We want the load balancer to never serve sites that are broken. That can happen automatically if every application has a health.json endpoint that the load balancer can check.
Level of urgency
[ ] High
[x] Moderate
[ ] Low
Why is this maintenance needed?
Right now it's easy to deploy to one of x servers, have that deploy fail, and thus have the site go down, unless we do a laborious load balancer dance.
Acceptance criteria
[ ] There's a /health.json endpoint.
[ ] It checks at least the following services (which will be checked by nginx)
What maintenance needs to be done?
We want the load balancer to never serve sites that are broken. That can happen automatically if every application has a health.json endpoint that the load balancer can check.
Level of urgency
Why is this maintenance needed?
Right now it's easy to deploy to one of x servers, have that deploy fail, and thus have the site go down, unless we do a laborious load balancer dance.
Acceptance criteria
/health.json
endpoint.Implementation notes, if any
An example implementation: https://github.com/pulibrary/lae-blacklight/pull/380/files
First Step
Install the health-monitor-rails gem to the gem file, as per the above PR.
Health monitor rails gem: https://github.com/lbeder/health-monitor-rails