Wikimedia Search Team/Meeting minutes/2015-05-26

2015-05-26

Checkins

James
- Worked on API docs
- Wrote a simple WDQS graph explorer. Not deployed yet. Stas will help James with that.
- Will continue with API docs. Goal is to make it as quick and painless as possible for new devs
Stas
- Getting a few results from beta query service. Some bug reports. Some performance concerns.
- Will work on bugs and performance issues, and will streamline its deployment process
Nik
- Worked on query_string replacement. Wants to integrate with Cirrus and use on real log data.
- Should break into more granular tasks
Oliver
- Priority has been bug fixes.
- Dashboard is showing live data, which revealed several problems.
- Still some brittleness when tables are scrubbed. Should be fixed at some point.
- Clearing out legal stuff, some of which probably should have been handled by reading.
  - 40-point task is almost done, which will allow full-time work on S&D
- Reviewing tech tasks from job candidates (45 plausible in the pipeline)
  - Blocked on some steps due to budget, but moving good candidates along where possible
- Still slowed by analytics hardware limitations
  - There is twice as much non-production work as CPU capacity for it
  - Since new hardware is 3+ months away, Tomasz will get involved in trying to find a solution
  - KS to set up further meetings
Max
- Mostly preparing to demo maps during hackathon
  - Heard from Yuri that it went well, with lots of interest (including Lila)
- Looking to revise hardware request to bare minimum to try to fit budget
- Maps are “not dead yet”, so will keep working on them.
Tomasz
- Catching up on inbox; connecting w/Wes to understand budget.

Workboard Walkthrough

https://phabricator.wikimedia.org/tag/search-team/

Not necessary yet; will probably become part of the agenda later

Scrum of Scrums

Are we blocked?
- Nothing
Are we blocking?
- https://phabricator.wikimedia.org/tag/blocked-on-search/

Info/To Discuss

From last week
- Should James learn more about metrics/measurement?
  - "No"
  - There are three different aspects to this problem:
    - Figuring out what we should measure to evaluate our success. -> Erik
    - Implementing event logging to measure those things. -> Engineering
    - Creating dashboards and visualisations of that data. -> Oliver
- Who can help Stas with Java stuff?
  - We don’t have an answer for this - that is really something we need to work on
  - Wes should have answers within a week or so
Proposal to keep feature-specific phabricator projects around
- No objections, so will proceed
Oliver proposes that all logging should go to the analytics cluster
- Examples: Elastic, WDQS.
- Need to spread the knowledge of how to do it (Oliver will coach Stas)

FYI

Wikimania (July 15-19)
- (need to insert relevant team page here)
Some folks will attend the Open Air conf next Thursday (SF)