radis/docker-compose.override.yml.example at main · openradx/radis · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
# Observability overlay for RADIS.
#
# Tags every signal sent to OpenObserve with:
#   service.name           = the stack name (e.g. radis_prod, radis_staging)
#                            from OTEL_SERVICE_NAME=${STACK_NAME}
#   service.component      = the compose service that emitted it
#                            from SERVICE_COMPONENT, read by adit_radis_shared/telemetry.py
#   service.namespace      = openradx
#   deployment.environment = production by default; change per stack (see below)
#
# PER-STACK SETUP:
#   1. cp docker-compose.override.yml.example docker-compose.override.yml
#   2. If this is a staging or development deployment, edit the
#      `deployment.environment=production` value inside `OTEL_RESOURCE_ATTRIBUTES`
#      below to `staging` or `development`:
#        sed -i 's/deployment.environment=production/deployment.environment=staging/' \
#          docker-compose.override.yml
#
# STACK_NAME is dynamically resolved by adit-radis-shared's CommandHelper and
# injected into the compose subprocess env by cli compose-up / compose-build /
# stack-deploy. No new .env entry required.

x-observability-net: &observability-net
  networks:
    - default
    - openradx-observability

x-observability-env: &observability-env
  OTEL_EXPORTER_OTLP_ENDPOINT: ${OTEL_EXPORTER_OTLP_ENDPOINT:-}
  OTEL_PYTHON_DJANGO_EXCLUDED_URLS: "health/,static/.*"
  OTEL_SERVICE_NAME: ${STACK_NAME}
  OTEL_RESOURCE_ATTRIBUTES: "service.namespace=openradx,deployment.environment=production"
  # Swarm interpolates {{.Task.Slot}} to the replica ordinal (1, 2, 3…); plain
  # `docker compose up` passes the literal through, which telemetry.py drops.
  TASK_SLOT: "{{.Task.Slot}}"
  # Cap OTLP HTTP exporter retry deadline at 2 seconds. The default is 10s
  # with exponential backoff, which previously exceeded Docker's 10s
  # stop-grace-period during redeploys: when otel-collector was shut down
  # before the apps, the OTel batch processors blocked on flush() trying
  # to reach an unreachable collector, causing apps to be SIGKILLed
  # (exit 137) instead of shutting down cleanly.
  OTEL_EXPORTER_OTLP_TIMEOUT: ${OTEL_EXPORTER_OTLP_TIMEOUT:-2}

networks:
  openradx-observability:
    external: true

services:
  init:
    <<: *observability-net
    environment:
      <<: *observability-env
      SERVICE_COMPONENT: init

  web:
    <<: *observability-net
    environment:
      <<: *observability-env
      SERVICE_COMPONENT: web

  default_worker:
    <<: *observability-net
    environment:
      <<: *observability-env
      SERVICE_COMPONENT: default_worker

  llm_worker:
    <<: *observability-net
    environment:
      <<: *observability-env
      SERVICE_COMPONENT: llm_worker