Page MenuHomePhabricator

Create a standard puppet module for service-runner services
Closed, ResolvedPublic

Description

The https://github.com/wikimedia/service-runner project establishes a standard interface for node services. This means that each service (or collection of services) has the same commandline arguments, uses the same config format & starts up the same way.

We should be able to exploit this regularity to use a standard puppet module to set up each of these services. A lot of the templates should be identical with those needed for the Debian package (see T89900), so lets try to share as much of the template code as possible.

Generic service tasks

Can be done by Debian package or puppet

  • Set up an init script / systemd unit
    • Can use the restbase init as a template
    • Parameters:
      • service name: used for paths (config in /etc/{service}/, init in /etc/init.d/{service}), user, group
  • user / group setup: use service name
  • normally, no local logging (apart from syslog) should be needed any more - we use logstash
  • could set up an /etc/defaults/{service} file

Unique to Puppet / trebuchet

  • set up generic groups & sudo rules per service to allow members of a svc-{service}-deployers group to:
    • service start, stop, restart
    • execute commands as the service user / group: attach the node debugger, trigger heap dumps
    • (ideally) perform common debug tasks like checking network connections per process with netstat -p

Unique to trebuchet-based deploys

  • set up the full trebuchet system (and make sure it actually works)

Per-service settings

  • config.yaml has at least some per-service settings that need to be modified for each instance of the service. We can create a thin puppet module for each service that contains a thin init.pp for the defaults and the config.yaml template. The template should ideally be shared with / provided by the service itself. It would also be desirable if modifications to this puppet template would not require global +2 in puppet, although I'm not sure how that would work.

Event Timeline

GWicke raised the priority of this task from to Needs Triage.
GWicke updated the task description. (Show Details)
GWicke subscribed.
GWicke set Security to None.
GWicke edited subscribers, added: ori, Joe, faidon and 2 others; removed: Aklapper.
GWicke added a subscriber: greg.
mobrovac claimed this task.

Indeed @GWicke :) Done.