From 089c6eb5c02b9dc03479fce3893ba95e566a3b4f Mon Sep 17 00:00:00 2001
From: Joseph Magly <1159087+jmagly@users.noreply.github.com>
Date: Fri, 28 Aug 2026 10:46:24 -0400
Subject: [PATCH] docs: clarify shared-host admission budget

---
 docs/deployment/shared-gpu-host.md | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/docs/deployment/shared-gpu-host.md b/docs/deployment/shared-gpu-host.md
index 63f740d..0202db2 100644
--- a/docs/deployment/shared-gpu-host.md
+++ b/docs/deployment/shared-gpu-host.md
@@ -95,7 +95,10 @@ new admission.
 Readiness occurs after model allocation, never at UI startup. Set
 `OBLITERATUS_GPU_HEARTBEAT_SECONDS` to change the default 15-second heartbeat,
 and `OBLITERATUS_GPU_ADMISSION_TIMEOUT_SECONDS` to change the default 30-second
-fail-closed acknowledgment timeout.
+fail-closed acknowledgment timeout. Shared hosts that must drain another model
+server should use a larger measured deadline (120 seconds in the example below)
+and require the supervisor's broker timeout plus heartbeat handshake to finish
+several seconds before it.
 
 A minimal systemd setup lets the supervisor create the local boundary without
 giving the application scheduling authority:
@@ -108,7 +111,7 @@ RuntimeDirectory=obliteratus-gpu-lifecycle
 RuntimeDirectoryMode=0770
 Environment=OBLITERATUS_GPU_LIFECYCLE_DIR=/run/obliteratus-gpu-lifecycle
 Environment=OBLITERATUS_GPU_HEARTBEAT_SECONDS=15
-Environment=OBLITERATUS_GPU_ADMISSION_TIMEOUT_SECONDS=30
+Environment=OBLITERATUS_GPU_ADMISSION_TIMEOUT_SECONDS=120
 ExecStart=/srv/obliteratus/current/.venv/bin/obliteratus ui --host 127.0.0.1
 ```