Issue Station
Sign In
Tools hereToolbox

vllm - 💡(How to fix) Fix [Bug]: Fix two crashes that prevent DeepSeek-V4-Flash + OffloadingConnector [1 comments, 1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
vllm-project/vllm#43093•Fetched 2026-05-20 03:39:53
View on GitHub
Comments
1
Participants
1
Timeline
2
Reactions
0
Author
panpan0000panpan0000
Participants
panpan0000panpan0000
Timeline (top)
commented ×1labeled ×1

Error Message

Error 1 — stale block_size causes divisibility check to fail

AssertionError: gpu_block_size=64 not divisible by hash_block_size=256.

Code Example

# Error 1 — stale block_size causes divisibility check to fail
AssertionError: gpu_block_size=64 not divisible by hash_block_size=256.

---

# Error 2 — only visible after fixing Error 1
AssertionError: If 'block_size' is specified in kv_connector_extra_config,
there must be at least one KV cache group, and all groups must have the same block size.

---

vllm serve deepseek-ai/DeepSeek-V4-Flash \
  --block-size 256 \
  --kv-cache-dtype fp8 \
  --no-disable-hybrid-kv-cache-manager \
  --kv-transfer-config '{\"kv_connector\":\"OffloadingConnector\",\"kv_role\":\"kv_both\",\"kv_connector_extra_config\":{\"block_size\":256}}'
RAW_BUFFERClick to expand / collapse

Your current environment

H200 + v0.21.0 + deepseek-v4-flash

🐛 Describe the bug

[kv_offload] Fix stale cache_config.block_size used as hash_block_size in OffloadingSpec

Running DeepSeek-V4-Flash with KV offloading enabled causes the engine to crash immediately at startup with one of these assertion errors:

# Error 1 — stale block_size causes divisibility check to fail
AssertionError: gpu_block_size=64 not divisible by hash_block_size=256.
# Error 2 — only visible after fixing Error 1
AssertionError: If 'block_size' is specified in kv_connector_extra_config,
there must be at least one KV cache group, and all groups must have the same block size.

A typical failing command looks like:

vllm serve deepseek-ai/DeepSeek-V4-Flash \
  --block-size 256 \
  --kv-cache-dtype fp8 \
  --no-disable-hybrid-kv-cache-manager \
  --kv-transfer-config '{\"kv_connector\":\"OffloadingConnector\",\"kv_role\":\"kv_both\",\"kv_connector_extra_config\":{\"block_size\":256}}'

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#logging issue#authentication issue#prompt issue#agent setup#task chaining

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

vllm - 💡(How to fix) Fix [Bug]: Fix two crashes that prevent DeepSeek-V4-Flash + OffloadingConnector [1 comments, 1 participants]