search.yaml

# Copyright (c) 2017, the Dart project authors.  Please see the AUTHORS file
# for details. All rights reserved. Use of this source code is governed by a
# BSD-style license that can be found in the LICENSE file.

runtime: custom
env: flex
service: search

resources:
  cpu: 8
  # This is the max memory we can request for 8 cpus.
  # https://cloud.google.com/appengine/docs/flexible/reference/app-yaml?tab=python
  memory_gb: 51.6
  disk_size_gb: 25

#manual_scaling:
#  instances: 1

automatic_scaling:
  min_num_instances: 6
  max_num_instances: 10
  # The service does most of the work on a single-core CPU, and because of that, the utilization
  # metrics is usually under 0.05.
  # Full single-core utilization is about 0.125 (1/cpu_count):
  # - starting a new isolate maxes out that utilization during the index loading
  # - serving requests may max it out too
  # This metrics is observed to go up to 0.20.
  # Threshold is calibrated to be triggered if the new isolate is starting + the existing is above 50% load.
  cpu_utilization:
    target_utilization: 0.175
  # The number of seconds that the autoscaler should wait before it starts collecting information from a new instance.
  # Skipping the collection in the first 20 minutes.
  cool_down_period_sec: 1200

skip_files:
- ^\.git/.*$

liveness_check:
  path: '/liveness_check'

readiness_check:
  path: '/readiness_check'
  check_interval_sec: 15
  app_start_timeout_sec: 1800