Wikimedia Release Engineering Team/Deployment pipeline/2017-09-05
Appearance
2017-09-05
[edit]Last Time
[edit]Tasks
[edit]- paths rfc
- Proof helm as a potential k8s deployment tool
- Helm for local dev on minikube? wmctl for...?
- Using other services as prebuilt images
- rebuild images as you development
- different use
- helm to create a standardized deployment for staging and production
- changes configuration variables for hostname, secrets, etc
- Keeps things similar like how deployment-prep *should* work :)
- Pod structure for logging and monitoring https://phabricator.wikimedia.org/T170120
- will eventually need feedback (e.g., services re: logging)
Quarterly goals
[edit]- RelEng (thcipriani/dan) talked about stuff https://etherpad.wikimedia.org/p/blubberlist
- upgrading 17.05 in labs
- may have to figure out a workaround without new packaging in wikimedia repo
- target stretch not jessie
- upgrading 17.05 in labs
- How do people deploy to production?
- Workflow for the deployer, implementing using the tools we've chosen
- would be good for design document
- to be written
- leaving placeholders for folks
- K8s version in prod/staging at 1.7.4