# Settings of the s6 provider for Puppet's service type.
#
# S6_PUPPET_START_TIMEOUT
#   How long a longrun may take after "svctl start" or "svctl restart" until
#   it is up -- in seconds. Default 30.
#
#   Why this setting exists: "svctl start" returns as soon as s6-rc has
#   spawned the run script. A service that needs a moment to open its socket
#   is not up yet then -- and asking straight away gave "svctl start reported
#   success but the service is not running" for a perfectly healthy service
#   (postfix on dns-1, 23.09.2026: check rc 0, spool complete, LISTEN :25).
#   Puppet counted an error and restarted it on the next run; the service
#   flapped once per run.
#
#   30 seconds by default, because puppetserver needs about 12 until 8140 is
#   open (measured on puppet-1.tick1). Raise the value here for a slower
#   service -- a service that is not up after the limit is reported with the
#   s6-svstat line, which tells "slow" from "dead".
#
#   The environment beats this file: S6_PUPPET_START_TIMEOUT=60 puppet apply ...
#S6_PUPPET_START_TIMEOUT=30
