sgl-project/sglang · critical · ValueError

Kimi expert-pack header is truncated

Error message

Kimi expert-pack header is truncated

What it means

The first GGML_PACK_HEADER.size bytes read from the pack file are shorter than the fixed header struct, i.e. the file is smaller than a valid header. The pack is unusable — almost certainly truncated or empty.

Source

Thrown at python/sglang/srt/layers/moe/expert_pack.py:800

        self.entries: dict[tuple[int, int, int], ExpertPackEntry] = {}
        self.object_offsets: dict[tuple[int, int], int] = {}
        object_payload_bytes = int(pack_manifest["object_bytes"])
        previous_end = int(pack_manifest["data_start"])
        role_offsets: dict[str, int] = {}
        role_nbytes = {
            role: int(roles[role]["expert_bytes"]) for role in KIMI_PHYSICAL_ROLES
        }
        running_role_offset = 0
        for role in KIMI_PHYSICAL_ROLES:
            role_offsets[role] = running_role_offset
            running_role_offset += role_nbytes[role]
        if running_role_offset != object_payload_bytes:
            raise ValueError("Kimi expert-pack role sizes do not match object bytes")

        with self.path.open("rb", buffering=0) as stream:
            raw_header = stream.read(GGML_PACK_HEADER.size)
            if len(raw_header) != GGML_PACK_HEADER.size:
                raise ValueError("Kimi expert-pack header is truncated")
            index_digest.update(raw_header)
            magic, version, header_size, index_count, data_start = (
                GGML_PACK_HEADER.unpack(raw_header)
            )
            if (
                magic != GGML_PACK_MAGIC
                or version != 1
                or header_size != GGML_PACK_HEADER.size
                or index_count != expected_entry_count
                or data_start != int(pack_manifest["data_start"])
            ):
                raise ValueError("Kimi expert-pack header does not match its manifest")

            for index in range(index_count):
                raw_entry = stream.read(GGML_PACK_ENTRY.size)
                if len(raw_entry) != GGML_PACK_ENTRY.size:
                    raise ValueError("Kimi expert-pack index is truncated")
                index_digest.update(raw_entry)

View on GitHub (pinned to 0132848349)

Solutions

1. Verify the file path points at the real pack
2. Re-download/recreate the pack and check size against the manifest before loading

Example fix

// before
pack = KimiExpertPack(Path('/data/model.pack.tmp'))
// after
assert p.stat().st_size == manifest['expert_pack']['size']
pack = KimiExpertPack(p)

import struct
HEADER_SIZE = struct.calcsize('<5Q')  # match GGML_PACK_HEADER.size
if pack_path.stat().st_size < HEADER_SIZE:
    raise RuntimeError('pack file too small to contain a header')

When it happens

Trigger: Opening a 0-byte or few-byte file (failed download, interrupted write, wrong path pointing at a stub file).

Common situations: See trigger scenarios.

Related errors

• expert-pack index is truncated
• Kimi active routed MoE layers must be exactly 1..92
• Kimi manifest expert-pack path does not match pack_path
• Kimi expert-pack size does not match its manifest
• Kimi expert-pack physical role order is unsupported

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/bdf31bcc008d4691. Report an issue: GitHub.