as a version of Trove datastore created in Step 7.
  9. Launch a Trove instance with the datastore version created in Step 8.

Do you think the above workflow makes sense? Feel free to let me know if you have any confusion.

Hope this is helpful.