Deployment at TAMU

Owned by Brandon Tharp
Feb 21, 2018
2 min read

Observations and Questions:

At the TAMU Librares, we were able to get folio running in approximately 20 docker containers.  The github projects of folioorg/folio-ansible and folioorg/folio-install along with other documentation in folio’s github page were used as a guide.  Here are some of the observations/questions we came up with.

  1. There are published container images in dockerhub.  However, these have no documentation.  There is no information on how to run the images as containers. Can we update the published images with documentation?
  2. It seems that a lot of the documentation and setup/install scripts for folio make assumptions that may not work in a multi-node scenario. What can we do moving forward to ensure that the documentation and various scripts are more portable for use in multi-node production scenarios?
  3. The only way to know what backend modules you need to install is by pulling/registering all available Modules into Okapi. This is a very resource intensive behavior, which I am hesitant to perform in production.  Is this the intended behavior moving forward? 
  4. I didn’t know that you had to enable the okapi module for the tenant for certain things to work.  Is this documented somewhere? If so where?
  5. How do you know what apps/modules to build stripes with?
  6. Installing, building and running stripes can be difficult. Sometimes the stripes builds fine, other times it has random build errors.  What is the plan for deployment of stripes? Will there be an official release of stripes or will we encourage people to customize it for their needs?
  7. The install&simulate=true curl command that spits out the required modules for your tenant does not work unless you register all Modules into Okapi beforehand.  Is this the intended workflow? 
  8. Deploying modules outside of okapi is much different and more difficult, specifically in regard to:
    1. Dependencies (ports, how to run containers, options you can set with each module)
    2. Database Initialization
    3. Database sample data
    4. Registration, deployment and enabling in okapi (hard to automate???)
    5. Which comes first database creation, module running or registration, deployment and enabling in okapi
  9. Is there something we can do to reduce this complexity for a better workflow in a distributed, multi-node production environment?