Skip to content

Latest commit

 

History

History
27 lines (25 loc) · 1.33 KB

TODO.md

File metadata and controls

27 lines (25 loc) · 1.33 KB

TODO

plan for de-centralized crawling

  • add column to match table "processed_at"
  • normal match pulling/updated at is unaffected, new matches have processed_at=null
  • match pulling is distributed across many machines
  • single "cruncher" machine periodically checks for new matches, updates players/global, then sets matches as processed
  • no conditions on processing matches, do away with old system of latest_match_id
  • to save space, old processed matches can have their response column set to null

misc todo

  • explore idea of pulling most users on-demand (when their page is loaded)
  • make error logger a "with" wrapper?
    • sends email on close if exiting block abnormally or state altered
  • find long term solution for match table
    • when pulling matches and saving to db, only save if time > player ticks
    • migrate to archive based on match ticks
      • add new column to match, backfill gradually
  • improve player lookup page/dropdown
  • explicit setup-everything script (python, uses envar)
  • cron to simulate status page, email me if anything down
  • integration tests against temp db
  • remove all un-subscribed players, remove subscribed field?
    • meh, non-urgent
  • ween/combine player.updated_at and player.match_updated_at
  • dynamically disable adding players if cookie gooes down? store status in db?
  • async the cfn requests