Wikimedia Release Engineering Team/Checkin archive/20191030
Appearance
2019-10-30
[edit]Vacations/Important dates
[edit]- November 1 (All Saints' Day): Željko, Antoine holiday
- November 8–10: WikiCon North America (James)
- November 11 (Veterans' Day): US Staff
- November 12–15: TechConf19
- November 16–20: Offsite
- November 21–30: Jeena
- November 25–29: Dan
- November 28–29 (Thanksgiving): US Staff
- December 1–2: Jeena
- December 6: Lars (Finnish Independence Day)
- December 25–31 (Christmas): US Staff (+ everyone?)
- December 25–26: Lars, Željko, Christmas
- 1 January (New Year's Day): US Staff, Lars, Željko
- 20 January (Martin Luther King Jr. Day): US Staff
Rotating positions
[edit]Train
[edit]- Maniphest query for deployment blocker tasks: https://phabricator.wikimedia.org/maniphest/query/s3KW8bpsXhYF/#R
- 14 Oct - wmf.2 - Jeena + Lars
- 21 Oct - wmf.3 - Lars + Brennen
- 28 Oct - wmf.4 - Brennen + Mukunda
- 4 Nov - wmf.5 - Mukunda + Antoine
- 11 Nov - wmf.6 - No Train (Tech conf)
- 18 Nov - wmf.7 - No Train (Off site)
- 25 Nov - wmf.8 - Group0 on Tuesday Brennen + Mukunda (Thanksgiving)
- 2 Dec - wmf.8 part 2 - Wednesday Group1 + Thursday Group2
- 9 Dec - wmf.10 - TBD
- 16 Dec - wmf.11 - TBD
- 23 Dec - wmf.12 - No Train or any other deploys (Winterval)
- 30 Dec - wmf.13 - No Train or any other deploys (Still Winterval)
- 6 Jan - wmf.14 - TBD
- 13 Jan - wmf.15 - TBD
SoS
[edit]- Aug 14 onward: Zeljko 🎸 🎷 \o/
Team Business
[edit]Timespent spreadsheet
[edit]- For the avoidance of doubt: fill out the sheet week number for the previous week
- spreadsheet for week 43 (2019-10-21): https://docs.google.com/spreadsheets/d/1YS_2eUQWUOnygshMfxgGPu93GKGCO5t00yjUREQ5B-I/edit#gid=337909193
- reminder: your time spent doesn't need to total to 100%
Clinic Duty Proposal
[edit]- https://www.mediawiki.org/wiki/Wikimedia_Release_Engineering_Team/Clinic_Duty
- Not done thcipriani JFDI®
November Offsite/Summit
[edit]- TechConf19: November 12–15
- Offsite: November 16–20
- Arrive the 15th
- 16th is our fun day/outing
- 17–20 are working days
- 21st fly home
- https://etherpad.wikimedia.org/p/EngProdTeam-Offsite-2019-11-planning
- Outing day! Vote please
Book club
[edit]Monthly reflection on accomplishments - October '19 edition
[edit]- https://www.mediawiki.org/wiki/Wikimedia_Release_Engineering_Team/Monthly_notable_accomplishments
- Add as you have them!
October
[edit]- Dev images are now automatically created as part of postmerge via the pipeline for:
- Parsoid
- Soon: RestBASE
- (different from RESTbase? ;-))
- Selenium documentation updated https://www.mediawiki.org/wiki/Selenium/Node.js
- Quibble 0.0.36 released https://lists.wikimedia.org/pipermail/wikitech-l/2019-October/092658.html
- Quibble 0.0.37 released https://lists.wikimedia.org/pipermail/wikitech-l/2019-October/092660.html
- Quibble 0.0.38 & 0.0.39 released for mediawiki/tools/api-testing
- Introducing Phatality - Streamlined error reporting from Kibana to Phabricator https://phabricator.wikimedia.org/phame/post/view/177/introducing_phatality/
- HHVM removed from CI and MediaWiki.
- Gerrit is on gerrit1001 now
- … and so is most of the code review. ;-) :)
- Unforked Jenkins Job Builder
Incoming/Needs attention
[edit]- Wikidata load issues; NFA for us right now. T236928 // T234948 // …
- Branch cut didn't work with the new script, had to use the old one.
Scrum of Scrums for next week
[edit]Internal team notes, not for pasting into the SoS notes
[edit]Incoming from last week (aka this morning)
[edit]Outgoing next week
[edit]- Blocked:
- Site Reliability Engineering: logstash-beta.wmflabs.org does not receive any mediawiki events https://phabricator.wikimedia.org/T233134
- [Low priority] Fundraising Tech: Code review on https://gerrit.wikimedia.org/r/544071 appreciated to unblock https://phabricator.wikimedia.org/T192167
- Updates:
- v3 of architecture document for new continuous integration wants review and feedback https://www.mediawiki.org/wiki/User:LarsWirzenius/NewCI
- Train Health
- Last week: 1.35.0-wmf.3 - phab:T233851
- This week: 1.35.0-wmf.4 - phab:T233852
- Next week: 1.35.0-wmf.5 - phab:T233853 - last train before two week break
Callouts
[edit]- Release Engineering
- Next week train last train (1.35.0-wmf.5 - phab:T233853) is the last train before two week break
- v3 of architecture document for new continuous integration wants review and feedback https://www.mediawiki.org/wiki/User:LarsWirzenius/NewCI
Train status and happenings
[edit]- No Gerrit GC cycle during branch cut. How was branch cut?
Standup!
[edit]- Make sure your tasks are correct on our kanban board!
Antoine
[edit]Last week
[edit]- Best thing
- JJB to be used directly from integration/config: tox -e jenkins-jobs -- <arguments>
- Will probably phase out integration/jenkins-job-builder
- 🎉
- Will probably phase out integration/jenkins-job-builder
- Couple of JJB patch upstream to speed it up (from 28s down to 4s on my machine)
- Fixed Zuul gearman monitoring probe. Raised threshold
- sync-up Slides for tech conf
- JJB to be used directly from integration/config: tox -e jenkins-jobs -- <arguments>
- Worst thing
- Docker containers on Stretch are 20% slower than Jessie ones T236675
- liw: could this be a spectre problem?
- hashar: I think there is a problem with that security patch and docker
- Docker containers on Stretch are 20% slower than Jessie ones T236675
- What I planned vs did
- Unplannned
This week
[edit]- What I plan to do
- Blockers
Other
[edit]- Not there friday Nov 1st (holiday: All Saints)
Brennen
[edit]Last week
[edit]- Best thing
- Decreasing sense of terror about train.
- Worst thing
- Train.
- What I planned vs did
- Done Train backup for Lars / train lead this week
- Done Sort out all hands travel
- Dev environment hackery
- Not done Check out k3sup?
- Done Think about TechConf local dev sessions / discuss
- And also In progress
- Not done Improve sshecret pull req: https://github.com/thcipriani/sshecret/pull/2
- Decided after discussion that current state is Good Enough™, Tyler merged
- May devote some future yak shaving time
- Unplannned
- T236333: Install wikidiff2 into mediawiki-quibble-api-testing-vendor-mysql-php72-docker
- Discovered Antoine's jjb changes firsthand
- Help on some local-charts bugfixen:
- T236333: Install wikidiff2 into mediawiki-quibble-api-testing-vendor-mysql-php72-docker
This week
[edit]- What I plan to do
- Train, including some updates to docs
- Improve a log-tailing script and document
- Write a "This Week in Logspam"
- Work on tech conf session
- Review pending local-charts patches
- Take a crack at TimedMediaHandler on local-charts
- Train, including some updates to docs
- Blockers
Other
[edit]
Dan
[edit]Last week
[edit]- Best thing
- Learned a lot about statistical forecasting
- Worst thing
- Learned a lot about statistical forecasting
- What I planned vs did
- Finished (enough) the Seakeeper proposal for presentation to SRE Service Ops
- Presented proposal to SRE Service Ops
- Unplannned
- Fixed up Jenkins build reports
This week
[edit]- What I plan to do
- Following conversations about Zuul's Dependent Pipeline Manager and how it will need to fit with New CI, looking into it a bit
- Calculating overhead
- Looking at implementation
- Revisiting notes from Dec 2018 offsite and how we might test it's efficacy
- Start working with Lars on pipelinelib user documentation
- Following conversations about Zuul's Dependent Pipeline Manager and how it will need to fit with New CI, looking into it a bit
- Blockers
Other
[edit]
Greg
[edit]Last week
[edit]- Best thing
- I have a desk!
- Worst thing
- Thought I lost the hardware for the desk for a day, luckily my partner is smarter than I
- What I planned vs did
- TechConf: mostly on track, the final countdown begins ( https://www.youtube.com/watch?v=9jK-NcRmVcw )
- Unplannned
- Things
This week
[edit]- What I plan to do
- TechConf!
- Offsite
- Blockers
Other
[edit]James
[edit]Last week
[edit]- Best thing
- WikidataCon.
- Worst thing
- WikidataCon-related travel.
- What I planned vs did
- Done WikiDataCon from tomorrow onwards; back in the US next Wednesday.
- Not done YAML/static builds of dblists; not had the time yet today to deploy the next step.
- Unplannned
- Some extra PHP 7.0 / 7.1 CI removal patches for MW-related libraries.
- Additional linting: we now lint require() JS calls to work with ResourceLoader https://github.com/wikimedia/eslint-plugin-mediawiki
- Some more clean-up/support for PHPUnit upgrade, Phan upgrade, removal of globals, etc. T192167, T159283, T160811, …
- Built some early images for PHP74 CI testing; paused waiting for PHP 7.4.0 release. T236132
- Explored replacing Travis CI with GitHub Actions CI for some of our GitHub-hosted repos. https://github.com/wikimedia/less.php/pull/18
- Helped Analytics by putting a "Statistics" link in the footer of every page: T235803.
This week
[edit]- What I plan to do
- Static YAML generation of dblists
- Static YAML generation of InitialiseSettings
- LocalDev support
- TechConf planning
- Blockers
- Waiting on feedback from FR-Tech on whether we can proceed with PHPUnit stuff as it effects them (pinged in SoS) T192167
Other
[edit]- Working from the East Coast from Sunday onwards.
Jean-Rene
[edit]Last week
[edit]- Best thing
- Worst thing
- What I planned vs did
- Unplannned
This week
[edit]- What I plan to do
- Blockers
Other
[edit]
Jeena
[edit]Last week
[edit]- Best thing
- fixed some bugs in local-charts
- Worst thing
- planning is hard
- What I planned vs did
- In progress tech conf session planning
- Not done changes to mw core image publish pipeline to log more and fail if prereq script errors
- Not done re-open patchsets to build restbase dev image
- Not done timespent spreadsheet
- In progress work with zeljko on getting local-charts running
- Unplanned
This week
[edit]- What I plan to do
- tech conf session planning
- changes to mw core image publish pipeline to log more and fail if prereq script errors
- re-open patchsets to build restbase dev image
- timespent spreadsheet
- fix admin username/password issue for mw core on local-charts
- Blockers
- need Marko to review my deployment-charts restrouter patch (thcipriani: IIRC he's OOO today....)
Other
[edit]
Lars
[edit]Last week
[edit]- Best thing
- Ran the train and didn't break anything too badly
- Worst thing
- Realized that my technical writing isn't very useful
- What I planned vs did
- Done Train, 1.35.0-wmf.3
- In progress Skim Go book, write a small program from scratch: fetch a web page, diff against previously fetched version.
- Done Attempt to set up study groups, or possibly SIGs, for Go and Rust.
- Unplanned
- First meeting with SRE on hardware to new CI and general sync on this.
- First meeting in rebooted series of Deployment Pipeline meetings.
This week
[edit]- What I plan to do
- Ponder what came out of the SRE meeting on CI hardware meeting, and what I can learn from that, and what I can do for future discussssions on this.
- Help prepare CI/CD related sessions at Tech Conf.
- Respond to email thread started by Tyler. Especially: suggest goals and desired outcomes of the session and other related discussions from Tech Conf.
- Fill out https://phabricator.wikimedia.org/T234641 best I can.
- Sketch a distillation of my CI architecture document.
- Update and fill out my personal travel checklist for Tech Conf.
- Read up about last year's Tech Conf to be better prepared for this years'.
- Install at least a rudimentary, "hello world" level Argo some K8s somewhere.
- Skim Go book, write a small program from scratch: fetch a web page, diff against previously fetched version.
- Have first Go study club meeting, on coming Monday. https://office.wikimedia.org/wiki/Go
- Stretch: Arrange first meeting for Rust study club. https://office.wikimedia.org/wiki/Rust
- Stretch: Look at Zuul's dependent pipeline manager and think hard about it for new CI.
- Blockers
Other
[edit]- Train conductor hat discussion was fun, again. Or is it train engineer hats?
- http://monica-anderson.com/autumnsong.html
Mukunda
[edit]Last week
[edit]- Best thing
- Making good progress on phabricator redundancy and failover stuff with dzahn. phd (async job queue) is now running on two servers.
- Worst thing
- All of my planned tasks are Not done because of a lot of time sunk on unplanned stuff.
- What I planned vs did
- Not done Test out phabricator's built-in search engin
- Not done Drum up interest in patch-triage (email wikitech-l, not sure what else to do)
- Working on it...
- Get branch.py ready for full automation in releases-jenkins
- still blocked on code review
- Not done Deploy phatality (after https://gerrit.wikimedia.org/r/#/c/545593/ merges)
- Doing this today
- Not done Deploy phabricator
- Tried with scap, it didn't work. Turns out it was my fault (scap.cfg change fixed it)
- Unplanned
- Phabricator dump is broken, so far I'm stumped: https://phabricator.wikimedia.org/T236507
- Spent a bunch of time helping volker_e and converting desig/style-guide to scap: https://phabricator.wikimedia.org/T235677
This week
[edit]- What I plan to do
- Still working on branch.py, it didn't quite work as expected during the train branch cut this week, hope to have it working next time.
- Deploy phabricator (with scap)
- Deploy Phatality
- Get branch.py working
- Be the backup train conductor
- Blockers
Other
[edit]
Tyler
[edit]Last week
[edit]- Best thing
- New CI moving/public/momentum!
- Worst thing
- 113 Résumés + cover letters
- What I planned vs did
- Done Resumé review
- Not done Work out clinic duty rotation
- Done Next bookclub email
- Done Gerrit/JVM tuning to make use of Moar Memory™
- Unplanned
- Figured out Thanksgiving train (kinda/sorta)
- Design moving to git-LFS
This week
[edit]- What I plan to do
- Work out clinic duty rotation
- Techconf session planning
- Gerrit review notes splitbrains
- More gerrit tuning
- Blockers
Other
[edit]
Zeljko
[edit]Last week
[edit]- Best thing
- local-charts powered mediawiki works on my machine!
- Worst thing
- /me grumbles something containing words: savings, daylight, time
- What I planned vs did
- In progress ERROR webdriver: Request failed due to Error: session not created: Chrome version must be between 70 and 73 https://phabricator.wikimedia.org/T234610
- In progress All repositories with Selenium tests should use wdio-mediawiki https://phabricator.wikimedia.org/T199113
- Unplanned
- mediawiki-vagrant
- MediaWiki-Vagrant should use the same Node.js version as Wikimedia production https://phabricator.wikimedia.org/T217113
- `npm run selenium` fails on MW Vagrant (SyntaxError: Unexpected identifier; due to Node 6) https://phabricator.wikimedia.org/T236282
- local-charts
- Can not log in to mediawiki https://phabricator.wikimedia.org/T236776
- Error: error installing: the server could not find the requested resource https://phabricator.wikimedia.org/T233960
- `Error: validation failed` during `make deploy` https://phabricator.wikimedia.org/T236471
- `make install` fails with `Error: caskroom/cask was moved. Tap homebrew/cask-cask instead.` https://phabricator.wikimedia.org/T236393
- `make install` fails with `“minikube-darwin-amd64” cannot be opened because the developer cannot be verified.` https://phabricator.wikimedia.org/T236402
- mediawiki-vagrant
This week
[edit]- What I plan to do
- Wikimedia Technical Conference 2019 Session: System level testing: patterns and anti-patterns with Selenium https://phabricator.wikimedia.org/T234635
- Release new version of wdio-mediawiki https://phabricator.wikimedia.org/T236927
- All repositories with Selenium tests should use wdio-mediawiki https://phabricator.wikimedia.org/T199113
- T236620 Refactor GrowthExperiments Homepage Selenium test https://phabricator.wikimedia.org/T236620
- Blockers
Other
[edit]- [QA] Proposal to close this list https://lists.wikimedia.org/pipermail/qa/2019-October/002744.html
- Jeena, Brennen (local-charts)
- Can not log in to mediawiki https://phabricator.wikimedia.org/T236776
- Capitalize MediaWiki user (from admin to Admin) https://gerrit.wikimedia.org/r/c/releng/local-charts/+/546988
- TLDR section in README https://gerrit.wikimedia.org/r/c/releng/local-charts/+/546990
- Antoine
- ERROR webdriver: Request failed due to Error: session not created: Chrome version must be between 70 and 73 https://phabricator.wikimedia.org/T234610