Skip to content

Upstream SGLang fix for disaggregated prefill TP-rank desync / 向 SGLang 上游提交分离式预填充 TP rank 不同步修复 #2151

Description

@cquil11

Track an upstream SGLang fix for the local patch added in #2127.

resolve_waiting_queue_bootstrap() can build different candidate sets across TP ranks, causing some ranks to enter the collective while others skip it. The local workaround limits candidates to non-aborted pending_bootstrap requests.

Reference: #2127 (comment)

中文说明

跟踪 #2127 中本地补丁对应的 SGLang 上游修复。resolve_waiting_queue_bootstrap() 可能在不同 TP rank 上生成不一致的候选集合,导致 collective 不同步。当前临时方案仅保留未中止的 pending_bootstrap 请求。

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    Status
    No status

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions