上传知识库文档后，数据处理报错，Embedding 配置总是无法成功

code=105000004 message=SearchStore operation failed: store search store failed, err: [slices2Columns] embed failed, [Ark] CreateEmbeddings error: RequestError code: 400, err: can't fetch endpoint sts token without endpoint, request_id: 2026030501415500006DDEBD1014A9A123

{
"code": 400,
"msg": "embed test string failed: error, status code: 400, status: 400 Bad Request, message: %!s(\u003cnil\u003e), body: {\"object\":\"error\",\"message\":\"Model \\\"Qwen3-Embedding-8B\\\" does not support matryoshka representation, changing output dimensions will lead to poor results.\",\"type\":\"BadRequestError\",\"param\":null,\"code\":400}"
}

No source file, test, or entry point is identified in the issue. Start by reproducing the failure through knowledge-base document upload and the embedding configuration test, then inspect the reported endpoint-token and Qwen3-Embedding-8B errors. Done means document processing and embedding configuration succeed without either reported 400 error.

Written by the indexing model from the issue text.