forked from mtusi2026/mtusi2026lab7
Перенесена реализация lab6; бизнес-API без изменений, транспорт — HTTPS. - Внешний HTTPS (public-ca) на 127.0.0.1:8443; HSTS, TLS >= 1.2. - Обязательное mTLS LB -> каждая API-реплика: API проверяет цепочку internal-ca и SAN клиента (allowlist: lb); health только на 127.0.0.1. - TLS с проверкой сервера ко всем зависимостям: PostgreSQL verify-full (pg_hba только hostssl), HTTPS S3, AMQPS, OTLP/HTTPS с client cert, mTLS Collector -> Prometheus/Loki/Tempo, Grafana -> datasources, scrape. - OpenBao 2.7.2 не в dev-mode: Raft на volume, init/unseal оператором, auth/cert с ролью и policy на сервис, TTL 15 мин + renew/re-login. - Отдельные роли PostgreSQL с грантами по таблицам; разные ключи S3 и пользователи брокера выдаются из OpenBao. - Argon2id (m=19 MiB, t=2, p=1, замер на стенде) и rehash при входе. - Случайный X-Instance-Id; hardening контейнеров (non-root, cap_drop ALL, no-new-privileges, read-only, лимиты); TCP_NODELAY на сокете API. - Модель угроз, trust matrix, отрицательные тесты TLS/mTLS, API, OpenBao, hardening; evidence, в т.ч. прогон из чистого clone.
843 lines
31 KiB
YAML
843 lines
31 KiB
YAML
# Lab7 = lab6 + защита: HTTPS снаружи, mTLS LB->API, TLS ко всем зависимостям, OpenBao, hardening.
|
||
#
|
||
# клиент --HTTPS :8443 (public-ca)--> lb --mTLS (identity lb, internal-ca)--> api x2
|
||
# api/relay/worker/janitor --TLS verify-full--> db (PostgreSQL, отдельные роли *_svc)
|
||
# --HTTPS--> s3 (SeaweedFS, отдельные ключи)
|
||
# relay/worker --AMQPS 5671--> broker (RabbitMQ, отдельные пользователи и ACL)
|
||
# все процессы приложения --OTLP/HTTPS + client cert--> otel-collector --mTLS--> prometheus/loki/tempo
|
||
# grafana --HTTPS + client cert--> prometheus/loki/tempo; оператор --HTTPS :3000--> grafana
|
||
# api/relay/worker/janitor/migrate/s3-init/ops --TLS + cert auth--> openbao (KV v2, scoped policy)
|
||
#
|
||
# Сети (все internal, кроме frontend и grafana-ui): apinet (lb <-> api), backend (данные),
|
||
# telemetry, vault (OpenBao только для процессов приложения). LB не видит БД/S3/брокер/OpenBao.
|
||
# На хост опубликованы ТОЛЬКО 127.0.0.1:8443 (HTTPS API) и 127.0.0.1:3000 (HTTPS Grafana).
|
||
#
|
||
# Запуск (подробно — REPORT.md):
|
||
# ./scripts/init-secrets.sh # пароли, PKI; CA keys -> каталог оператора
|
||
# ./scripts/bao-init.sh # ОПЕРАТОР: init/unseal OpenBao, policies, cert roles, KV
|
||
# docker compose --env-file .env up -d --build --wait
|
||
name: mtusi-lab7
|
||
|
||
x-app-image: &app-image
|
||
image: mtusi-messenger:lab7
|
||
pull_policy: never
|
||
|
||
# Hardening общий для всех сервисов: без повышения привилегий (setuid-бинарники бесполезны).
|
||
x-sec: &sec
|
||
security_opt: ["no-new-privileges:true"]
|
||
|
||
# Процессы приложения: non-root (uid 10001 из образа), без capabilities, read-only rootfs,
|
||
# запись только в tmpfs /tmp (upload O_TMPFILE, рабочие файлы worker, heartbeat, instance-id).
|
||
x-app-sec: &app-sec
|
||
security_opt: ["no-new-privileges:true"]
|
||
cap_drop: [ALL]
|
||
read_only: true
|
||
user: "10001:10001"
|
||
|
||
x-otel-env: &otel-env
|
||
OTEL_SDK_DISABLED: ${OTEL_SDK_DISABLED:-false}
|
||
OTEL_EXPORTER_OTLP_ENDPOINT: https://otel-collector:4318
|
||
OTEL_EXPORTER_OTLP_PROTOCOL: http/protobuf
|
||
# Lab7: CA для проверки Collector и client cert сервиса (mTLS ingest).
|
||
OTEL_EXPORTER_OTLP_CERTIFICATE: /run/secrets/internal-ca.crt
|
||
OTEL_EXPORTER_OTLP_CLIENT_CERTIFICATE: /run/secrets/service.crt
|
||
OTEL_EXPORTER_OTLP_CLIENT_KEY: /run/secrets/service.key
|
||
OTEL_RESOURCE_ATTRIBUTES: deployment.environment.name=${APP_ENV:-lab}
|
||
OTEL_TRACES_SAMPLER: ${OTEL_TRACES_SAMPLER:-parentbased_always_on}
|
||
OTEL_TRACES_SAMPLER_ARG: ${OTEL_TRACES_SAMPLER_ARG:-1.0}
|
||
OTEL_METRIC_EXPORT_INTERVAL: ${OTEL_METRIC_EXPORT_INTERVAL:-10000}
|
||
OTEL_EXPORTER_OTLP_TIMEOUT_SECONDS: ${OTEL_EXPORTER_OTLP_TIMEOUT_SECONDS:-5}
|
||
OTEL_BSP_MAX_QUEUE_SIZE: ${OTEL_BSP_MAX_QUEUE_SIZE:-2048}
|
||
OTEL_BLRP_MAX_QUEUE_SIZE: ${OTEL_BLRP_MAX_QUEUE_SIZE:-2048}
|
||
OTEL_PYTHON_SDK_INTERNAL_METRICS_ENABLED: ${OTEL_PYTHON_SDK_INTERNAL_METRICS_ENABLED:-true}
|
||
|
||
# Секреты сервисов — из OpenBao: cert auth (свой leaf) -> token со scoped policy -> KV v2.
|
||
x-bao-env: &bao-env
|
||
BAO_ADDR: https://openbao:8200
|
||
BAO_CACERT: /run/secrets/internal-ca.crt
|
||
BAO_KV_MOUNT: kv
|
||
BAO_STARTUP_WAIT_SECONDS: ${BAO_STARTUP_WAIT_SECONDS:-300}
|
||
|
||
x-db-env: &db-env
|
||
POSTGRES_HOST: db
|
||
POSTGRES_PORT: "5432"
|
||
POSTGRES_DB: ${POSTGRES_DB:-messenger}
|
||
DB_SSLMODE: verify-full
|
||
|
||
x-s3-env: &s3-env
|
||
S3_ENDPOINT: https://s3:8333
|
||
S3_REGION: ${S3_REGION:-us-east-1}
|
||
S3_BUCKET: ${S3_BUCKET:-messenger-attachments}
|
||
S3_CONNECT_TIMEOUT_SECONDS: ${S3_CONNECT_TIMEOUT_SECONDS:-2}
|
||
S3_READ_TIMEOUT_SECONDS: ${S3_READ_TIMEOUT_SECONDS:-5}
|
||
|
||
x-job-env: &job-env
|
||
PIPELINE_VERSION: ${PIPELINE_VERSION:-1}
|
||
JOB_MAX_ATTEMPTS: ${JOB_MAX_ATTEMPTS:-5}
|
||
JOB_TIMEOUT_SECONDS: ${JOB_TIMEOUT_SECONDS:-60}
|
||
JOB_LEASE_SECONDS: ${JOB_LEASE_SECONDS:-30}
|
||
JOB_RETRY_BASE_SECONDS: ${JOB_RETRY_BASE_SECONDS:-2}
|
||
JOB_RETRY_MAX_SECONDS: ${JOB_RETRY_MAX_SECONDS:-60}
|
||
MAX_UPLOAD_BYTES: ${MAX_UPLOAD_BYTES:-10485760}
|
||
MAX_IMAGE_PIXELS: ${MAX_IMAGE_PIXELS:-40000000}
|
||
|
||
services:
|
||
# ===================================================================== secrets
|
||
# OpenBao НЕ в dev-mode: file storage на volume baodata, TLS-listener, после старта sealed.
|
||
# Init/unseal — оператор (scripts/bao-init.sh, scripts/bao-unseal.sh); unseal shares и root token
|
||
# лежат в каталоге оператора вне проекта и в контейнер не монтируются.
|
||
# healthcheck `bao status`: exit 0 только когда initialized + unsealed (sealed -> 2 -> unhealthy).
|
||
openbao:
|
||
image: openbao/openbao:2.7.2@sha256:5046911075e137e154ce49ab051502f4ec7cf367d484ae4a4dce665d9b90561f
|
||
entrypoint: ["bao", "server", "-config=/openbao/config/openbao.hcl"]
|
||
init: true
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
read_only: true
|
||
environment:
|
||
BAO_ADDR: https://127.0.0.1:8200
|
||
BAO_CACERT: /run/secrets/internal-ca.crt
|
||
SKIP_SETCAP: "true"
|
||
volumes:
|
||
- baodata:/openbao/file
|
||
- ./infra/openbao/openbao.hcl:/openbao/config/openbao.hcl:ro
|
||
tmpfs: ["/tmp:size=16m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: openbao_crt, target: service.crt}
|
||
- {source: openbao_key, target: service.key}
|
||
networks: [vault]
|
||
restart: unless-stopped
|
||
healthcheck:
|
||
test: ["CMD", "bao", "status"]
|
||
interval: 5s
|
||
start_interval: 1s
|
||
timeout: 3s
|
||
retries: 3
|
||
start_period: 10s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 256M}
|
||
|
||
# ===================================================================== данные
|
||
# PostgreSQL: ssl=on, pg_hba только hostssl + scram-sha-256 (plaintext reject). Стартовый root
|
||
# нужен entrypoint-у образа (chown данных, копия ключа TLS с правами 0600), затем uid postgres.
|
||
db:
|
||
image: postgres:16.4-alpine@sha256:5660c2cbfea50c7a9127d17dc4e48543eedd3d7a41a595a2dfa572471e37e64c
|
||
entrypoint: ["/bin/sh", "/etc/postgresql/start.sh"]
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
cap_add: [CHOWN, DAC_OVERRIDE, FOWNER, SETGID, SETUID]
|
||
environment:
|
||
POSTGRES_DB: ${POSTGRES_DB:-messenger}
|
||
POSTGRES_USER: postgres
|
||
POSTGRES_PASSWORD_FILE: /run/secrets/postgres_password
|
||
POSTGRES_INITDB_ARGS: "--auth-local=peer --auth-host=scram-sha-256"
|
||
secrets:
|
||
- postgres_password
|
||
- {source: db_crt, target: service.crt}
|
||
- {source: db_key, target: service.key}
|
||
volumes:
|
||
- pgdata:/var/lib/postgresql/data
|
||
- ./infra/postgres/pg_hba.conf:/etc/postgresql/pg_hba.conf:ro
|
||
- ./infra/postgres/start.sh:/etc/postgresql/start.sh:ro
|
||
networks: [backend]
|
||
restart: unless-stopped
|
||
healthcheck:
|
||
# от имени ОС-пользователя postgres (локальный сокет, peer): под root был бы «Peer authentication failed»
|
||
test: ["CMD-SHELL", "su-exec postgres pg_isready -q -U postgres -d \"$${POSTGRES_DB}\""]
|
||
interval: 2s
|
||
timeout: 3s
|
||
retries: 30
|
||
stop_grace_period: 30s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.0", memory: 512M}
|
||
|
||
# S3: SeaweedFS, S3 API только HTTPS :8333 (сертификат s3). master/volume/filer — 127.0.0.1
|
||
# контейнера (закрытый локальный hop, снаружи недостижим).
|
||
s3:
|
||
image: chrislusf/seaweedfs:4.50@sha256:c2658820e763294a9e9de7b36c81e59b879e3459148351aee4e31a126ca64860
|
||
command:
|
||
- server
|
||
- -dir=/data
|
||
- -ip=127.0.0.1
|
||
- -ip.bind=127.0.0.1
|
||
- -master.volumeSizeLimitMB=1024
|
||
- -master.electionTimeout=2s
|
||
- -volume.preStopSeconds=2
|
||
- -volume.max=0
|
||
- -s3
|
||
- -s3.ip.bind=0.0.0.0
|
||
- -s3.port=8333
|
||
- -s3.cert.file=/run/secrets/service.crt
|
||
- -s3.key.file=/run/secrets/service.key
|
||
- -s3.config=/run/secrets/s3_identities.json
|
||
- -s3.iam=false
|
||
- -s3.port.iceberg=0
|
||
- -s3.port.lance=0
|
||
- -s3.concurrentUploadLimitMB=64
|
||
- -filer.concurrentUploadLimitMB=64
|
||
- -volume.concurrentUploadLimitMB=64
|
||
- -s3.cacheCapacityMB=16
|
||
- -s3.readerCacheSizeMB=32
|
||
# entrypoint образа: chown /data и su-exec на пользователя seaweed (дальше процесс не root).
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
cap_add: [CHOWN, SETGID, SETUID]
|
||
secrets:
|
||
- s3_identities.json
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: s3_crt, target: service.crt}
|
||
- {source: s3_key, target: service.key}
|
||
environment:
|
||
GOMEMLIMIT: 512MiB
|
||
volumes:
|
||
- s3data:/data
|
||
networks: [backend]
|
||
restart: unless-stopped
|
||
healthcheck:
|
||
test: ["CMD-SHELL", "curl -sf -m 2 --cacert /run/secrets/internal-ca.crt https://127.0.0.1:8333/healthz -o /dev/null && curl -sf -m 2 -X PUT --data-binary ok http://127.0.0.1:8888/.healthcheck/probe -o /dev/null"]
|
||
interval: 5s
|
||
start_interval: 1s
|
||
timeout: 3s
|
||
retries: 3
|
||
start_period: 60s
|
||
stop_grace_period: 30s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.0", memory: 768M}
|
||
|
||
# RabbitMQ: только AMQPS 5671 и HTTPS-метрики 15691 (client cert обязателен). Узел rabbit@localhost:
|
||
# epmd и Erlang distribution слушают только 127.0.0.1 контейнера (кластерного канала нет).
|
||
broker:
|
||
image: rabbitmq:4.1.4-alpine@sha256:b736d649308e1b3e1a116c3f36986b605ee3d03e88f10166be2900083d2e63f2
|
||
hostname: broker
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
cap_add: [CHOWN, DAC_OVERRIDE, FOWNER, SETGID, SETUID]
|
||
environment:
|
||
RABBITMQ_NODENAME: rabbit@localhost
|
||
ERL_EPMD_ADDRESS: 127.0.0.1
|
||
RABBITMQ_SERVER_ADDITIONAL_ERL_ARGS: "-kernel inet_dist_use_interface {127,0,0,1}"
|
||
volumes:
|
||
- brokerdata:/var/lib/rabbitmq
|
||
- ./infra/rabbitmq/rabbitmq.conf:/etc/rabbitmq/conf.d/90-lab7.conf:ro
|
||
secrets:
|
||
- {source: rabbitmq_definitions.json, target: rabbitmq_definitions.json}
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: broker_crt, target: service.crt}
|
||
- {source: broker_key, target: service.key}
|
||
networks: [backend, telemetry]
|
||
restart: unless-stopped
|
||
healthcheck:
|
||
test: ["CMD-SHELL", "rabbitmq-diagnostics -q check_running && rabbitmq-diagnostics -q check_local_alarms && rabbitmq-diagnostics -q check_port_listener 5671 && rabbitmqctl -q -p messenger list_queues name | grep -q attachments.process.v1"]
|
||
interval: 10s
|
||
start_interval: 2s
|
||
timeout: 15s
|
||
retries: 5
|
||
start_period: 60s
|
||
stop_grace_period: 30s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.0", memory: 512M}
|
||
|
||
# ===================================================================== bootstrap jobs
|
||
migrate:
|
||
<<: [*app-image, *app-sec]
|
||
build:
|
||
context: .
|
||
args:
|
||
APP_VERSION: ${APP_VERSION:-1}
|
||
pull_policy: missing
|
||
command: ["python", "/app/src/migrate.py"]
|
||
environment:
|
||
<<: [*db-env, *bao-env]
|
||
POSTGRES_USER: postgres
|
||
# migrate импортирует server.py (STORE/MIGRATIONS_DIR); S3-клиент создаётся, но не используется.
|
||
S3_ENDPOINT: https://s3:8333
|
||
S3_CLIENT_REQUIRED: "0"
|
||
BAO_AUTH_ROLE: migrate
|
||
BAO_KV_PATH: messenger/migrate
|
||
tmpfs: ["/tmp:size=8m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: migrate_crt, target: service.crt}
|
||
- {source: migrate_key, target: service.key}
|
||
networks: [backend, vault]
|
||
healthcheck: {disable: true}
|
||
restart: "no"
|
||
depends_on:
|
||
db: {condition: service_healthy}
|
||
openbao: {condition: service_healthy}
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 128M}
|
||
|
||
s3-init:
|
||
<<: [*app-image, *app-sec]
|
||
command: ["python", "/app/src/s3_bootstrap.py"]
|
||
environment:
|
||
<<: [*s3-env, *bao-env]
|
||
BAO_AUTH_ROLE: s3-init
|
||
BAO_KV_PATH: messenger/s3-init
|
||
tmpfs: ["/tmp:size=8m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: s3_init_crt, target: service.crt}
|
||
- {source: s3_init_key, target: service.key}
|
||
networks: [backend, vault]
|
||
healthcheck: {disable: true}
|
||
restart: "no"
|
||
depends_on:
|
||
s3: {condition: service_healthy}
|
||
openbao: {condition: service_healthy}
|
||
migrate: {condition: service_completed_successfully}
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 128M}
|
||
|
||
# ===================================================================== приложение
|
||
api:
|
||
<<: [*app-image, *app-sec]
|
||
expose: ["8443"]
|
||
environment:
|
||
<<: [*db-env, *s3-env, *job-env, *otel-env, *bao-env]
|
||
POSTGRES_USER: api_svc
|
||
BAO_AUTH_ROLE: api
|
||
BAO_KV_PATH: messenger/api
|
||
APP_HOST: 0.0.0.0
|
||
APP_PORT: "8443"
|
||
MTLS_ALLOWED_CLIENTS: lb
|
||
HEALTH_HOST: 127.0.0.1
|
||
HEALTH_PORT: "8081"
|
||
APP_ENV: ${APP_ENV:-lab}
|
||
SESSION_TTL_SECONDS: ${SESSION_TTL_SECONDS:-3600}
|
||
IDEMPOTENCY_TTL_SECONDS: ${IDEMPOTENCY_TTL_SECONDS:-86400}
|
||
ARGON2_MEMORY_KIB: ${ARGON2_MEMORY_KIB:-19456}
|
||
ARGON2_ITERATIONS: ${ARGON2_ITERATIONS:-2}
|
||
ARGON2_PARALLELISM: ${ARGON2_PARALLELISM:-1}
|
||
DB_POOL_SIZE: ${DB_POOL_SIZE:-10}
|
||
DRAIN_SECONDS: ${DRAIN_SECONDS:-3}
|
||
SHUTDOWN_TIMEOUT_SECONDS: ${SHUTDOWN_TIMEOUT_SECONDS:-10}
|
||
PASSWORD_HASH_CONCURRENCY: ${PASSWORD_HASH_CONCURRENCY:-2}
|
||
PASSWORD_HASH_WAIT_SECONDS: ${PASSWORD_HASH_WAIT_SECONDS:-0.5}
|
||
DB_HEALTH_INTERVAL_SECONDS: ${DB_HEALTH_INTERVAL_SECONDS:-1}
|
||
READY_STALE_SECONDS: ${READY_STALE_SECONDS:-3}
|
||
UPLOAD_TIMEOUT_SECONDS: ${UPLOAD_TIMEOUT_SECONDS:-60}
|
||
JSON_BODY_TIMEOUT_SECONDS: ${JSON_BODY_TIMEOUT_SECONDS:-10}
|
||
UPLOAD_CONCURRENCY: ${UPLOAD_CONCURRENCY:-8}
|
||
DECODE_CONCURRENCY: ${DECODE_CONCURRENCY:-1}
|
||
JPEG_PROBE_MEMORY_MB: ${JPEG_PROBE_MEMORY_MB:-384}
|
||
JPEG_PROBE_CPU_SECONDS: ${JPEG_PROBE_CPU_SECONDS:-10}
|
||
JPEG_PROBE_TIMEOUT_SECONDS: ${JPEG_PROBE_TIMEOUT_SECONDS:-15}
|
||
S3_PUT_CONCURRENCY: ${S3_PUT_CONCURRENCY:-3}
|
||
FAULT_INJECTION: ${FAULT_INJECTION:-0}
|
||
FAULT_SLOW_HANDLER_SECONDS: ${FAULT_SLOW_HANDLER_SECONDS:-1.5}
|
||
OTEL_SERVICE_NAME: messenger-api
|
||
tmpfs: ["/tmp:size=160m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: api_crt, target: service.crt}
|
||
- {source: api_key, target: service.key}
|
||
# У API нет доступа к брокеру (задание пишется в outbox PostgreSQL).
|
||
networks: [apinet, backend, telemetry, vault]
|
||
restart: unless-stopped
|
||
stop_grace_period: 20s
|
||
depends_on:
|
||
db: {condition: service_healthy}
|
||
openbao: {condition: service_healthy}
|
||
migrate: {condition: service_completed_successfully}
|
||
s3-init: {condition: service_completed_successfully}
|
||
deploy:
|
||
replicas: ${API_REPLICAS:-2}
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 640M, pids: 256}
|
||
|
||
relay:
|
||
<<: [*app-image, *app-sec]
|
||
command: ["python", "/app/src/relay.py"]
|
||
environment:
|
||
<<: [*db-env, *job-env, *otel-env, *bao-env]
|
||
POSTGRES_USER: relay_svc
|
||
BAO_AUTH_ROLE: relay
|
||
BAO_KV_PATH: messenger/relay
|
||
SERVICE_NAME: relay
|
||
RELAY_POLL_SECONDS: ${RELAY_POLL_SECONDS:-1}
|
||
REAPER_INTERVAL_SECONDS: ${REAPER_INTERVAL_SECONDS:-5}
|
||
JOB_REDRIVE_AFTER_SECONDS: ${JOB_REDRIVE_AFTER_SECONDS:-900}
|
||
OUTBOX_RETENTION_SECONDS: ${OUTBOX_RETENTION_SECONDS:-86400}
|
||
RELAY_FAULT: ${RELAY_FAULT:-}
|
||
OTEL_SERVICE_NAME: messenger-relay
|
||
tmpfs: ["/tmp:size=8m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: relay_crt, target: service.crt}
|
||
- {source: relay_key, target: service.key}
|
||
networks: [backend, telemetry, vault]
|
||
healthcheck:
|
||
test: ["CMD", "python", "-c", "import os,sys,time; sys.exit(0 if time.time() - os.path.getmtime('/tmp/relay.heartbeat') < 30 else 1)"]
|
||
interval: 10s
|
||
timeout: 3s
|
||
retries: 3
|
||
start_period: 30s
|
||
restart: unless-stopped
|
||
stop_grace_period: 15s
|
||
depends_on:
|
||
migrate: {condition: service_completed_successfully}
|
||
broker: {condition: service_healthy}
|
||
openbao: {condition: service_healthy}
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.25", memory: 128M, pids: 64}
|
||
|
||
# Upload-worker: ограничен по CPU/RAM (cgroup) и по времени/памяти задания (подпроцесс transform:
|
||
# RLIMIT_AS/RLIMIT_CPU + JOB_TIMEOUT_SECONDS).
|
||
worker:
|
||
<<: [*app-image, *app-sec]
|
||
command: ["python", "/app/src/worker.py"]
|
||
environment:
|
||
<<: [*db-env, *s3-env, *job-env, *otel-env, *bao-env]
|
||
POSTGRES_USER: worker_svc
|
||
BAO_AUTH_ROLE: worker
|
||
BAO_KV_PATH: messenger/worker
|
||
SERVICE_NAME: worker
|
||
S3_READ_TIMEOUT_SECONDS: ${WORKER_S3_READ_TIMEOUT_SECONDS:-10}
|
||
WORKER_CONCURRENCY: ${WORKER_CONCURRENCY:-2}
|
||
WORKER_PREFETCH: ${WORKER_PREFETCH:-1}
|
||
TRANSFORM_MEMORY_LIMIT_MB: ${TRANSFORM_MEMORY_LIMIT_MB:-768}
|
||
TRANSFORM_CPU_SECONDS: ${TRANSFORM_CPU_SECONDS:-30}
|
||
WORKER_TMP_DIR: /tmp/work
|
||
WORKER_FAULT_INJECTION: ${WORKER_FAULT_INJECTION:-0}
|
||
FAULT_SLOW_SECONDS: ${FAULT_SLOW_SECONDS:-20}
|
||
OTEL_SERVICE_NAME: messenger-worker
|
||
tmpfs: ["/tmp:size=96m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: worker_crt, target: service.crt}
|
||
- {source: worker_key, target: service.key}
|
||
networks: [backend, telemetry, vault]
|
||
healthcheck:
|
||
test: ["CMD", "python", "-c", "import os,sys,time; sys.exit(0 if time.time() - os.path.getmtime('/tmp/worker.heartbeat') < 30 else 1)"]
|
||
interval: 10s
|
||
timeout: 3s
|
||
retries: 3
|
||
start_period: 30s
|
||
restart: unless-stopped
|
||
stop_grace_period: 75s
|
||
depends_on:
|
||
migrate: {condition: service_completed_successfully}
|
||
s3-init: {condition: service_completed_successfully}
|
||
broker: {condition: service_healthy}
|
||
openbao: {condition: service_healthy}
|
||
deploy:
|
||
replicas: ${WORKER_REPLICAS:-2}
|
||
resources:
|
||
limits: {cpus: "1.0", memory: 1G, pids: 128}
|
||
|
||
janitor:
|
||
<<: [*app-image, *app-sec]
|
||
command: ["python", "/app/src/janitor.py"]
|
||
environment:
|
||
<<: [*db-env, *s3-env, *otel-env, *bao-env]
|
||
POSTGRES_USER: janitor_svc
|
||
BAO_AUTH_ROLE: janitor
|
||
BAO_KV_PATH: messenger/janitor
|
||
ATTACHMENT_GRACE_SECONDS: ${ATTACHMENT_GRACE_SECONDS:-3600}
|
||
JANITOR_INTERVAL_SECONDS: ${JANITOR_INTERVAL_SECONDS:-300}
|
||
STAGING_RETENTION_SECONDS: ${STAGING_RETENTION_SECONDS:-86400}
|
||
OTEL_SERVICE_NAME: messenger-janitor
|
||
tmpfs: ["/tmp:size=8m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: janitor_crt, target: service.crt}
|
||
- {source: janitor_key, target: service.key}
|
||
networks: [backend, telemetry, vault]
|
||
healthcheck:
|
||
test: ["CMD", "python", "-c", "import os,sys,time; sys.exit(0 if time.time() - os.path.getmtime('/tmp/janitor.heartbeat') < 90 else 1)"]
|
||
interval: 15s
|
||
timeout: 3s
|
||
retries: 2
|
||
start_period: 20s
|
||
restart: unless-stopped
|
||
depends_on:
|
||
migrate: {condition: service_completed_successfully}
|
||
s3-init: {condition: service_completed_successfully}
|
||
openbao: {condition: service_healthy}
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.25", memory: 160M, pids: 64}
|
||
|
||
# Ingress: HTTPS :8443 наружу (только 127.0.0.1 хоста), mTLS к API. Не видит БД/S3/OpenBao.
|
||
lb:
|
||
image: haproxy:3.0.5-alpine@sha256:43a2591a6a62a41dd22edb4452880e9c70de489fbd7c54ffb52d035cccbfe949
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
read_only: true
|
||
volumes:
|
||
- ./infra/haproxy/haproxy.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
|
||
environment:
|
||
LOGIN_RATE_LIMIT: ${LOGIN_RATE_LIMIT:-20}
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: lb_public_pem, target: lb-public.pem}
|
||
- {source: lb_pem, target: lb.pem}
|
||
ports:
|
||
- "127.0.0.1:${LB_PORT:-8443}:8443"
|
||
networks: [frontend, apinet, telemetry]
|
||
restart: unless-stopped
|
||
healthcheck:
|
||
test: ["CMD", "haproxy", "-c", "-q", "-f", "/usr/local/etc/haproxy/haproxy.cfg"]
|
||
interval: 10s
|
||
timeout: 3s
|
||
retries: 3
|
||
depends_on:
|
||
api: {condition: service_started}
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 128M}
|
||
|
||
# Операторский CLI: docker compose --profile tools run --rm ops <команда>
|
||
ops:
|
||
<<: [*app-image, *app-sec]
|
||
profiles: [tools]
|
||
entrypoint: ["python", "/app/src/ops.py"]
|
||
environment:
|
||
<<: [*db-env, *s3-env, *job-env, *bao-env]
|
||
POSTGRES_USER: ops_svc
|
||
BAO_AUTH_ROLE: ops
|
||
BAO_KV_PATH: messenger/ops
|
||
SERVICE_NAME: ops
|
||
tmpfs: ["/tmp:size=8m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: ops_crt, target: service.crt}
|
||
- {source: ops_key, target: service.key}
|
||
networks: [backend, vault]
|
||
healthcheck: {disable: true}
|
||
restart: "no"
|
||
|
||
# Проверки ВНУТРИ стенда (отрицательные mTLS/TLS-тесты, права S3, варианты). Только для тестов:
|
||
# получает тестовые identities (intruder, rogue-lb) и — для положительной проверки mTLS — lb.
|
||
checks:
|
||
<<: [*app-image, *app-sec]
|
||
profiles: [tools]
|
||
entrypoint: ["python"]
|
||
environment:
|
||
<<: *s3-env
|
||
PYTHONDONTWRITEBYTECODE: "1"
|
||
tmpfs: ["/tmp:size=64m,mode=1777"]
|
||
volumes:
|
||
- ./student-tests:/app/student-tests:ro
|
||
- ./tests:/app/tests:ro
|
||
- ./contracts:/app/contracts:ro
|
||
- ./lab.json:/app/lab.json:ro
|
||
- ./evidence:/app/evidence
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: public_ca, target: public-ca.crt}
|
||
- {source: lb_crt, target: lb.crt}
|
||
- {source: lb_key, target: lb.key}
|
||
- {source: intruder_crt, target: intruder.crt}
|
||
- {source: intruder_key, target: intruder.key}
|
||
- {source: rogue_ca, target: rogue-ca.crt}
|
||
- {source: rogue_lb_crt, target: rogue-lb.crt}
|
||
- {source: rogue_lb_key, target: rogue-lb.key}
|
||
- s3_api_access_key
|
||
- s3_api_secret_key
|
||
- s3_worker_access_key
|
||
- s3_worker_secret_key
|
||
- s3_janitor_access_key
|
||
- s3_janitor_secret_key
|
||
networks: [apinet, backend, telemetry, vault]
|
||
healthcheck: {disable: true}
|
||
restart: "no"
|
||
|
||
# ===================================================================== телеметрия (lab6 + TLS)
|
||
otel-collector:
|
||
image: otel/opentelemetry-collector-contrib:0.163.2@sha256:7f22d2a6748110ffb6a59bc0df448bccbf501cdb667d610ede4c61d3863d5849
|
||
command: ["--config=/etc/otelcol/collector.yaml"]
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
read_only: true
|
||
volumes:
|
||
- ./infra/otel/collector.yaml:/etc/otelcol/collector.yaml:ro
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: otel_collector_crt, target: service.crt}
|
||
- {source: otel_collector_key, target: service.key}
|
||
environment:
|
||
GOMEMLIMIT: 200MiB
|
||
networks: [telemetry]
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 256M}
|
||
|
||
prometheus:
|
||
image: prom/prometheus:v3.13.5@sha256:aafa9d694e6a883c02db516d9f6e5b84e14195f49a2a6456a4cca815edda72b9
|
||
command:
|
||
- --config.file=/etc/prometheus/prometheus.yml
|
||
- --web.config.file=/etc/prometheus/web.yml
|
||
- --storage.tsdb.path=/prometheus
|
||
- --storage.tsdb.retention.time=15d
|
||
- --storage.tsdb.retention.size=2GB
|
||
- --web.enable-otlp-receiver
|
||
- --enable-feature=exemplar-storage
|
||
- --web.listen-address=:9090
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
read_only: true
|
||
volumes:
|
||
- ./infra/prometheus/prometheus.yml:/etc/prometheus/prometheus.yml:ro
|
||
- ./infra/prometheus/web.yml:/etc/prometheus/web.yml:ro
|
||
- ./infra/prometheus/client.yml:/etc/prometheus/client.yml:ro
|
||
- ./infra/prometheus/rules:/etc/prometheus/rules:ro
|
||
- promdata:/prometheus
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: prometheus_crt, target: service.crt}
|
||
- {source: prometheus_key, target: service.key}
|
||
networks: [telemetry]
|
||
restart: unless-stopped
|
||
healthcheck:
|
||
test: ["CMD", "promtool", "check", "ready", "--url=https://prometheus:9090", "--http.config.file=/etc/prometheus/client.yml"]
|
||
interval: 10s
|
||
start_interval: 2s
|
||
timeout: 3s
|
||
retries: 5
|
||
start_period: 30s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 512M}
|
||
|
||
loki:
|
||
image: grafana/loki:3.6.19@sha256:0f9cbc340cbe33ee498e654075034336962057db072f57b37f253e14d476443a
|
||
command: ["-config.file=/etc/loki/loki.yaml"]
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
read_only: true
|
||
volumes:
|
||
- ./infra/loki/loki.yaml:/etc/loki/loki.yaml:ro
|
||
- lokidata:/loki
|
||
tmpfs: ["/tmp:size=32m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: loki_crt, target: service.crt}
|
||
- {source: loki_key, target: service.key}
|
||
networks: [telemetry]
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 512M}
|
||
|
||
tempo:
|
||
image: grafana/tempo:2.10.10@sha256:12b0820f13ffa10ce956b6e9acc3b03f3cffded99109be6cc09d106fe56e5d38
|
||
command: ["-config.file=/etc/tempo/tempo.yaml"]
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
read_only: true
|
||
volumes:
|
||
- ./infra/tempo/tempo.yaml:/etc/tempo/tempo.yaml:ro
|
||
- tempodata:/var/tempo
|
||
tmpfs: ["/tmp:size=32m,mode=1777"]
|
||
secrets:
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: tempo_crt, target: service.crt}
|
||
- {source: tempo_key, target: service.key}
|
||
networks: [telemetry]
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 512M}
|
||
|
||
# Grafana: HTTPS (public-ca) на 127.0.0.1:3000; к datasources — HTTPS + client cert grafana.
|
||
grafana:
|
||
image: grafana/grafana:12.4.14@sha256:ae509d9ae8c1021ca32ec370a9ca975d40165eda9fccd1e1102d317ff8787cd7
|
||
<<: *sec
|
||
cap_drop: [ALL]
|
||
environment:
|
||
GF_SERVER_PROTOCOL: https
|
||
GF_SERVER_CERT_FILE: /run/secrets/public.crt
|
||
GF_SERVER_CERT_KEY: /run/secrets/public.key
|
||
GF_SERVER_MIN_TLS_VERSION: TLS1.2
|
||
GF_SECURITY_ADMIN_USER: admin
|
||
GF_SECURITY_ADMIN_PASSWORD__FILE: /run/secrets/grafana_admin_password
|
||
GF_SECURITY_COOKIE_SECURE: "true"
|
||
GF_SECURITY_COOKIE_SAMESITE: strict
|
||
GF_SECURITY_DISABLE_GRAVATAR: "true"
|
||
GF_USERS_ALLOW_SIGN_UP: "false"
|
||
GF_AUTH_ANONYMOUS_ENABLED: "false"
|
||
GF_ANALYTICS_REPORTING_ENABLED: "false"
|
||
GF_ANALYTICS_CHECK_FOR_UPDATES: "false"
|
||
GF_ANALYTICS_CHECK_FOR_PLUGIN_UPDATES: "false"
|
||
GF_NEWS_NEWS_FEED_ENABLED: "false"
|
||
GF_PLUGINS_PREINSTALL_DISABLED: "true"
|
||
GF_DASHBOARDS_DEFAULT_HOME_DASHBOARD_PATH: /var/lib/grafana/dashboards/messenger.json
|
||
secrets:
|
||
- grafana_admin_password
|
||
- {source: internal_ca, target: internal-ca.crt}
|
||
- {source: public_ca, target: public-ca.crt}
|
||
- {source: grafana_crt, target: service.crt}
|
||
- {source: grafana_key, target: service.key}
|
||
- {source: grafana_public_crt, target: public.crt}
|
||
- {source: grafana_public_key, target: public.key}
|
||
volumes:
|
||
- ./infra/grafana/provisioning:/etc/grafana/provisioning:ro
|
||
- ./infra/grafana/dashboards:/var/lib/grafana/dashboards:ro
|
||
- grafanadata:/var/lib/grafana
|
||
ports:
|
||
- "127.0.0.1:${GRAFANA_PORT:-3000}:3000"
|
||
networks: [telemetry, grafana-ui]
|
||
restart: unless-stopped
|
||
healthcheck:
|
||
test: ["CMD", "curl", "-sf", "-m", "2", "--cacert", "/run/secrets/public-ca.crt", "https://localhost:3000/api/health", "-o", "/dev/null"]
|
||
interval: 10s
|
||
start_interval: 2s
|
||
timeout: 3s
|
||
retries: 6
|
||
start_period: 60s
|
||
depends_on:
|
||
prometheus: {condition: service_started}
|
||
loki: {condition: service_started}
|
||
tempo: {condition: service_started}
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.5", memory: 384M}
|
||
|
||
secrets:
|
||
internal_ca:
|
||
file: ./secrets/pki/internal-ca.crt
|
||
public_ca:
|
||
file: ./secrets/pki/public-ca.crt
|
||
lb_public_pem:
|
||
file: ./secrets/pki/lb-public.pem
|
||
lb_pem:
|
||
file: ./secrets/pki/lb.pem
|
||
grafana_public_crt:
|
||
file: ./secrets/pki/grafana-public.crt
|
||
grafana_public_key:
|
||
file: ./secrets/pki/grafana-public.key
|
||
lb_crt:
|
||
file: ./secrets/pki/lb.crt
|
||
lb_key:
|
||
file: ./secrets/pki/lb.key
|
||
api_crt:
|
||
file: ./secrets/pki/api.crt
|
||
api_key:
|
||
file: ./secrets/pki/api.key
|
||
worker_crt:
|
||
file: ./secrets/pki/worker.crt
|
||
worker_key:
|
||
file: ./secrets/pki/worker.key
|
||
relay_crt:
|
||
file: ./secrets/pki/relay.crt
|
||
relay_key:
|
||
file: ./secrets/pki/relay.key
|
||
janitor_crt:
|
||
file: ./secrets/pki/janitor.crt
|
||
janitor_key:
|
||
file: ./secrets/pki/janitor.key
|
||
migrate_crt:
|
||
file: ./secrets/pki/migrate.crt
|
||
migrate_key:
|
||
file: ./secrets/pki/migrate.key
|
||
s3_init_crt:
|
||
file: ./secrets/pki/s3-init.crt
|
||
s3_init_key:
|
||
file: ./secrets/pki/s3-init.key
|
||
ops_crt:
|
||
file: ./secrets/pki/ops.crt
|
||
ops_key:
|
||
file: ./secrets/pki/ops.key
|
||
grafana_crt:
|
||
file: ./secrets/pki/grafana.crt
|
||
grafana_key:
|
||
file: ./secrets/pki/grafana.key
|
||
db_crt:
|
||
file: ./secrets/pki/db.crt
|
||
db_key:
|
||
file: ./secrets/pki/db.key
|
||
s3_crt:
|
||
file: ./secrets/pki/s3.crt
|
||
s3_key:
|
||
file: ./secrets/pki/s3.key
|
||
broker_crt:
|
||
file: ./secrets/pki/broker.crt
|
||
broker_key:
|
||
file: ./secrets/pki/broker.key
|
||
otel_collector_crt:
|
||
file: ./secrets/pki/otel-collector.crt
|
||
otel_collector_key:
|
||
file: ./secrets/pki/otel-collector.key
|
||
prometheus_crt:
|
||
file: ./secrets/pki/prometheus.crt
|
||
prometheus_key:
|
||
file: ./secrets/pki/prometheus.key
|
||
loki_crt:
|
||
file: ./secrets/pki/loki.crt
|
||
loki_key:
|
||
file: ./secrets/pki/loki.key
|
||
tempo_crt:
|
||
file: ./secrets/pki/tempo.crt
|
||
tempo_key:
|
||
file: ./secrets/pki/tempo.key
|
||
openbao_crt:
|
||
file: ./secrets/pki/openbao.crt
|
||
openbao_key:
|
||
file: ./secrets/pki/openbao.key
|
||
# тестовые identities (только сервис checks)
|
||
intruder_crt:
|
||
file: ./secrets/pki/test/intruder.crt
|
||
intruder_key:
|
||
file: ./secrets/pki/test/intruder.key
|
||
rogue_ca:
|
||
file: ./secrets/pki/test/rogue-ca.crt
|
||
rogue_lb_crt:
|
||
file: ./secrets/pki/test/rogue-lb.crt
|
||
rogue_lb_key:
|
||
file: ./secrets/pki/test/rogue-lb.key
|
||
# серверная сторона хранилищ и тестовые ключи S3
|
||
postgres_password:
|
||
file: ./secrets/postgres_password
|
||
grafana_admin_password:
|
||
file: ./secrets/grafana_admin_password
|
||
s3_identities.json:
|
||
file: ./secrets/s3_identities.json
|
||
rabbitmq_definitions.json:
|
||
file: ./secrets/rabbitmq_definitions.json
|
||
s3_api_access_key:
|
||
file: ./secrets/s3_api_access_key
|
||
s3_api_secret_key:
|
||
file: ./secrets/s3_api_secret_key
|
||
s3_worker_access_key:
|
||
file: ./secrets/s3_worker_access_key
|
||
s3_worker_secret_key:
|
||
file: ./secrets/s3_worker_secret_key
|
||
s3_janitor_access_key:
|
||
file: ./secrets/s3_janitor_access_key
|
||
s3_janitor_secret_key:
|
||
file: ./secrets/s3_janitor_secret_key
|
||
|
||
networks:
|
||
frontend: {}
|
||
apinet:
|
||
internal: true
|
||
backend:
|
||
internal: true
|
||
telemetry:
|
||
internal: true
|
||
vault:
|
||
internal: true
|
||
grafana-ui: {}
|
||
|
||
volumes:
|
||
baodata: {}
|
||
pgdata: {}
|
||
s3data: {}
|
||
brokerdata: {}
|
||
promdata: {}
|
||
lokidata: {}
|
||
tempodata: {}
|
||
grafanadata: {}
|