Checklist for deployment of initial "production" dispatch infrastructure
- Obtain dedicated WMCS horizon project space/quota (we'll use onfire project)
- Containers (or puppetized service with VMs)
- Postgres backend
- DNS entries
- Service/host monitoring T326842
- Data protection (e.g. data exported and backed up) T326843
- HA/maintenance considerations
20220823
Status update: with https://gerrit.wikimedia.org/r/c/operations/puppet/+/824449 and https://gerrit.wikimedia.org/r/c/operations/puppet/+/824448 we can stand up a database primary and dispatch's frontend (behind SSO) and scheduler. Items pending:
- Write a dispatch plugin to read username from a request header.
- Worth checking with upstream if they'd be interested in having it too. https://github.com/Netflix/dispatch/issues/2447
- Ship said plugin with dispatch container
- Write a sync script between LDAP and dispatch API to set user permissions based on group (e.g. ops is dispatch admins)
- Make sure an admin user is created for API access for the script above