lab4: вложения в приватном S3 (SeaweedFS), метаданные в PostgreSQL

- compose: сервис s3 (SeaweedFS 4.46 по digest, volume s3data, внутренняя сеть storage,
  web-интерфейсы только на 127.0.0.1 внутри контейнера), одноразовый s3-init;
- три пары ключей S3: admin (только bootstrap), api (Read+Write своего bucket),
  reconciler (+List); bootstrap идемпотентен и проверяет приватность и права;
- миграция 0004: attachments (pending -> ready, случайный object_key), message_attachments,
  post_message_v2 с проверкой вложений в той же транзакции;
- POST /chats/{id}/attachments: лимиты до чтения/декодирования, потоковая проверка PNG
  (CRC + zlib до точного размера), JPEG через djpeg -strict, семафор загрузок;
- GET метаданных и /content (стриминг из S3, без redirect), readiness с S3,
  HAProxy: деградированный режим без S3 через /health/ready/core;
- janitor незавершённых загрузок с grace period.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Alexey Bodanyuk
2026-10-25 13:26:01 +03:00
co-authored by Claude Opus 5.5
parent d916d09f7b
commit 335df4b2b2
19 changed files with 2586 additions and 69 deletions
+6 -1
View File
@@ -24,7 +24,7 @@ def write(name: str, value: str) -> None:
print(f"keep secrets/{name}")
return
path.write_text(value + "\n", encoding="utf-8", newline="\n")
# Readable by the container users (postgres, valkey, app uid 10001) through the
# Readable by the container users (postgres, valkey, seaweedfs, app uid 10001) through the
# compose bind mount. Local synthetic lab credentials only.
os.chmod(path, 0o644)
print(f"created secrets/{name}")
@@ -40,6 +40,11 @@ def main() -> int:
write("valkey_password", secrets.token_urlsafe(24))
write("database_url", f"postgresql://messenger:{read('postgres_password')}@postgres:5432/messenger")
write("cache_url", f"redis://:{read('valkey_password')}@valkey:6379/0")
# Lab 4: three separate S3 key pairs (SeaweedFS identities, infra/seaweedfs/start.sh):
# admin = bucket bootstrap only; api = Read+Write on the bucket; reconciler = + List.
for who in ("admin", "api", "reconciler"):
write(f"s3_{who}_access_key_id", "AK" + secrets.token_hex(9).upper())
write(f"s3_{who}_secret_access_key", secrets.token_hex(20))
return 0
+5 -4
View File
@@ -1,6 +1,6 @@
#!/bin/sh
# Restart hook for `make test-persistence ACTION_SCRIPT=scripts/restart.sh`.
# Restarts EVERY process of the stand: LB, all API replicas, PostgreSQL and Valkey.
# Restarts EVERY process of the stand: LB, all API replicas, PostgreSQL, Valkey and S3 (SeaweedFS).
# Volumes, schema and data are preserved: no `down -v`, no FLUSH, no re-seed.
# Output is suppressed by the test; run it by hand to see the steps.
set -eu
@@ -8,10 +8,11 @@ cd "$(dirname "$0")/.."
# 1. Entry point and API first: no request is served while the stores restart underneath.
docker compose stop lb api
# 2. Stores: clean shutdown (PostgreSQL checkpoint, Valkey AOF fsync), then start again.
docker compose restart postgres valkey
# 2. Stores: clean shutdown (PostgreSQL checkpoint, Valkey AOF fsync, SeaweedFS volume/filer
# flush), then start again. The S3 bytes live on the named volume s3data.
docker compose restart postgres valkey s3
# 3. API replicas (wait for healthy stores via depends_on), then the LB. HAProxy resolves
# the new replica addresses through Docker DNS by itself - no stale backend list.
docker compose start api
docker compose start lb
echo "restart.sh: lb, api (all replicas), postgres and valkey restarted with volumes preserved"
echo "restart.sh: lb, api (all replicas), postgres, valkey and s3 restarted with volumes preserved"