Hosted Server FPS Loop Sleep

Hosted Server FPS Loop Sleep

Status

Checked 2026-06-24 after fetching origin and Miksuu master. Repo docs branch HEAD@4a1bc13edde6 is source-unchanged from d0161083 / a27086cd for the checked hosted-FPS paths and keeps the branch-local two-publisher DR-19 fix in Chernarus and maintained Vanilla: both serverFpsGUI.sqf:1 and monitorServerFPS.sqf:1 exit before their loops on !isDedicated. Current stable/B74.1 origin/master@f8a76de34 / origin/claude/b74.1-aicom@f8a76de34, current B74.2 origin/claude/b74.2-aicom@21b62b04, B69 8d465fce, B74 b23f557f, live origin/feat/client-fps@709258e7, live origin/perf/asr-ai-server-fps@07609ef4 and historical a96fdda2 use a different fixed shape: one guarded serverFpsGUI.sqf publisher and no maintained-root Server/Module/serverFPS/monitorServerFPS.sqf file. Miksuu master@b8389e748243 and origin/perf/quick-wins@0076040f still keep the old two unguarded publisher loops in checked maintained roots. Current origin exposes no release/* heads, and no origin/dev/july-update-hosted-server-fps-loop-fix head was found on 2026-06-24. Arma 2 OA smoke is still pending.

Current Branch Scope

Ref / root scope	FPS publisher shape	Development meaning
Repo docs branch HEAD@4a1bc13edde6 Chernarus and maintained Vanilla	Source unchanged from d0161083 / a27086cd: Server/GUI/serverFpsGUI.sqf:1 and Server/Module/serverFPS/monitorServerFPS.sqf:1 both exit on !isDedicated; Init_Server.sqf:578,595 still starts both scripts. serverFpsGUI.sqf:6-7 publishes SERVER_FPS_GUI; monitorServerFPS.sqf:5-6 publishes WFBE_VAR_SERVER_FPS; Client_UpdateRHUD.sqf:113 reads SERVER_FPS_GUI.	Branch-local source carries the low-risk two-publisher guard. Smoke dedicated publish plus hosted/listen no-spin before release-ready wording.
Current stable/B74.1 origin/master@f8a76de34 / origin/claude/b74.1-aicom@f8a76de34, B69 8d465fce and B74 b23f557f	Both maintained roots guard serverFpsGUI.sqf at :4, publish SERVER_FPS_GUI at :9-10, and Client_UpdateRHUD.sqf:113,130 reads that variable; current stable starts it from Chernarus Init_Server.sqf:950 / Vanilla :937, while B69/B74 start it at :937 in both roots. The removed-monitor note is current stable Chernarus :996-998 / Vanilla :983-985; B69/B74 use :983-985; monitorServerFPS.sqf is absent.	Stable/B74-shaped refs are no longer old-loop-shaped. Smoke the single-publisher target branch rather than porting the docs-branch two-publisher shape blindly. The planned July dev branch must be created or restored before code work because no current remote head was found.
Current B74.2 origin/claude/b74.2-aicom@21b62b04	Same single-publisher FPS shape as current stable in both roots. Source Chernarus line-drifts the server start/removal anchors to Init_Server.sqf:983 and :1029-1031; maintained Vanilla stays stable-shaped at :937 and :983-985; serverFpsGUI.sqf:4,9-10 and Client_UpdateRHUD.sqf:113,130 are unchanged for the FPS contract.	B74.2 does not reopen the hosted-FPS loop. Its checked deltas are unrelated source-Chernarus AICOM/GUER/RHUD cap context plus line drift; treat maintained Vanilla as stable-shaped for this lane.
Live FPS-named branches origin/feat/client-fps@709258e7 and origin/perf/asr-ai-server-fps@07609ef4	Both maintained roots keep the single guarded serverFpsGUI.sqf:4,9-10 shape and no maintained-root monitor file. feat/client-fps starts the publisher at Chernarus Init_Server.sqf:719 / Vanilla :612; perf/asr-ai-server-fps starts it at :579 in both roots.	Do not infer old-loop status from the branch names. They are single-publisher-shaped and still need target-branch smoke before release wording.
Historical release commit a96fdda2	Both maintained roots guard serverFpsGUI.sqf at :4, publish SERVER_FPS_GUI at :9-10, and Client_UpdateRHUD.sqf:113,130 reads it; Init_Server.sqf:579 starts it, :595-597 documents monitor removal, and monitorServerFPS.sqf is absent.	Historical release evidence matches stable for this runtime surface, but current origin exposes no live release/* head. Smoke the target branch before release-complete claims.
Miksuu master@b8389e748243 and origin/perf/quick-wins@0076040f	Both checked roots keep serverFpsGUI.sqf:4 and monitorServerFPS.sqf:2 guarding only the body inside while {true}; the only sleep remains inside that branch. Miksuu starts both scripts at Init_Server.sqf:578,595; perf starts them at :573,590 in Chernarus and :578,595 in maintained Vanilla.	These are the remaining old-loop targets. Port either the docs two-publisher guard or the stable single-publisher cleanup deliberately, then propagate maintained Vanilla and smoke.

What I Read

• Missions/[55-2hc]warfarev2_073v48co.chernarus/Server/GUI/serverFpsGUI.sqf
• Missions/[55-2hc]warfarev2_073v48co.chernarus/Server/Module/serverFPS/monitorServerFPS.sqf
• Missions/[55-2hc]warfarev2_073v48co.chernarus/Server/Init/Init_Server.sqf
• Missions/[55-2hc]warfarev2_073v48co.chernarus/Client/Client_UpdateRHUD.sqf
• Source/Vanilla/modded search results for SERVER_FPS_GUI and WFBE_VAR_SERVER_FPS

Original Bug Shape

Before the DR-19 patch, Init_Server.sqf started two FPS publisher scripts:

Source	Behavior
Old-shape init, still visible on Miksuu/perf	Starts Server\GUI\serverFpsGUI.sqf and Server\Module\serverFPS\monitorServerFPS.sqf.
Old-shape Server/GUI/serverFpsGUI.sqf:1-12	Runs while {true}; publishes SERVER_FPS_GUI only inside if (isDedicated); sleeps only inside that same branch.
Old-shape Server/Module/serverFPS/monitorServerFPS.sqf:1-8	Runs while {true}; publishes WFBE_VAR_SERVER_FPS only inside if (isDedicated); sleeps only inside that same branch.
Client/Client_UpdateRHUD.sqf:113-125	Source/Vanilla RHUD reads SERVER_FPS_GUI.

That means dedicated servers published every 8 seconds, but hosted/listen servers entered both loops and skipped the only sleep. The result was two unslept scheduled loops doing no useful work.

Patch Shape

The patch keeps dedicated behavior and removes the hosted/listen busy-spin:

• Add if (!isDedicated) exitWith {}; before each while {true}.
• Remove the inner if (isDedicated) branch.
• Leave SERVER_FPS_GUI and WFBE_VAR_SERVER_FPS publishing cadence unchanged on dedicated servers.
• Do not consolidate the two publishers yet; source/Vanilla only consume SERVER_FPS_GUI, but stale/modded mission folders still contain WFBE_VAR_SERVER_FPS consumers.

Changed branch-local docs/source files (docs/developer-wiki-index; stable/release use the one-publisher cleanup instead):

• Missions/[55-2hc]warfarev2_073v48co.chernarus/Server/GUI/serverFpsGUI.sqf
• Missions/[55-2hc]warfarev2_073v48co.chernarus/Server/Module/serverFPS/monitorServerFPS.sqf

Propagated maintained Vanilla files:

• Missions_Vanilla/[61-2hc]warfarev2_073v48co.takistan/Server/GUI/serverFpsGUI.sqf
• Missions_Vanilla/[61-2hc]warfarev2_073v48co.takistan/Server/Module/serverFPS/monitorServerFPS.sqf

Validation

Branch-local source/Vanilla validation done:

• This docs branch's Chernarus serverFpsGUI.sqf has one early !isDedicated exit, one loop, one sleep 8 and no inner if (isDedicated) branch.
• This docs branch's Chernarus monitorServerFPS.sqf has one early !isDedicated exit, one loop, one sleep 8 and no inner if (isDedicated) branch.
• Current stable/B74.1 origin/master@f8a76de34, current B74.2 21b62b04, B69 8d465fce, B74 b23f557f, live feat/client-fps@709258e7, live perf/asr-ai-server-fps@07609ef4 and historical a96fdda2 have the guarded single serverFpsGUI.sqf shape at :4-12 and lack the maintained-root Server/Module/serverFPS/monitorServerFPS.sqf file. Init anchors drift by branch: stable/B74.1 Chernarus :950,:996-998, stable/B74.1 Vanilla :937,:983-985, B74.2 Chernarus :983,:1029-1031, B74.2 Vanilla :937,:983-985, client-fps Chernarus :719,:765-767, client-fps Vanilla :612,:628-630, ASR FPS and historical release :579,:595-597.
• Miksuu b8389e74 and origin/perf/quick-wins 0076040f remain old-loop-shaped in checked maintained roots.
• Vanilla Takistan has the same early-exit shape after the propagation run.
• dotnet run in Tools/LoadoutManager now works from work\a; use A2WASP_SKIP_ZIP=1 for propagation-only runs so missing 7za remains non-blocking.
• git diff --check passes.

Pending smoke:

• Dedicated: confirm RHUD/server FPS updates roughly every 8 seconds.
• Dedicated: confirm no RPT errors from either FPS publisher.
• Hosted/listen: confirm no scheduled busy-spin from these scripts and HUD behavior remains acceptable when server FPS is not published.

Handoff

For Codex/future code owner:

• Treat this as the low-risk DR-19 implementation. It intentionally does not remove monitorServerFPS.sqf or merge WFBE_VAR_SERVER_FPS into SERVER_FPS_GUI.
• If cleaning docs/source further, decide whether to keep the two-publisher compatibility path or adopt the stable/release single-publisher cleanup. Check stale/modded WFBE_VAR_SERVER_FPS consumers before consolidating publishers.
• Modded mission folders still need their broader generated/forked maintenance model resolved before hand edits.

Wave P contract note: in current source Chernarus, RHUD/FPS HUD reads SERVER_FPS_GUI (Client/Client_UpdateRHUD.sqf:113), while WFBE_VAR_SERVER_FPS is published by Server/Module/serverFPS/monitorServerFPS.sqf:5-6 with no player-UI reader found in the source mission. Consolidation is therefore a separate compatibility cleanup, not part of the hosted/listen busy-loop fix.

For Claude:

• Good contradiction check: inspect hosted/listen behavior in Arma 2 OA after this patch and verify no other Init_Server.sqf loop has the same while true + branch-only sleep shape.

Continue Reading

• Previous: Performance opportunity sweep
• Next: AI, headless and performance