[PD] Prevent decode scheduler from blocking on ZMQ sends to a stalled prefill peer (#31144)

Co-authored-by: Shangming Cai <csmthu@gmail.com>
This commit is contained in:
SovietPower
2026-07-25 12:53:41 +08:00
committed by GitHub
co-authored by Shangming Cai
parent ebcb74abd4
commit 6a046fad09
6 changed files with 188 additions and 116 deletions
@@ -1428,6 +1428,11 @@ SGLang supports various environment variables that can be used to configure its
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>Consecutive heartbeat failures tolerated before a peer is considered dead.</td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}><code>2</code></td>
</tr>
<tr>
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><code>SGLANG_DISAGGREGATION_ZMQ_SEND_TIMEOUT</code></td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>Send timeout (seconds) for decode's ZMQ sockets to prefill peers. Raise if healthy sends exceed the default.</td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}><code>1</code></td>
</tr>
<tr>
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><code>SGLANG_DISAGGREGATION_BOOTSTRAP_ENTRY_CLEANUP_INTERVAL</code></td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>Interval (seconds) for cleaning up stale bootstrap entries.</td>