Page MenuHomePhabricator

Evaluate requirements to productionise Sockpuppet Detection API
Closed, ResolvedPublic

Description

Context

In the long term, Chris Albon's team will develop and deploy an ML production pipeline. In the short term, this model should be made available for use by Product teams.

Requirements

  • Document as tasks steps required to dockerise the Sockpuppet Detection model and host on kubernetes
  • Document as tasks steps required to make Sockpuppet Detection model available via the API gateway (will not be done)
  • Evaluate effort to complete work

Event Timeline

WDoranWMF created this task.
WDoranWMF moved this task from Backlog to Ready on the Platform Team Workboards (Green) board.
WDoranWMF removed a project: Anti-Harassment.

Document as tasks steps required to dockerise the Sockpuppet Detection model and host on cloudVPS
Document as tasks steps required to make Sockpuppet Detection model available via the API gateway

We shouldn't/can't/I REALLY HOPE won't expose services running on cloudVPS via production infrastructure. The two environments have vastly different stability requirements and guarantees. If the features dependent on this new API require production level of support and stability, the new service has to run on production k8s cluster and thus meet production standards. If the cloudVPS level is OK, the service and it's API should be hosted there. Let's not mix the two.

WDoranWMF updated the task description. (Show Details)
hnowlan updated the task description. (Show Details)

Moving the work on this ticket to T265722.