Wikimedia Release Engineering Team/Deployment pipeline/2018-09-13
Last Time
editGeneral
editRelEng
edit- Played with https://github.com/zotero/translation-server-v2
- Minor tweak needed in https://github.com/zotero/translation-server/blob/master/package.json#L33
- Needs Node8+ obvs
- Otherwise npm test works fine and blubber/pipeline will work fine (for test at least -- as far as I got)
- wdqs-frontend docker image
- showed off blubber in IRC
- lots of unknowns, just reading this morning
- https://phabricator.wikimedia.org/T192006
- Optimizing resources in integration cluster
- Goal was to have fewer CI instances to make better use of the docker-cache
- Disk space
- Create a new mount at /var/lib/docker to store docker images seperately from core os and workspaces
- Stats collection
- Have used the jenkins api with a google spreadsheet
- statsd was missing metadata so was less useful
- Aggregating build results in logstash (currently blocked on firewall tasks)
- Advice needed for prometheus
- May be as simple as creating a prometheus scraper
- Probably need grafana on top to create graphs -- dan likes beautiful graphs
- depends on exported-resources, integration has a puppetmaster, but may not have a puppetdb
- Create a prometheus instance in integration
SREs
editBlubber upgraded to 0.5.0 Kubernetes clusters upgraded 1.10.6
Goal planning started. SREs will have a goal for this. Zotero v2, graphoid, blubberoid (if releng is still interested)
- TODO: RelEng team meeting discussion
Services
edit- zotero :)