Wikimedia Release Engineering Team/Deployment pipeline/2018-08-02
Appearance
Last Time
[edit]General
[edit]- Soon™: sprint for graphoid deployment
RelEng
[edit]- Dan working on shared code library: https://gerrit.wikimedia.org/g/integration/pipelinelib/
- Establish shared library for pipeline code used in Jenkins
- ^ blocked on Jenkins fails to checkout shared Groovy library integration/pipelinelib
- Currently blocked (I think it's mostly me [thcipriani] that needs to review code here?)
- Move verify stage from Minikube to CI k8s namespace in production context
- I can definitely deploy to CI namespace
- Helm test failing Alex investigated
- What's happening with DNS for production? we are not doing DNS in the production cluster :D
- Workarounds? Use dns for the CI namespace?
- DNS in CI namespace
- Use the IP from helm (if available)
- Refactor service-checker image to be able to locate the mathoid container
- TODO: task CI user access to logs for namespace
- Sending to logstash?
- Graphoid
- Left some feedback on https://gerrit.wikimedia.org/r/#/c/mediawiki/services/graphoid/+/434471/
- Was able to get it to fail in the same way it's currently failing in CI with latest Blubber https://www.gymaholic.co/motivation/its-okay-to-fail-consistent-doesnt-mean
SREs
[edit]- Updated mathoid to chart 0.0.9 (and soon to 0.0.10)
- Upgraded all cluster to kubernetes 1.9.9
- Packaged and testing kubernetes 1.10.6
- Not sure when the upgrade is going to happen though
- Alex going on vacation \o/
- We got your phone number \o/