P2 infra: add Loki + Promtail for centralized logs

[SlavaKuntsov]

Phase / Priority / Area

Phase 14 · Priority P2 · Area infra

Зависимости

Blocked by: #15 (unified JSON log с trace_id)

Scope

deploy/docker-compose.yml:1, deploy/loki/loki.yml (new), deploy/promtail/promtail.yml (new), deploy/grafana/provisioning/datasources/datasource.yml (Loki datasource), .env.example (LOKI_PORT)

Goal

Централизованный сбор логов вместо docker logs | grep. В Grafana Explore → Loki запрос {service="gateway"} |= "test123" находит все логи с trace_id из всех сервисов.

Acceptance Criteria

• Сервисы loki:3100, promtail в compose, Loki datasource в Grafana provisioned
• curl -H X-Request-ID:test123 /api/v1/videos → в Grafana Explore Loki запрос {service=~".*"} |= "test123" возвращает логи gateway/metadata/upload/auth
• JSON логи с trace_id/request_id сохраняются (зависит от P2 infra: enforce unified JSON log with trace_id #15)

Проверки

curl http://localhost:3100/ready, curl -G http://localhost:3100/loki/api/v1/query --data-urlencode 'query={service="gateway"}', make up && curl -H X-Request-ID:test123 ... → Loki UI

Out of scope

Долгое хранение / S3 для Loki (бэклог)

[SlavaKuntsov]

Pull request opened

#39 — will auto-close via Closes #38 on merge.

Verification

• make lint-go — 0 issues
• make lint-py — success
• make lint-front — ok
• make test-go/test-py — pass
• make up --build -d — all Healthy (prometheus, loki, grafana)
• make e2e — PASS
• curl -H X-Request-ID:test123 → logs contain trace_id; Loki query {service="gateway"} |= "test123" — success

Remaining risks

• Loki allow_structured_metadata false (v11), Promtail docker.sock, rabbitmq_queue_depth Gauge 0 — см. PR Risks