434 lines
18 KiB
YAML
434 lines
18 KiB
YAML
name: messenger-lab7
|
||
|
||
x-app: &app
|
||
build: .
|
||
image: messenger-lab7-app:dev # уникальное имя: на общем Docker-host не перепутать с чужими стендами
|
||
networks: [backend]
|
||
read_only: true
|
||
cap_drop: [ALL]
|
||
security_opt: [no-new-privileges:true]
|
||
tmpfs: ["/tmp:size=64m,mode=1777"]
|
||
ulimits:
|
||
core: { soft: 0, hard: 0 }
|
||
volumes: [./secrets/tls/service-ca:/run/tls:ro]
|
||
logging: &logging
|
||
driver: json-file
|
||
options: { max-size: "10m", max-file: "3" } # stdout-логи ограничены; история логов — в Loki
|
||
|
||
# OpenTelemetry SDK api и worker (src/telemetry.ts). service.name задаётся у каждого сервиса отдельно.
|
||
# OTEL_SDK_DISABLED=true — приложение без telemetry (сравнение, отладка).
|
||
x-otel-env: &otel-env
|
||
OTEL_EXPORTER_OTLP_ENDPOINT: https://otel-collector:4318
|
||
OTEL_EXPORTER_OTLP_PROTOCOL: ${OTEL_EXPORTER_OTLP_PROTOCOL:-http/protobuf}
|
||
OTEL_RESOURCE_ATTRIBUTES: ${OTEL_RESOURCE_ATTRIBUTES:-deployment.environment.name=lab}
|
||
OTEL_SDK_DISABLED: ${OTEL_SDK_DISABLED:-false}
|
||
OTEL_PROPAGATORS: tracecontext # W3C traceparent/tracestate; baggage клиентов не принимается
|
||
OTEL_METRIC_EXPORT_INTERVAL: "10000" # мс; окна rate() от 1 минуты
|
||
OTEL_BSP_SCHEDULE_DELAY: "1000"
|
||
OTEL_BLRP_SCHEDULE_DELAY: "1000"
|
||
OTEL_NODE_EXPERIMENTAL_SDK_METRICS: "true" # otel.sdk.*: сколько spans/логов SDK обработал и отбросил
|
||
|
||
x-s3-env: &s3-env
|
||
S3_ENDPOINT: https://s3:9000
|
||
S3_REGION: ${S3_REGION:-us-east-1}
|
||
S3_BUCKET: messenger-attachments
|
||
NODE_EXTRA_CA_CERTS: /run/tls/ca.crt
|
||
|
||
x-jobs-env: &jobs-env
|
||
MAX_IMAGE_PIXELS: ${MAX_IMAGE_PIXELS:-40000000}
|
||
PIPELINE_VERSION: ${PIPELINE_VERSION:-1}
|
||
JOB_MAX_ATTEMPTS: ${JOB_MAX_ATTEMPTS:-5}
|
||
JOB_TIMEOUT_SECONDS: ${JOB_TIMEOUT_SECONDS:-60}
|
||
JOB_RETRY_BASE_MS: ${JOB_RETRY_BASE_MS:-2000}
|
||
JOB_REPUBLISH_SECONDS: ${JOB_REPUBLISH_SECONDS:-600}
|
||
FAILPOINTS: ${FAILPOINTS:-0} # 1 только на тестовом стенде: файл failpoint-<точка>.png один раз роняет процесс
|
||
|
||
services:
|
||
postgres:
|
||
image: postgres:18.6-alpine@sha256:10a5c69a83035ab5239f5620d708520e8b28f5aa1f2b20143e6024532edadfb5
|
||
entrypoint: ["/bin/sh", "/etc/lab/entrypoint.sh"]
|
||
environment:
|
||
POSTGRES_USER: messenger
|
||
POSTGRES_DB: messenger
|
||
POSTGRES_PASSWORD_FILE: /run/secrets/postgres_password
|
||
secrets: [postgres_password]
|
||
volumes:
|
||
- pgdata:/var/lib/postgresql
|
||
- ./infra/postgres:/etc/lab:ro
|
||
- ./secrets/tls/postgres:/run/identity:ro
|
||
healthcheck:
|
||
test: ["CMD-SHELL", "pg_isready -h 127.0.0.1 -U messenger -d messenger"]
|
||
interval: 2s
|
||
timeout: 2s
|
||
retries: 20
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "2", memory: 1g }
|
||
stop_grace_period: 30s
|
||
restart: unless-stopped
|
||
security_opt: [no-new-privileges:true]
|
||
networks: [backend]
|
||
|
||
openbao:
|
||
image: openbao/openbao:2.5.4@sha256:436eaf9778cad75507ff70ea26ace30dcbe15606e619ac3823495663d7f7c115
|
||
user: "100:1000"
|
||
entrypoint: ["/bin/bao"]
|
||
command: ["server", "-config=/etc/lab/server.hcl"]
|
||
volumes:
|
||
- ./infra/openbao/server.hcl:/etc/lab/server.hcl:ro
|
||
- ./secrets/tls/openbao:/run/identity:ro
|
||
- bao-data:/openbao/file
|
||
- bao-audit:/openbao/logs
|
||
networks: [frontend, backend]
|
||
ports: ["127.0.0.1:${BAO_PORT:-8200}:8200"]
|
||
read_only: true
|
||
cap_drop: [ALL]
|
||
security_opt: [no-new-privileges:true]
|
||
tmpfs: ["/tmp:size=16m,mode=1777"]
|
||
mem_limit: 256m
|
||
cpus: 0.5
|
||
restart: unless-stopped
|
||
|
||
# Filer, master и volume на loopback в том же контейнере. Наружу слушает только TLS S3.
|
||
s3:
|
||
image: chrislusf/seaweedfs:4.48@sha256:4e61d15fd35994cb1e43e1e553dff106794841fd9a99ade2fc8c8bfce4d7872d
|
||
command: ["server", "-dir=/data", "-ip=127.0.0.1", "-ip.bind=127.0.0.1", "-filer", "-s3",
|
||
"-master.volumeSizeLimitMB=128", "-volume.minFreeSpace=64MiB", "-s3.ip.bind=0.0.0.0", "-s3.port=9000",
|
||
"-s3.config=/run/s3/identities.json", "-s3.cert.file=/run/identity/tls.crt", "-s3.key.file=/run/identity/tls.key",
|
||
"-s3.iam=false", "-s3.port.iceberg=0", "-s3.port.lance=0", "-s3.autoCreateBucket=false",
|
||
"-s3.concurrentUploadLimitMB=256"]
|
||
volumes:
|
||
- s3data:/data
|
||
- ./secrets/runtime/s3:/run/s3:ro
|
||
- ./secrets/tls/s3:/run/identity:ro
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
healthcheck:
|
||
test: ["CMD-SHELL", "SSL_CERT_FILE=/run/tls/ca.crt wget -qO /dev/null https://s3:9000/healthz"]
|
||
interval: 2s
|
||
timeout: 2s
|
||
retries: 30
|
||
mem_limit: 768m
|
||
cpus: 1
|
||
restart: unless-stopped
|
||
security_opt: [no-new-privileges:true]
|
||
networks: [backend]
|
||
|
||
# Одноразовые сервисы до старта реплик: схема БД (advisory lock) и bucket (повторный запуск ничего не меняет).
|
||
migrate:
|
||
<<: *app
|
||
command: ["node", "src/main.ts", "migrate"]
|
||
environment:
|
||
DATABASE_URL_FILE: /run/secrets/database_url
|
||
secrets: [database_url]
|
||
restart: "no"
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "0.5", memory: 128m }
|
||
depends_on:
|
||
postgres: { condition: service_healthy }
|
||
|
||
s3-init:
|
||
<<: *app
|
||
command: ["node", "src/main.ts", "s3-init"]
|
||
environment:
|
||
<<: *s3-env
|
||
S3_ACCESS_KEY_ID_FILE: /run/secrets/s3_access_key_id
|
||
S3_SECRET_ACCESS_KEY_FILE: /run/secrets/s3_secret_access_key
|
||
secrets:
|
||
- { source: s3_admin_access_key_id, target: s3_access_key_id }
|
||
- { source: s3_admin_secret_access_key, target: s3_secret_access_key }
|
||
restart: "no"
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "0.5", memory: 128m }
|
||
depends_on:
|
||
s3: { condition: service_healthy }
|
||
|
||
# Брокер: RabbitMQ, quorum-очередь на диске (volume rabbitmq-data), publisher confirms, ручной ack.
|
||
# hostname фиксирует имя узла rabbit@rabbitmq: данные в volume привязаны к нему и переживают пересоздание.
|
||
rabbitmq:
|
||
image: rabbitmq:4.3.6-alpine@sha256:6931816af1a943f471c038b278835d36692d55dec769232739296bb91b7de0f5
|
||
hostname: rabbitmq
|
||
volumes:
|
||
- rabbitmq-data:/var/lib/rabbitmq
|
||
- ./infra/rabbitmq/rabbitmq.conf:/etc/rabbitmq/conf.d/90-messenger.conf:ro
|
||
- ./infra/rabbitmq/enabled_plugins:/etc/rabbitmq/enabled_plugins:ro
|
||
- ./secrets/tls/rabbitmq:/run/identity:ro
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/tls/telemetry-ca:/run/telemetry:ro
|
||
secrets: [rabbitmq_definitions]
|
||
healthcheck:
|
||
test: ["CMD", "rabbitmq-diagnostics", "-q", "check_port_connectivity"]
|
||
interval: 5s
|
||
timeout: 10s
|
||
retries: 30
|
||
start_period: 20s
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "1", memory: 512m }
|
||
stop_grace_period: 30s
|
||
restart: unless-stopped
|
||
networks: [backend]
|
||
|
||
api:
|
||
<<: *app
|
||
environment:
|
||
<<: [*s3-env, *jobs-env, *otel-env]
|
||
OTEL_SERVICE_NAME: messenger-api
|
||
BAO_ROLE: api
|
||
ARGON2_MEMORY_KIB: ${ARGON2_MEMORY_KIB:-19456}
|
||
ARGON2_ITERATIONS: ${ARGON2_ITERATIONS:-2}
|
||
ARGON2_PARALLELISM: ${ARGON2_PARALLELISM:-1}
|
||
SESSION_TTL_SECONDS: ${SESSION_TTL_SECONDS:-3600}
|
||
DB_POOL_MAX: ${DB_POOL_MAX:-8}
|
||
SHUTDOWN_DRAIN_MS: ${SHUTDOWN_DRAIN_MS:-2500}
|
||
MAX_UPLOAD_BYTES: ${MAX_UPLOAD_BYTES:-10485760}
|
||
volumes:
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/tls/api:/run/identity:ro
|
||
- ./secrets/pki:/run/pki:ro
|
||
depends_on:
|
||
migrate: { condition: service_completed_successfully }
|
||
s3-init: { condition: service_completed_successfully }
|
||
healthcheck:
|
||
test: ["CMD", "wget", "-qO", "/dev/null", "http://127.0.0.1:8081/"]
|
||
interval: 2s
|
||
timeout: 2s
|
||
retries: 3
|
||
# Lab6: старт с OpenTelemetry SDK (загрузка instrumentations через ESM-hook) на занятом хосте — до ~25 с.
|
||
# Успешная проверка сразу делает контейнер healthy; start_period только прощает неудачи при старте.
|
||
start_period: 120s
|
||
deploy:
|
||
replicas: ${API_REPLICAS:-2}
|
||
resources:
|
||
limits: { cpus: "${API_CPUS:-1}", memory: "${API_MEMORY:-256m}" }
|
||
stop_grace_period: 15s
|
||
restart: unless-stopped # упавшая реплика поднимается сама; после `docker stop` остаётся остановленной
|
||
|
||
# Worker: отдельный процесс без HTTP, конкурирующие consumers одной очереди. WORKER_CONCURRENCY = prefetch =
|
||
# одновременных заданий на реплику; память ограничена контейнером, libvips — один поток, без кэша.
|
||
worker:
|
||
<<: *app
|
||
command: ["node", "--import", "./src/telemetry.ts", "src/main.ts", "worker"]
|
||
environment:
|
||
<<: [*s3-env, *jobs-env, *otel-env]
|
||
OTEL_SERVICE_NAME: messenger-worker
|
||
BAO_ROLE: worker
|
||
WORKER_CONCURRENCY: ${WORKER_CONCURRENCY:-2}
|
||
volumes:
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/tls/worker:/run/identity:ro
|
||
depends_on:
|
||
migrate: { condition: service_completed_successfully }
|
||
s3-init: { condition: service_completed_successfully }
|
||
rabbitmq: { condition: service_healthy }
|
||
deploy:
|
||
replicas: ${WORKER_REPLICAS:-2}
|
||
resources:
|
||
limits: { cpus: "${WORKER_CPUS:-1}", memory: "${WORKER_MEMORY:-768m}" }
|
||
stop_grace_period: 15s # SIGTERM: consumer отменяется, текущие задания до 10 с, затем выход
|
||
restart: unless-stopped
|
||
|
||
# Периодическая сверка БД ↔ S3 с grace period (src/reconcile.ts). Свои S3-ключи: объекты + листинг bucket.
|
||
reconciler:
|
||
<<: *app
|
||
command: ["node", "src/main.ts", "reconcile", "--apply", "--every=${RECONCILE_INTERVAL_SECONDS:-300}"]
|
||
environment:
|
||
<<: *s3-env
|
||
BAO_ROLE: reconciler
|
||
RECONCILE_GRACE_SECONDS: ${RECONCILE_GRACE_SECONDS:-900}
|
||
STAGING_RETENTION_SECONDS: ${STAGING_RETENTION_SECONDS:-86400}
|
||
volumes:
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/tls/reconciler:/run/identity:ro
|
||
depends_on:
|
||
migrate: { condition: service_completed_successfully }
|
||
s3-init: { condition: service_completed_successfully }
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "0.5", memory: 128m }
|
||
restart: unless-stopped
|
||
|
||
lb:
|
||
image: haproxy:3.2.26-alpine@sha256:5893840576071e6217cd52ebdcc14f454c6f7a631056fa6f4e0f40d3cac4215b
|
||
volumes:
|
||
- ./infra/haproxy/haproxy.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
|
||
- ./secrets/tls/ingress:/run/public:ro
|
||
- ./secrets/tls/lb-client:/run/client:ro
|
||
- ./secrets/tls/lb:/run/identity:ro
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/tls/telemetry-ca:/run/telemetry:ro
|
||
ports: ["127.0.0.1:${LB_PORT:-8443}:8443"]
|
||
read_only: true
|
||
cap_drop: [ALL]
|
||
security_opt: [no-new-privileges:true]
|
||
depends_on:
|
||
api: { condition: service_healthy }
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "${LB_CPUS:-1}", memory: 128m }
|
||
restart: unless-stopped
|
||
networks: [frontend, backend]
|
||
|
||
# HAProxy проверяет CN и CRL перед передачей OTLP по loopback в общем network namespace.
|
||
otel-collector:
|
||
image: haproxy:3.2.26-alpine@sha256:5893840576071e6217cd52ebdcc14f454c6f7a631056fa6f4e0f40d3cac4215b
|
||
volumes:
|
||
- ./infra/otel/ingress.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
|
||
- ./secrets/tls/otel-collector:/run/identity:ro
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/pki:/run/pki:ro
|
||
networks: [backend, telemetry]
|
||
read_only: true
|
||
cap_drop: [ALL]
|
||
security_opt: [no-new-privileges:true]
|
||
mem_limit: 128m
|
||
cpus: 0.5
|
||
restart: unless-stopped
|
||
logging: *logging
|
||
|
||
collector:
|
||
image: otel/opentelemetry-collector-contrib:0.163.2@sha256:7f22d2a6748110ffb6a59bc0df448bccbf501cdb667d610ede4c61d3863d5849
|
||
network_mode: "service:otel-collector"
|
||
volumes:
|
||
- ./infra/otel/collector.yaml:/etc/otelcol-contrib/config.yaml:ro
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/tls/collector:/run/client:ro
|
||
environment:
|
||
GOMEMLIMIT: 180MiB
|
||
read_only: true
|
||
cap_drop: [ALL]
|
||
security_opt: [no-new-privileges:true]
|
||
mem_limit: 256m
|
||
cpus: 1
|
||
restart: unless-stopped
|
||
logging: *logging
|
||
|
||
prometheus:
|
||
image: prom/prometheus:v3.15.1@sha256:fa3302279710be2f21a8301ec17a8b4f2301fd829bd113839b3b6b02b6f191a2
|
||
command:
|
||
- --config.file=/etc/prometheus/prometheus.yml
|
||
- --storage.tsdb.path=/prometheus
|
||
- --storage.tsdb.retention.time=15d
|
||
- --storage.tsdb.retention.size=1GB
|
||
- --web.enable-otlp-receiver
|
||
- --web.config.file=/etc/prometheus/web.yml
|
||
- --enable-feature=created-timestamp-zero-ingestion # новая серия счётчика начинается с 0: rate() видит первые ошибки
|
||
- --query.lookback-delta=1m # данные остановленной реплики перестают учитываться через минуту, а не через 5
|
||
volumes:
|
||
- ./infra/prometheus:/etc/prometheus:ro
|
||
- ./secrets/tls/prometheus:/run/identity:ro
|
||
- ./secrets/tls/telemetry-ca:/run/telemetry:ro
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/tls/prometheus-scrape:/run/client:ro
|
||
- prometheus-data:/prometheus
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "1", memory: 512m }
|
||
stop_grace_period: 30s
|
||
restart: unless-stopped
|
||
logging: *logging
|
||
networks: [telemetry]
|
||
|
||
loki:
|
||
image: grafana/loki:3.7.9@sha256:0db216b630ae1f348a1584c67f3cead8b4c1fc839169a39501db664744158e28
|
||
command: ["-config.file=/etc/loki/loki.yaml"]
|
||
volumes:
|
||
- ./infra/loki/loki.yaml:/etc/loki/loki.yaml:ro
|
||
- ./secrets/tls/loki:/run/identity:ro
|
||
- ./secrets/tls/telemetry-ca:/run/telemetry:ro
|
||
- loki-data:/loki
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "1", memory: 384m }
|
||
stop_grace_period: 30s
|
||
restart: unless-stopped
|
||
logging: *logging
|
||
networks: [telemetry]
|
||
|
||
tempo:
|
||
image: grafana/tempo:2.10.10@sha256:12b0820f13ffa10ce956b6e9acc3b03f3cffded99109be6cc09d106fe56e5d38
|
||
command: ["-config.file=/etc/tempo/tempo.yaml"]
|
||
volumes:
|
||
- ./infra/tempo/tempo.yaml:/etc/tempo/tempo.yaml:ro
|
||
- ./secrets/tls/tempo:/run/identity:ro
|
||
- ./secrets/tls/telemetry-ca:/run/telemetry:ro
|
||
- ./secrets/tls/collector-ca:/run/collector-ca:ro
|
||
- tempo-data:/var/tempo
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "1", memory: 512m }
|
||
stop_grace_period: 30s
|
||
restart: unless-stopped
|
||
logging: *logging
|
||
networks: [telemetry]
|
||
|
||
# Grafana: интерфейс и запросы к хранилищам; datasources и dashboard импортируются из Git (provisioning).
|
||
grafana:
|
||
image: grafana/grafana:13.2.5@sha256:9f7aa1e1e3246aade5792d35a738049f7f32f89cd51d1956cf426a4e86268a84
|
||
entrypoint:
|
||
- /bin/sh
|
||
- -ec
|
||
- |
|
||
export DS_CA="$$(cat /run/tls/ca.crt)" DS_CERT="$$(cat /run/client/tls.crt)" DS_KEY="$$(cat /run/client/tls.key)"
|
||
exec /run.sh
|
||
environment:
|
||
GF_SERVER_PROTOCOL: https
|
||
GF_SERVER_CERT_FILE: /run/identity/tls.crt
|
||
GF_SERVER_CERT_KEY: /run/identity/tls.key
|
||
GF_SECURITY_ADMIN_USER: admin
|
||
GF_SECURITY_ADMIN_PASSWORD__FILE: /run/secrets/grafana_admin_password
|
||
GF_USERS_ALLOW_SIGN_UP: "false"
|
||
GF_AUTH_ANONYMOUS_ENABLED: "false"
|
||
GF_SERVER_ROOT_URL: https://localhost:${GRAFANA_PORT:-3000}
|
||
GF_ANALYTICS_REPORTING_ENABLED: "false"
|
||
GF_ANALYTICS_CHECK_FOR_UPDATES: "false"
|
||
GF_ANALYTICS_CHECK_FOR_PLUGIN_UPDATES: "false"
|
||
GF_PLUGINS_PREINSTALL_DISABLED: "true"
|
||
GF_NEWS_NEWS_FEED_ENABLED: "false"
|
||
secrets: [grafana_admin_password]
|
||
volumes:
|
||
- ./infra/grafana/provisioning:/etc/grafana/provisioning:ro
|
||
- ./infra/grafana/dashboards:/var/lib/grafana/dashboards:ro
|
||
- ./secrets/tls/grafana:/run/identity:ro
|
||
- ./secrets/tls/service-ca:/run/tls:ro
|
||
- ./secrets/tls/grafana-query:/run/client:ro
|
||
- grafana-data:/var/lib/grafana
|
||
ports:
|
||
- "127.0.0.1:${GRAFANA_PORT:-3000}:3000"
|
||
healthcheck:
|
||
test: ["CMD-SHELL", "SSL_CERT_FILE=/run/tls/ca.crt wget -qO /dev/null https://localhost:3000/api/health"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 30
|
||
deploy:
|
||
resources:
|
||
limits: { cpus: "0.5", memory: 256m }
|
||
restart: unless-stopped
|
||
logging: *logging
|
||
networks: [frontend, telemetry]
|
||
|
||
networks:
|
||
frontend: {}
|
||
backend: { internal: true } # api, worker, postgres, rabbitmq и s3-шлюз недоступны с хоста и не имеют выхода наружу
|
||
telemetry: { internal: true } # collector → prometheus/loki/tempo ← grafana; приложение сюда не подключено
|
||
|
||
volumes:
|
||
bao-data: {}
|
||
bao-audit: {}
|
||
pgdata: {}
|
||
s3data: {}
|
||
rabbitmq-data: {}
|
||
prometheus-data: {}
|
||
loki-data: {}
|
||
tempo-data: {}
|
||
grafana-data: {}
|
||
|
||
secrets:
|
||
database_url: { file: ./secrets/runtime/database_url }
|
||
postgres_password: { file: ./secrets/runtime/postgres_password }
|
||
s3_admin_access_key_id: { file: ./secrets/runtime/s3_admin_access_key_id }
|
||
s3_admin_secret_access_key: { file: ./secrets/runtime/s3_admin_secret_access_key }
|
||
rabbitmq_definitions: { file: ./secrets/runtime/rabbitmq_definitions.json }
|
||
grafana_admin_password: { file: ./secrets/runtime/grafana_admin_password }
|