Starlight: Sprint 5 (April 8-19, 2019)

Starlight Sprint 5 Review Recording: April 19, 2019

What was the goal of Sprint 5?
Testing in staging environment at both campuses, bug fixes, theming (including the ability to apply a global theme), and getting Shibboleth configured for UCSB.

What is the milestone that this sprint is supporting?
Milestone 1 is to have an initial instance of the Spotlight platform in production for both campuses.

Accomplishments of Sprint 5:

  • Initial version of UCSD theme
  • UCSB theme revisions
  • Initial approach for global theming (blocked by upstream Spotlight PR)


  • UCSD successfully deployed to staging environment using Shibboleth
  • Added an initial healthcheck endpoint for the application. This will be needed when we eventually deploy Starlight in a Kubernetes environment
  • Collaborated with the Lark team to get a Helm chart for Solr deployed via the Lark Helm chart. This is work that Starlight will benefit from when it also transitions towards a Kubernetes/Helm based deployment.


  • Product Owner testing on both campuses identified a number of bugs
  • Team successfully resolved a number of bugs identified, including:
    - Resolved issue where the “feedback” menu item and drop-down contact form was no showing in header. This is connected to the feature that allows exhibit curators to specify a contact email address per exhibit  
    - The disabling of irrelevant and/or non-functional features (Citation, SMS This, Saved Searches). The non-functional features were determined to be caused by upstream bugs
    - Issue uploading individual items at UCSD
  • Discussed how best to approach testing on both campuses


  • Converted roles / playbooks to support Tower-compatible Ansible version
  • Reorganized passenger gems for idempotency
  • Ensure all required characters are properly percent encoded for database passwords
  • Enabled Shibboleth paths in apache's configuration


  • Updated Ruby to latest version (2.6.2)
  • Replaced deprecated “chromedriver-helper” dependency with new “webdrivers” alternative

Did we do everything we set out to accomplish?
Almost. There are a notable number of blocked tickets on the board. Some of these are related to bugs that were discovered during user testing, but several are related to external circumstances. These external circumstances have been major blockers for the team and include failure to merge upstream pull requests (meaning, that we have written the code and requested that it be added to the upstream Spotlight code base) and delays on the UCSB Shibboleth implementation.

By the end of next sprint, our goal is to give all stakeholders access to their respective installations.

GitLab link: