Wikimedia Release Engineering Team/Deployment pipeline/2018-10-11

From mediawiki.org

Last Time[edit]

General[edit]

  • Lars Wirzenius will be working on the project
    • introductions (mic not work, sorry)
    • Yes Done TODO: switch hangout over to Google Meet
  • Last week's TODO: create task to talk about how to track and install additional npm packages for all services
    • Yes Done Track and install additional npm packages for all service container images
    • _joe_: we should create an intermediate image as a docker layer
    • alex: we need these since these use binary packages, but I don't know where these layers should be created
    • dan: this could add a fair amount of bloat if not using multi-stage
    • alex: other servies use npm binary packages (maps) -- whatever we come up with should serve the needs of both those types of services and service-runner services
  • _joe_: when should new services go to k8s directly, we need docs, etc
  • alex: we have no docs now really, other than Marko
  • alex: helm charts are tricky, particularly CPU limits are had to come up with, tests needed and hard to do
  • TODO: start docs on Wikitech

RelEng[edit]

SREs[edit]

Add a very very first draft for zotero helm chart https://gerrit.wikimedia.org/r/#/c/operations/deployment-charts/+/466287/

Services[edit]

As Always[edit]