RIAK Workshop 2018

RIAK development workshop to discuss last year's activities and the coming year's roadmap.

The Zenith Bldg, 26 Spring Gardens, Manchester M2 1AB
Report to Floor 2 and buzz the intercom.

As per last year I think the general shape of the day will be:

• Welcome
• GDPR
• Introductions
• Agenda setting (from the items below and anything else on the day)
• Discussions / agreements
• Lunch
• Discussions / agreements
• End of day
• Repeat

As with last year it will be easy going and fluid.

If people can look to arrive 09:30 – 10:00, and we’ll get going once we’re all gathered.

https://www.meetup.com/RIAK-Development-Roadmap-Workshop/events/254984975/

Current Items to Discuss:

• Discussion around Basho name - website, github, what work has been done so far, what needs to be completed.
• OTP20 work.
• Riak 3.0.
• Replication - future (has it got one?) / work completed at bet365 (connection balancing, per cluster filtering, deletes).
• 2.2.7a-d replication issues– lessons learned
• What are the forums expectations of bet365 as custodians – are we doing enough ?
• Time series - a time to kill ?
• Periodicals and articles – riak.com / bet365 tech blog articles / reddit / hacker news / other ? (as in what medium should we use)
• Clients - deprecating unmaintained clients? Dev owners?
• Go client.
• Unmaintained dependency cull - ensemble, js, control etc.
• Relinquish Basho forks in favour of canonical repos. Many of the deps have been taken from upstream repo for the OTP work, may as well leave in situ? Make riak lean.
• Big Sets.
• Node latency monitoring - couple of different approaches, bet365 have completed some work on this, the NHS has also, with a different approach.
• How to work more cohesively toward common goals.
• A short reminder of the bugs and fixes that made it into 2.2.5, 2.2.6, 2.2.7a-d;
• Report on the testing and production-readiness of leveled - Quviq's work on property-based testing leveled, riak_test and latest volume testing results
• FSM and kv_vnode changes to utilise HEAD requests
• Overview of TictacAAE and its production readiness as an alternative anti-entropy mechanism
• Soft-limits (aggressive vnode queue monitoring)
• Mapfold (range-based object folds)
• End-to-end repl replacement (and questions about whether some of the repl control logic should be removed from such replacements)