Isolate cloud model access before enabling product RAG workflows
Some checks failed
verify / verify (push) Has been cancelled

The API and ingestion tools now use a fixed internal model gateway while
governed profiles, embedding cache assignments, traceable citations, and
stable API errors establish the boundaries required by later workflows.

Constraint: The current Alibaba Cloud workspace rejects all three live model calls with authentication failures
Rejected: Give the API or seed tools the Bailian key and direct egress | combines database access, cloud credentials, and public network access
Rejected: Mix offline and Bailian vectors in one demo namespace | makes profile activation and retrieval ambiguous
Confidence: high
Scope-risk: moderate
Reversibility: clean
Directive: Keep Bailian credentials and egress exclusive to model-gateway and create a new immutable profile hash for any embedding identity change
Tested: make verify; 121 backend tests; 14 frontend tests; fresh and populated Alembic upgrade-downgrade-upgrade; two idempotent offline seeds; Docker health and HTTP retrieval; isolated provider smoke
Not-tested: Successful live Bailian responses because the supplied workspace credential currently fails authentication
This commit is contained in:
2026-07-13 04:09:06 +08:00
parent 99b7df64ea
commit 75592af33a
28 changed files with 3932 additions and 254 deletions

View File

@@ -443,7 +443,7 @@ embedding_cache_key = SHA256(
chunk_embedding_assignment = UNIQUE(chunk_id, embedding_profile_hash)
```
相同文本可以安全复用向量缓存,但每个 chunk 都必须建立独立 assignment不能用缓存键替代任务/行唯一键。模型别名无法解析到稳定 revision 时,每次冻结实验显式设置 `cache_epoch`避免别名漂移后复用旧向量。激活前再次核对当前 `embedding_text_sha256`、完整 profile 与 assignment任何一项变化都生成新索引版本。
相同文本可以安全复用向量缓存,但每个 chunk 都必须建立独立 assignment不能用缓存键替代任务/行唯一键。模型别名无法解析到稳定 revision 时,每次冻结实验显式设置 `cache_epoch`并把它纳入不可变的 `embedding_profile_hash`epoch 变化必须创建新 profile禁止原地修改 profile 后继续复用旧缓存。激活前再次核对当前 `embedding_text_sha256`、完整 profile 与 assignment任何一项变化都生成新索引版本。
`endpoint_identity_hash` 由规范化后的工作空间/部署端点身份计算,仅保存哈希而不在实验导出中暴露真实工作空间。这样同地域、同模型别名但不同 MaaS 工作空间的向量不会错误共用缓存。