Wikimedia Release Engineering Team/SSD Sync Up/2019-06-11

2019-06-11

Last Time: 2019-06-04

Deployment Pipeline

Workboard

• Pipeline image build cleanup
  •  Done thcipriani to review and merge
  • TODO follow up for contint1001

• Add/reserve a Jenkins node for the pipeline's trigger jobs
  • Seems like a simple one that's a good CI admin task for the uninitiated, takers?
  • Brennen to take a shot
  • Rough steps
    • setup m1.small (or whatever the small isntance is) in https://horizon.wikimedia.org in the integration project
    • Add a new node in our jenkins setup https://integration.wikimedia.org/ci/ (cargo cult other nodes)
    • Add the "trigger" label to the node
    • Modify the trigger jobs in integration/config to use "trigger" (https://gerrit.wikimedia.org/r/plugins/gitiles/integration/config/+/master/jjb/service-pipeline.yaml#5 ... and other "trigger" jobs)

• Blubber Policy file && Pipeline config validation(/docs)
  • thcipriani licks cookies

• Doc writers meeting
  • TODO james setting up regular meeting
  • TODO brennen writing an email to propose date for regular meeting

• Kartotherian (map rendering / OSM) and Tilerator
  • kartotherian is a meta repository of tightly coupled bits of code
  • is now a mono repo
  • Still using upstream Kartotherian
  • Moving hosting to docker/k8s
  • TODO Bring this up with SRE/serviceops

New CI

• https://wmf-gitlab3.vm.liw.fi/ - use it or else!
  • though it's likely to change randomly as Lars learns more and prototypes things
  • please try to break this :)
  • Tsk, no support for British English. ;-)
    • Upstream issue. :)

Local Development

• Fix Install Script For Some Versions of Ubuntu
  • prereqs: Remove docker installation steps should merge, anyone with a Mac want to give it a quick test?

• Define a base docker-pkg template and .pipeline/blubber.yaml for mediawiki/core
  • Unblocked now that dev images are published?
    • B: I was going to say it'd be fine to merge https://gerrit.wikimedia.org/r/c/mediawiki/core/+/508392 now, but on reflection I'm not sure.

• Move local-charts helm charts to a chart repository
  • Anything needed? Should task be broken down further?

• Create an interface for the local-charts ecosystem
  • Mukunda and I (thcipriani) played with skaffold on k3s last Wednesday
  • skaffold seems like overkill unless we have a build step -- also pushes to repository by default -- seems to assume you're using GKE
  • jeena: manages multiple release at one time -- enter into skaffold.yaml configuration
  • MediaWiki, parsoid, restbase, other -- 3 different sets of those with different configuration
  • TODO mukunda to update task with details

• k3s
  • installs directly on bare metal
  • minimal version of k8s, less complexity
  • no harder to setup than minikube
  • kubeDNS? unsure
  • hides kubectl behind k3 kubectl
  • solved with

    #!/bin/bash
    exec k3 kubectl