forked from mtusi2026/mtusi2026lab7
785 lines
30 KiB
YAML
785 lines
30 KiB
YAML
# Стенд lab7: тот же мессенджер lab6, защищённый на всех границах доверия (contracts/trust-matrix.md).
|
||
#
|
||
# клиент --TLS--> lb (HAProxy, 127.0.0.1:8443) --mTLS (identity lb)--> api x2
|
||
# api/worker --TLS verify-full--> postgres (роли messenger_api / messenger_worker)
|
||
# api/worker --HTTPS--> s3 (SeaweedFS) (разные ключи и префиксы)
|
||
# api/worker --mTLS + SASL EXTERNAL--> broker (пользователи = CN: api публикует, worker читает)
|
||
# api --TLS + ACL-пароль--> valkey (кэш сессий, лимиты входа)
|
||
# api/worker --mTLS (allowlist api,worker)--> otlp-gw --127.0.0.1--> otel-collector
|
||
# otel-collector --mTLS--> prometheus / tsdb-gw (loki, tempo) <--mTLS-- grafana (HTTPS 127.0.0.1:3000 для оператора)
|
||
# api/worker --mTLS cert auth--> openbao (секреты: пароль роли БД, ключи S3, пароль Valkey)
|
||
#
|
||
# Запуск из чистого clone (подробно — REPORT.md):
|
||
# make secrets pki && docker compose up -d --wait openbao && make bao-init bao-unseal bao-bootstrap && make up
|
||
#
|
||
# Сети (все internal, кроме public и ops — они только для опубликованных на 127.0.0.1 портов):
|
||
# public lb -> 127.0.0.1:8443 (API), :8444 (API с коротким TTL), :8404 (stats)
|
||
# edge lb <-> api
|
||
# backend api, worker, migrate, s3-init <-> postgres, valkey, s3, broker
|
||
# otlp api, worker -> otlp-gw (otel-collector)
|
||
# telemetry otlp-gw/otel-collector, lb (scrape) <-> prometheus, tsdb-gw (loki, tempo) <- grafana
|
||
# vault api, worker, ops -> openbao
|
||
# ops grafana -> 127.0.0.1:3000
|
||
#
|
||
# Hardening (x-hardened): no-new-privileges, cap_drop ALL, read-only корень + tmpfs там, где совместимо.
|
||
# Исключения (stateful/entrypoint как root) объяснены у сервисов и в REPORT.md.
|
||
# Секреты: TLS-ключ каждого сервиса смонтирован только в его контейнер (secrets/tls/<identity>);
|
||
# пароли и ключи приложения — только в OpenBao. Docker socket, privileged и host network не используются.
|
||
name: mtusi-lab7
|
||
|
||
x-build: &build
|
||
context: .
|
||
args:
|
||
APP_VERSION: ${APP_VERSION:-dev}
|
||
|
||
x-hardened: &hardened
|
||
security_opt: ["no-new-privileges:true"]
|
||
cap_drop: [ALL]
|
||
|
||
# Только публичный CA-сертификат и CRL — во все сервисы.
|
||
x-ca: &ca ./secrets/tls/ca:/run/ca:ro
|
||
|
||
# Общие (несекретные) настройки процессов приложения. Секретов здесь нет: см. secretsource.py.
|
||
x-app-env: &app-env
|
||
TLS_CA_FILE: /run/ca/ca.crt
|
||
TLS_CRL_FILE: /run/ca/crl.pem
|
||
TLS_CERT_FILE: /run/tls/tls.crt
|
||
TLS_KEY_FILE: /run/tls/tls.key
|
||
# Только наш CA для всех исходящих HTTPS (requests внутри OTel-экспортёра читает эти переменные).
|
||
SSL_CERT_FILE: /run/ca/ca.crt
|
||
REQUESTS_CA_BUNDLE: /run/ca/ca.crt
|
||
CACHE_URL: valkeys://messenger@valkey:6379/0
|
||
BROKER_URL: amqps://broker:5671/messenger
|
||
BAO_ADDR: https://openbao:8200
|
||
OTEL_EXPORTER_OTLP_CERTIFICATE: /run/ca/ca.crt
|
||
OTEL_EXPORTER_OTLP_CLIENT_CERTIFICATE: /run/tls/tls.crt
|
||
OTEL_EXPORTER_OTLP_CLIENT_KEY: /run/tls/tls.key
|
||
# Пустые *_FILE: у приложения нет файлов-секретов (иначе config.py стал бы их искать).
|
||
DATABASE_URL_FILE: ""
|
||
CACHE_URL_FILE: ""
|
||
S3_ACCESS_KEY_ID_FILE: ""
|
||
S3_SECRET_ACCESS_KEY_FILE: ""
|
||
BROKER_URL_FILE: ""
|
||
|
||
x-api-env: &api-env
|
||
<<: *app-env
|
||
DATABASE_URL: postgresql://messenger_api@postgres:5432/messenger?sslmode=verify-full&sslrootcert=/run/ca/ca.crt
|
||
BAO_ROLE: messenger-api
|
||
BAO_SECRET_PATH: secret/data/messenger/api
|
||
TLS_ALLOWED_CLIENTS: lb
|
||
|
||
x-api-healthcheck: &api-healthcheck
|
||
# Внутренний слушатель 127.0.0.1:8081 контейнера (healthcheck.py): основной порт требует mTLS lb.
|
||
test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://127.0.0.1:8081/health/ready', timeout=2)"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 3
|
||
start_period: 40s
|
||
start_interval: 1s
|
||
|
||
x-api: &api
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
env_file: .env
|
||
read_only: true
|
||
# /tmp: alias реплики, временные файлы загрузок (UPLOAD_SPOOL_BYTES), точки отказа faults.py.
|
||
tmpfs: ["/tmp:size=128m,mode=1777"]
|
||
volumes:
|
||
- ./secrets/tls/api:/run/tls:ro
|
||
- *ca
|
||
networks: [edge, backend, otlp, vault]
|
||
|
||
services:
|
||
# ---------- инструменты оператора (profile tools, не часть работающего стенда) ----------
|
||
|
||
# Локальный CA и сертификаты: make pki (scripts/pki.py). Сеть не нужна.
|
||
pki:
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
build: *build
|
||
profiles: [tools]
|
||
user: "0:0" # пишет в ./secrets на хосте (bind mount)
|
||
entrypoint: ["python", "scripts/pki.py"]
|
||
command: ["init"]
|
||
environment:
|
||
PKI_ROOT: /srv/secrets
|
||
volumes:
|
||
- ./scripts:/srv/scripts:ro
|
||
- ./secrets:/srv/secrets
|
||
network_mode: none
|
||
read_only: true
|
||
|
||
# Операции OpenBao и ротация: make bao-* / rotate-db-secret (scripts/bao.py). Только сети vault и backend.
|
||
ops:
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
profiles: [tools]
|
||
pull_policy: never
|
||
user: "0:0" # пишет secrets/operator/ (shares, root token) с mode 600
|
||
entrypoint: ["python", "scripts/bao.py"]
|
||
command: ["status"]
|
||
environment:
|
||
BAO_ADDR: https://openbao:8200
|
||
BAO_CACERT: /srv/secrets/tls/ca/ca.crt
|
||
SECRETS_DIR: /srv/secrets
|
||
BAO_POLICY_DIR: /srv/infra/openbao/policies
|
||
DB_HOST: postgres
|
||
volumes:
|
||
- ./scripts:/srv/scripts:ro
|
||
- ./secrets:/srv/secrets
|
||
- ./infra/openbao/policies:/srv/infra/openbao/policies:ro
|
||
networks: [vault, backend]
|
||
read_only: true
|
||
|
||
# Контейнер для собственных проверок матрицы доверия изнутри сетей стенда (make tls-matrix, acl-matrix):
|
||
# сертификаты сервисов (secrets/tls) и тестовые сертификаты (secrets/tls-test) — только чтение.
|
||
sectest:
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
profiles: [tools]
|
||
pull_policy: never
|
||
entrypoint: ["python"]
|
||
command: ["--version"] # скрипт передаёт make (tls-matrix, acl-matrix)
|
||
environment:
|
||
PKI_DIR: /pki
|
||
volumes:
|
||
- ./student-tests:/srv/student-tests:ro
|
||
- ./secrets/tls:/pki/tls:ro
|
||
- ./secrets/tls-test:/pki/test:ro
|
||
networks: [edge, backend, otlp, telemetry, vault]
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=16m"]
|
||
|
||
# ---------- приложение ----------
|
||
|
||
# Миграции (роль-владелец messenger, пароль из файла — это одноразовый инструмент, не приложение).
|
||
migrate:
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
build: *build
|
||
env_file: .env
|
||
command: ["python", "-m", "messenger.migrate"]
|
||
environment:
|
||
DATABASE_URL: postgresql://messenger@postgres:5432/messenger?sslmode=verify-full&sslrootcert=/run/ca/ca.crt
|
||
DATABASE_URL_FILE: ""
|
||
DB_PASSWORD_FILE: /run/secrets/postgres_password
|
||
CACHE_URL_FILE: ""
|
||
S3_ACCESS_KEY_ID_FILE: ""
|
||
S3_SECRET_ACCESS_KEY_FILE: ""
|
||
BROKER_URL_FILE: ""
|
||
OTEL_SDK_DISABLED: "true"
|
||
secrets: [postgres_password]
|
||
volumes: [*ca]
|
||
networks: [backend]
|
||
read_only: true
|
||
depends_on:
|
||
postgres: {condition: service_healthy}
|
||
healthcheck:
|
||
disable: true
|
||
restart: "no"
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.50", memory: 128M}
|
||
|
||
api:
|
||
<<: *api
|
||
build: *build
|
||
environment:
|
||
<<: *api-env
|
||
depends_on:
|
||
migrate: {condition: service_completed_successfully}
|
||
s3-init: {condition: service_completed_successfully}
|
||
postgres: {condition: service_healthy}
|
||
valkey: {condition: service_healthy}
|
||
broker: {condition: service_healthy}
|
||
openbao: {condition: service_healthy}
|
||
otlp-gw: {condition: service_healthy}
|
||
healthcheck: *api-healthcheck
|
||
restart: unless-stopped
|
||
stop_grace_period: 15s
|
||
deploy:
|
||
replicas: ${API_REPLICAS:-2}
|
||
resources:
|
||
# ARGON2_MAX_CONCURRENCY=2 x 19 МиБ на хеш + процесс ~150 МиБ: 384 МиБ с запасом.
|
||
limits: {cpus: "1.00", memory: 384M}
|
||
|
||
worker:
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
pull_policy: never
|
||
env_file: .env
|
||
command: ["python", "-m", "messenger.worker"]
|
||
environment:
|
||
<<: *app-env
|
||
DATABASE_URL: postgresql://messenger_worker@postgres:5432/messenger?sslmode=verify-full&sslrootcert=/run/ca/ca.crt
|
||
BAO_ROLE: messenger-worker
|
||
BAO_SECRET_PATH: secret/data/messenger/worker
|
||
CACHE_URL: "" # кэш сессий worker не нужен
|
||
OTEL_SERVICE_NAME: messenger-worker
|
||
read_only: true
|
||
# /tmp: файлы попытки обработки (оригинал до 10 МиБ + варианты) и alive-файл healthcheck.
|
||
tmpfs: ["/tmp:size=192m,mode=1777"]
|
||
volumes:
|
||
- ./secrets/tls/worker:/run/tls:ro
|
||
- *ca
|
||
networks: [backend, otlp, vault]
|
||
depends_on:
|
||
migrate: {condition: service_completed_successfully}
|
||
s3-init: {condition: service_completed_successfully}
|
||
postgres: {condition: service_healthy}
|
||
broker: {condition: service_healthy}
|
||
openbao: {condition: service_healthy}
|
||
otlp-gw: {condition: service_healthy}
|
||
healthcheck:
|
||
test: ["CMD", "python", "-c", "import os, sys, time; sys.exit(0 if time.time() - os.path.getmtime('/tmp/worker.alive') < 30 else 1)"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 3
|
||
start_period: 40s
|
||
start_interval: 1s
|
||
restart: unless-stopped
|
||
stop_grace_period: 30s
|
||
deploy:
|
||
replicas: ${WORKER_REPLICAS:-1}
|
||
resources:
|
||
# Лимит на upload-worker (README, уровень 4): CPU 1 ядро, RAM 1 ГиБ на контейнер; внутри —
|
||
# WORKER_CONCURRENCY=2 попытки, каждая — дочерний процесс с RLIMIT_AS 640 МиБ и JOB_TIMEOUT_SECONDS.
|
||
limits: {cpus: "1.00", memory: 1G}
|
||
|
||
api-short-ttl:
|
||
<<: *api
|
||
profiles: [ttl]
|
||
pull_policy: never
|
||
environment:
|
||
<<: *api-env
|
||
SESSION_TTL_SECONDS: "2"
|
||
depends_on:
|
||
postgres: {condition: service_healthy}
|
||
valkey: {condition: service_healthy}
|
||
openbao: {condition: service_healthy}
|
||
healthcheck: *api-healthcheck
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.50", memory: 256M}
|
||
|
||
# ---------- вход ----------
|
||
|
||
lb:
|
||
<<: *hardened
|
||
image: haproxy:3.2.25-alpine
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=8m"]
|
||
ports:
|
||
- "127.0.0.1:8443:8443" # единственная точка входа API, только HTTPS
|
||
- "127.0.0.1:8444:8444" # API с коротким TTL (make expiry), тоже HTTPS через балансировщик
|
||
- "127.0.0.1:8404:8404" # статистика HAProxy по HTTPS: https://127.0.0.1:8404/stats
|
||
volumes:
|
||
- ./infra/haproxy/haproxy.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
|
||
- ./secrets/tls/lb:/run/tls:ro
|
||
- *ca
|
||
networks: [public, edge, telemetry]
|
||
depends_on:
|
||
api: {condition: service_started}
|
||
healthcheck:
|
||
test: ["CMD", "wget", "-q", "-T", "2", "-O", "/dev/null", "http://127.0.0.1:8405/lb-health"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 3
|
||
start_period: 20s
|
||
start_interval: 1s
|
||
restart: unless-stopped
|
||
stop_grace_period: 15s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.50", memory: 128M}
|
||
|
||
# ---------- секреты ----------
|
||
|
||
# OpenBao не в dev-mode (infra/openbao/bao.hcl): persistent storage, sealed после каждого запуска.
|
||
# Healthy = процесс отвечает (в том числе sealed): unseal — действие оператора (make bao-unseal).
|
||
# Наружу не опубликован; оператор работает через контейнер ops в сети vault.
|
||
openbao:
|
||
<<: *hardened
|
||
image: openbao/openbao:2.4.4
|
||
user: "10001:10001"
|
||
entrypoint: ["bao", "server", "-config=/openbao/config/bao.hcl"]
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=8m"]
|
||
volumes:
|
||
- ./infra/openbao/bao.hcl:/openbao/config/bao.hcl:ro
|
||
- ./secrets/tls/openbao:/run/tls:ro
|
||
- *ca
|
||
- baodata:/openbao/data
|
||
- baologs:/openbao/logs
|
||
networks: [vault]
|
||
depends_on:
|
||
openbao-init: {condition: service_completed_successfully}
|
||
healthcheck:
|
||
# bao status: 0 — unsealed, 2 — sealed, 1 — ошибка. Healthy при 0 и 2.
|
||
test: ["CMD-SHELL", "BAO_ADDR=https://127.0.0.1:8200 BAO_CACERT=/run/ca/ca.crt bao status >/dev/null 2>&1; test $$? -ne 1"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 12
|
||
start_period: 10s
|
||
restart: unless-stopped
|
||
stop_grace_period: 15s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.50", memory: 256M}
|
||
|
||
# Новый volume принадлежит root, а OpenBao работает от uid 10001: одноразово отдаём каталоги.
|
||
openbao-init:
|
||
image: mtusi-messenger:lab7
|
||
pull_policy: never
|
||
user: "0:0"
|
||
entrypoint: ["chown", "-R", "10001:10001", "/openbao/data", "/openbao/logs"]
|
||
security_opt: ["no-new-privileges:true"]
|
||
cap_drop: [ALL]
|
||
cap_add: [CHOWN, DAC_OVERRIDE, FOWNER]
|
||
volumes: ["baodata:/openbao/data", "baologs:/openbao/logs"]
|
||
network_mode: none
|
||
restart: "no"
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.25", memory: 64M}
|
||
|
||
# ---------- хранилища ----------
|
||
|
||
# PostgreSQL: только TLS (pg_hba.conf), раздельные роли. Ключ сервера копируется в каталог postgres
|
||
# с mode 600 (PostgreSQL не принимает ключ, доступный группе/всем — а bind mount с Windows такой).
|
||
# Исключение hardening: entrypoint образа стартует от root и переключается на postgres (gosu), поэтому
|
||
# нужны CHOWN/DAC_OVERRIDE/FOWNER/SETUID/SETGID; корень не read-only (stateful, /var/run/postgresql).
|
||
postgres:
|
||
image: postgres:17.10-bookworm
|
||
security_opt: ["no-new-privileges:true"]
|
||
cap_drop: [ALL]
|
||
cap_add: [CHOWN, DAC_OVERRIDE, FOWNER, SETUID, SETGID]
|
||
environment:
|
||
POSTGRES_USER: messenger
|
||
POSTGRES_DB: messenger
|
||
POSTGRES_PASSWORD_FILE: /run/secrets/postgres_password
|
||
entrypoint:
|
||
- sh
|
||
- -c
|
||
- >-
|
||
install -m 0600 -o postgres -g postgres /run/tls/tls.key /var/lib/postgresql/server.key &&
|
||
install -m 0644 -o postgres -g postgres /run/tls/tls.crt /var/lib/postgresql/server.crt &&
|
||
exec docker-entrypoint.sh postgres
|
||
-c ssl=on -c ssl_cert_file=/var/lib/postgresql/server.crt -c ssl_key_file=/var/lib/postgresql/server.key
|
||
-c ssl_min_protocol_version=TLSv1.2 -c hba_file=/etc/postgresql/pg_hba.conf
|
||
-c password_encryption=scram-sha-256
|
||
-c fsync=on -c synchronous_commit=on -c full_page_writes=on
|
||
secrets: [postgres_password, db_api_password, db_worker_password]
|
||
volumes:
|
||
- pgdata:/var/lib/postgresql/data
|
||
- ./infra/postgres/pg_hba.conf:/etc/postgresql/pg_hba.conf:ro
|
||
- ./infra/postgres/initdb:/docker-entrypoint-initdb.d:ro
|
||
- ./secrets/tls/postgres:/run/tls:ro
|
||
- *ca
|
||
healthcheck:
|
||
test: ["CMD-SHELL", "pg_isready -h 127.0.0.1 -U messenger -d messenger"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 12
|
||
start_period: 10s
|
||
restart: unless-stopped
|
||
stop_grace_period: 30s
|
||
networks: [backend]
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.00", memory: 512M}
|
||
|
||
valkey:
|
||
<<: *hardened
|
||
image: valkey/valkey:8.1.9
|
||
user: valkey # не root: entrypoint образа ничего не меняет и не переключает
|
||
command: ["valkey-server", "/usr/local/etc/valkey/valkey.conf"]
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=16m"]
|
||
volumes:
|
||
- ./infra/valkey/valkey.conf:/usr/local/etc/valkey/valkey.conf:ro
|
||
- ./secrets/tls/valkey:/run/tls:ro
|
||
- *ca
|
||
secrets: [valkey_acl]
|
||
healthcheck:
|
||
test: ["CMD-SHELL", "valkey-cli --tls --cacert /run/ca/ca.crt -h 127.0.0.1 -p 6379 --user health --pass none --no-auth-warning ping | grep -q PONG"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 12
|
||
restart: unless-stopped
|
||
networks: [backend]
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.50", memory: 128M}
|
||
|
||
# SeaweedFS: S3 только по HTTPS (:9000, сертификат s3). master/volume/filer — на 127.0.0.1 контейнера.
|
||
# Root внутри контейнера (так устроен образ), но без capabilities и с read-only корнем.
|
||
s3:
|
||
<<: *hardened
|
||
image: chrislusf/seaweedfs:4.44
|
||
# Не root: /data в образе принадлежит seaweed (1000). Root с cap_drop ALL (без DAC_OVERRIDE)
|
||
# писать туда не может, а возвращать capability ради этого незачем.
|
||
user: "1000:1000"
|
||
entrypoint: ["/usr/bin/weed"]
|
||
command: ["server", "-dir=/data", "-ip=127.0.0.1", "-ip.bind=127.0.0.1",
|
||
"-master.volumeSizeLimitMB=1024", "-volume.max=0", "-master.telemetry=false",
|
||
"-s3", "-s3.ip.bind=0.0.0.0", "-s3.port=9000", "-s3.iam=false",
|
||
"-s3.config=/run/secrets/s3_config",
|
||
"-s3.cert.file=/run/tls/tls.crt", "-s3.key.file=/run/tls/tls.key"]
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=32m"]
|
||
secrets: [s3_config]
|
||
volumes:
|
||
- s3data:/data
|
||
- ./secrets/tls/s3:/run/tls:ro
|
||
- *ca
|
||
healthcheck:
|
||
test: ["CMD", "curl", "-fsS", "--cacert", "/run/ca/ca.crt", "-o", "/dev/null", "https://127.0.0.1:9000/healthz"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 12
|
||
start_period: 15s
|
||
restart: unless-stopped
|
||
stop_grace_period: 15s
|
||
networks: [backend]
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.50", memory: 512M}
|
||
|
||
s3-init:
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
build: *build
|
||
command: ["python", "-m", "messenger.s3bootstrap"]
|
||
environment:
|
||
S3_ENDPOINT: ${S3_ENDPOINT:-https://s3:9000}
|
||
S3_REGION: ${S3_REGION:-us-east-1}
|
||
S3_BUCKET: ${S3_BUCKET:-messenger-attachments}
|
||
S3_ADMIN_ACCESS_KEY_ID_FILE: /run/secrets/s3_root_user
|
||
S3_ADMIN_SECRET_ACCESS_KEY_FILE: /run/secrets/s3_root_password
|
||
TLS_CA_FILE: /run/ca/ca.crt
|
||
secrets: [s3_root_user, s3_root_password]
|
||
volumes: [*ca]
|
||
read_only: true
|
||
networks: [backend]
|
||
depends_on:
|
||
s3: {condition: service_healthy}
|
||
healthcheck:
|
||
disable: true
|
||
restart: "no"
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.50", memory: 128M}
|
||
|
||
# RabbitMQ: только AMQPS 5671 + SASL EXTERNAL, пользователи и топология — infra/rabbitmq/definitions.json.
|
||
# Узел один: кластерного канала нет. Порт Erlang distribution (25672, нужен CLI rabbitmq-diagnostics)
|
||
# виден только во внутренней сети backend и закрыт cookie из volume (REPORT.md, «Ограничения»).
|
||
# Исключение hardening: entrypoint образа меняет владельца volume и переключается на rabbitmq (su-exec).
|
||
broker:
|
||
image: rabbitmq:4.2.9-alpine
|
||
hostname: broker
|
||
security_opt: ["no-new-privileges:true"]
|
||
cap_drop: [ALL]
|
||
cap_add: [CHOWN, DAC_OVERRIDE, FOWNER, SETUID, SETGID]
|
||
environment:
|
||
RABBITMQ_CONFIG_FILES: /etc/rabbitmq/conf.d
|
||
volumes:
|
||
- rabbitmqdata:/var/lib/rabbitmq
|
||
- ./infra/rabbitmq/20-messenger.conf:/etc/rabbitmq/conf.d/20-messenger.conf:ro
|
||
- ./infra/rabbitmq/definitions.json:/etc/rabbitmq/definitions.json:ro
|
||
- ./infra/rabbitmq/enabled_plugins:/etc/rabbitmq/enabled_plugins:ro
|
||
- ./secrets/tls/broker:/run/tls:ro
|
||
- *ca
|
||
healthcheck:
|
||
test: ["CMD", "rabbitmq-diagnostics", "-q", "ping"]
|
||
interval: 10s
|
||
timeout: 10s
|
||
retries: 6
|
||
start_period: 60s
|
||
start_interval: 2s
|
||
restart: unless-stopped
|
||
stop_grace_period: 30s
|
||
networks: [backend]
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.00", memory: 512M}
|
||
|
||
# ---------- наблюдаемость ----------
|
||
|
||
# mTLS-шлюз перед Collector (infra/otel/gateway.cfg): identity api/worker, остальным — 403.
|
||
# Владелец network namespace: otel-collector подключается к нему (network_mode: service:otlp-gw),
|
||
# поэтому имя otel-collector в сетях otlp и telemetry — это адрес шлюза.
|
||
otlp-gw:
|
||
<<: *hardened
|
||
image: haproxy:3.2.25-alpine
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=8m"]
|
||
volumes:
|
||
- ./infra/otel/gateway.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
|
||
- ./secrets/tls/otel-collector:/run/tls:ro
|
||
- *ca
|
||
networks:
|
||
otlp: {aliases: [otel-collector]}
|
||
telemetry: {aliases: [otel-collector]}
|
||
healthcheck:
|
||
test: ["CMD", "wget", "-q", "-T", "2", "-O", "/dev/null", "http://127.0.0.1:8406/healthz"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 6
|
||
start_period: 10s
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.25", memory: 64M}
|
||
|
||
otel-collector:
|
||
<<: *hardened
|
||
image: otel/opentelemetry-collector-contrib:0.140.1
|
||
command: ["--config=/etc/otelcol-contrib/config.yaml"]
|
||
network_mode: "service:otlp-gw" # слушает 127.0.0.1:14318 общего со шлюзом namespace
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=16m"]
|
||
volumes:
|
||
- ./infra/otel/collector.yaml:/etc/otelcol-contrib/config.yaml:ro
|
||
- ./secrets/tls/otel-collector:/run/tls:ro
|
||
- *ca
|
||
depends_on:
|
||
otlp-gw: {condition: service_healthy}
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.50", memory: 512M}
|
||
|
||
prometheus:
|
||
<<: *hardened
|
||
image: prom/prometheus:v3.7.3
|
||
command:
|
||
- --config.file=/etc/prometheus/prometheus.yml
|
||
- --web.config.file=/etc/prometheus/web.yml
|
||
- --storage.tsdb.path=/prometheus
|
||
- --storage.tsdb.retention.time=15d
|
||
- --web.enable-otlp-receiver
|
||
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
|
||
- ./secrets/tls/prometheus:/run/tls:ro
|
||
- *ca
|
||
- promdata:/prometheus
|
||
networks: [telemetry]
|
||
healthcheck:
|
||
test: ["CMD", "promtool", "check", "ready", "--url=https://127.0.0.1:9090", "--http.config.file=/etc/prometheus/client.yml"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 12
|
||
start_period: 20s
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.00", memory: 512M}
|
||
|
||
# mTLS-шлюз перед Loki и Tempo (infra/tsdb/gateway.cfg): у их HTTP-серверов нет allowlist identity,
|
||
# поэтому её проверяет шлюз — loki:3100 пускает otel-collector и grafana, tempo:3200 — grafana,
|
||
# tempo:4318 — otel-collector; остальным identity нашего CA — 403. Владелец network namespace:
|
||
# loki и tempo подключаются к нему (network_mode: service:tsdb-gw) и слушают только 127.0.0.1.
|
||
tsdb-gw:
|
||
<<: *hardened
|
||
image: haproxy:3.2.25-alpine
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=8m"]
|
||
volumes:
|
||
- ./infra/tsdb/gateway.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
|
||
- ./secrets/tls/loki:/run/tls-loki:ro
|
||
- ./secrets/tls/tempo:/run/tls-tempo:ro
|
||
- *ca
|
||
networks:
|
||
telemetry: {aliases: [loki, tempo]}
|
||
healthcheck:
|
||
test: ["CMD", "wget", "-q", "-T", "2", "-O", "/dev/null", "http://127.0.0.1:8407/healthz"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 6
|
||
start_period: 10s
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.25", memory: 64M}
|
||
|
||
loki:
|
||
<<: *hardened
|
||
image: grafana/loki:3.5.8
|
||
command: ["-config.file=/etc/loki/loki.yaml"]
|
||
network_mode: "service:tsdb-gw" # слушает 127.0.0.1:13100 общего со шлюзом namespace
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=32m"]
|
||
volumes:
|
||
- ./infra/loki/loki.yaml:/etc/loki/loki.yaml:ro
|
||
- lokidata:/loki
|
||
depends_on:
|
||
tsdb-gw: {condition: service_healthy}
|
||
restart: unless-stopped
|
||
stop_grace_period: 30s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.00", memory: 512M}
|
||
|
||
tempo-init:
|
||
image: grafana/tempo:2.9.0
|
||
user: root
|
||
entrypoint: ["chown", "10001:10001", "/var/tempo"]
|
||
security_opt: ["no-new-privileges:true"]
|
||
cap_drop: [ALL]
|
||
cap_add: [CHOWN]
|
||
volumes: ["tempodata:/var/tempo"]
|
||
network_mode: none
|
||
restart: "no"
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "0.25", memory: 64M}
|
||
|
||
tempo:
|
||
<<: *hardened
|
||
image: grafana/tempo:2.9.0
|
||
command: ["-config.file=/etc/tempo/tempo.yaml"]
|
||
network_mode: "service:tsdb-gw" # слушает 127.0.0.1:13200 и :14318 общего со шлюзом namespace
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=32m"]
|
||
volumes:
|
||
- ./infra/tempo/tempo.yaml:/etc/tempo/tempo.yaml:ro
|
||
- tempodata:/var/tempo
|
||
depends_on:
|
||
tsdb-gw: {condition: service_healthy}
|
||
tempo-init: {condition: service_completed_successfully}
|
||
restart: unless-stopped
|
||
stop_grace_period: 30s
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.00", memory: 512M}
|
||
|
||
# Grafana: HTTPS для оператора (https://127.0.0.1:3000, admin + пароль из secrets/grafana_admin_password),
|
||
# к хранилищам — mTLS со своим сертификатом grafana (provisioning/datasources).
|
||
grafana:
|
||
<<: *hardened
|
||
image: grafana/grafana:12.2.1
|
||
environment:
|
||
GF_SERVER_PROTOCOL: https
|
||
GF_SERVER_CERT_FILE: /run/tls/tls.crt
|
||
GF_SERVER_CERT_KEY: /run/tls/tls.key
|
||
GF_SERVER_DOMAIN: localhost
|
||
GF_SECURITY_ADMIN_USER: admin
|
||
GF_SECURITY_ADMIN_PASSWORD__FILE: /run/secrets/grafana_admin_password
|
||
GF_SECURITY_COOKIE_SECURE: "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_LOG_MODE: console
|
||
secrets: [grafana_admin_password]
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=32m", "/var/log/grafana:size=8m"]
|
||
volumes:
|
||
- ./infra/grafana/provisioning:/etc/grafana/provisioning:ro
|
||
- ./infra/grafana/dashboards:/var/lib/grafana/dashboards:ro
|
||
- ./secrets/tls/grafana:/run/tls:ro
|
||
- *ca
|
||
- grafanadata:/var/lib/grafana
|
||
ports:
|
||
- "127.0.0.1:3000:3000"
|
||
networks: [telemetry, ops]
|
||
depends_on:
|
||
prometheus: {condition: service_healthy}
|
||
loki: {condition: service_started}
|
||
tempo: {condition: service_started}
|
||
healthcheck:
|
||
# HTTPS-порт слушается (0BB8 = 3000, состояние 0A = LISTEN). Проверка сертификата здесь не нужна:
|
||
# healthcheck не клиент, он только смотрит, что процесс принимает соединения.
|
||
test: ["CMD-SHELL", "cat /proc/net/tcp /proc/net/tcp6 2>/dev/null | grep -q ':0BB8 [0-9A-F:]* 0A '"]
|
||
interval: 5s
|
||
timeout: 3s
|
||
retries: 20
|
||
start_period: 30s
|
||
restart: unless-stopped
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.00", memory: 512M}
|
||
|
||
# ---------- проверки ----------
|
||
|
||
# Интеграционные тесты хранилища и заданий — ролью-владельцем (тестам нужна схема целиком).
|
||
db-test:
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
profiles: [tools]
|
||
pull_policy: never
|
||
env_file: .env
|
||
environment:
|
||
DATABASE_URL: postgresql://messenger@postgres:5432/messenger?sslmode=verify-full&sslrootcert=/run/ca/ca.crt
|
||
DATABASE_URL_FILE: ""
|
||
DB_PASSWORD_FILE: /run/secrets/postgres_password
|
||
CACHE_URL_FILE: ""
|
||
S3_ACCESS_KEY_ID_FILE: ""
|
||
S3_SECRET_ACCESS_KEY_FILE: ""
|
||
BROKER_URL_FILE: ""
|
||
OTEL_SDK_DISABLED: "true"
|
||
secrets: [postgres_password]
|
||
networks: [backend]
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=32m"]
|
||
volumes: ["./student-tests:/srv/student-tests:ro", *ca]
|
||
command: ["python", "-m", "unittest", "discover", "-s", "student-tests/db", "-v"]
|
||
depends_on:
|
||
postgres: {condition: service_healthy}
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.00", memory: 512M}
|
||
|
||
imaging-test:
|
||
<<: *hardened
|
||
image: mtusi-messenger:lab7
|
||
profiles: [tools]
|
||
pull_policy: never
|
||
network_mode: none
|
||
read_only: true
|
||
tmpfs: ["/tmp:size=256m"]
|
||
volumes: ["./student-tests:/srv/student-tests:ro", "./tests/fixtures:/srv/tests/fixtures:ro"]
|
||
command: ["python", "-m", "unittest", "discover", "-s", "student-tests/worker", "-v"]
|
||
deploy:
|
||
resources:
|
||
limits: {cpus: "1.00", memory: 1G}
|
||
|
||
volumes:
|
||
pgdata: {}
|
||
s3data: {}
|
||
rabbitmqdata: {}
|
||
promdata: {}
|
||
lokidata: {}
|
||
tempodata: {}
|
||
grafanadata: {}
|
||
baodata: {} # с lab7: зашифрованное хранилище OpenBao
|
||
baologs: {} # с lab7: audit log OpenBao
|
||
|
||
secrets:
|
||
postgres_password: {file: ./secrets/postgres_password}
|
||
db_api_password: {file: ./secrets/db_api_password}
|
||
db_worker_password: {file: ./secrets/db_worker_password}
|
||
valkey_acl: {file: ./secrets/valkey_users.acl}
|
||
s3_config: {file: ./secrets/s3_config.json}
|
||
s3_root_user: {file: ./secrets/s3_root_user}
|
||
s3_root_password: {file: ./secrets/s3_root_password}
|
||
grafana_admin_password: {file: ./secrets/grafana_admin_password}
|
||
|
||
networks:
|
||
public: {}
|
||
edge: {internal: true}
|
||
backend: {internal: true}
|
||
otlp: {internal: true}
|
||
telemetry: {internal: true}
|
||
vault: {internal: true}
|
||
ops: {}
|