EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[PD] Add guidance for prefill bootstrap timeout (#7846)

Browse Source 
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
This commit is contained in:
Shangming Cai
2025-07-09 12:00:34 +08:00
committed by GitHub
parent 128f16a817
commit 64c5907e12
2 changed files with 42 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
docs/backend/pd_disaggregation.md
View File

@@ -56,13 +56,19 @@ PD Disaggregation with Mooncake supports the following environment variables for
|:--------:|:-----------:|:--------:
| **`SGLANG_DISAGGREGATION_THREAD_POOL_SIZE`** | Controls the total number of worker threads for KVCache transfer operations per TP rank | A dynamic value calculated by `int(0.75 * os.cpu_count()) // 8)`, which is limited to be larger than 4 and less than 12 to ensure efficiency and prevent thread race conditions |
| **`SGLANG_DISAGGREGATION_QUEUE_SIZE`** | Sets the number of parallel transfer queues. KVCache transfer requests from multiple decode instances will be sharded into these queues so that they can share the threads and the transfer bandwidth at the same time. If it is set to `1`, then we transfer requests one by one according to fcfs strategy | `4` |
| **`SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT`** | Timeout (seconds) for receiving destination KV indices during request initialization | `120` |
| **`SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT`** | Timeout (seconds) for receiving destination KV indices during request initialization | `300` |
If a greater mean TTFT is acceptable, you can `export SGLANG_DISAGGREGATION_BOOTSTRAP_TIMEOUT=600` (10 minutes) to relax the timeout condition.
Please be aware that this setting will cause prefill instances to take a longer time to clean up the affected memory resources when a running decode node loses connection.
#### Decode Server Configuration
| Variable | Description | Default |
|:--------:|:-----------:|:--------:
| **`SGLANG_DISAGGREGATION_HEARTBEAT_INTERVAL`** | Interval (seconds) between health checks to prefill bootstrap servers | `5.0` |
| **`SGLANG_DISAGGREGATION_HEARTBEAT_MAX_FAILURE`** | Consecutive heartbeat failures before marking prefill server offline | `2` |
| **`SGLANG_DISAGGREGATION_WAITING_TIMEOUT`** | Timeout (seconds) for receiving KV Cache after request initialization | `300` |
If a greater mean TTFT is acceptable, you can `export SGLANG_DISAGGREGATION_WAITING_TIMEOUT=600` (10 minutes) to relax the timeout condition.
## NIXL