Skip to content

History / Headless Client Scaling And Topology

Revisions

  • docs: sync release support page hashes

    rayswaynl committed Jul 1, 2026
  • docs: record runtime summary root binding gate

    @rayswaynl rayswaynl committed Jul 1, 2026
  • wiki: record runtime packet manifest gate proof

    @rayswaynl rayswaynl committed Jul 1, 2026
  • wiki: record runtime identity handoff guard

    @rayswaynl rayswaynl committed Jul 1, 2026
  • wiki: record summary release identity proof

    @rayswaynl rayswaynl committed Jul 1, 2026
  • wiki: record summary packet proof gate

    @rayswaynl rayswaynl committed Jul 1, 2026
  • wiki: record takistan fallback runtime gate

    @rayswaynl rayswaynl committed Jul 1, 2026
  • wiki: record aicom infantry fallback intake

    @rayswaynl rayswaynl committed Jul 1, 2026
  • docs: mark hc registry fix runtime-pending

    rayswaynl committed Jun 30, 2026
  • wiki: document cmdcon30 deadspawn fix + WEST-founding diagnosis + JIP filename rule - Deadspawn (FIXED, deployed): PR#122 editor-slot reaper in Init_Server.sqf deleteGroup'd JIP-selectable player-slot groups -> joiner had no wfbe_side -> enrollment self-heal exhausted -> stuck deadspawn. cmdcon30 REMOVED the reaper (back to pre-PR122 tag-only sweep). Plus getPosVisual->visiblePosition (A3-only), WFBE_Client_Logic isNil guard, WFBE_CONNECT_RETRY clear-on-disconnect. - WEST AI-commander founds 0 teams (DIAGNOSED / fix-pending; pre-existing on cmdcon28, NOT PR#122): synchronized WEST leader slot id 229 re-grabs the WEST HC -> reseat watcher re-announces ~6x -> connected-hc registry churn -> founding never stably founds while EAST founds normally. Not yet fixed. - Deploy rule (reinforced): client-side changes need a NEW pbo filename (A2-OA caches missions by filename; stale-serve makes correct fixes look inert). 18 pages updated across the JIP/HC/group-lifecycle/dead-code/compat/changelog refs. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

    rayswaynl committed Jun 30, 2026
  • Field notes: operating a 2nd HC on a live host Add a live-ops section to Headless-Client-Scaling-And-Topology covering the operational reality of running a 2nd HC (the in-code steps above are the easy part): - each extra HC needs its own interactive logon/Steam session; a non-interactive task launch fails with 0x80070520 and the HC process runs but never registers - trust the perf-audit headless:N signal, not process count or lobby slot tally - don't stack a manual restart on top of an automated restart cycle (HCs flap) - make deploy/repoint scripts idempotent so a same-build redeploy can't strand the server (the template find-replace no-op must be success, not a throw)

    @rayswaynl rayswaynl committed Jun 23, 2026
  • codex: refresh patrols hc topology route

    @rayswaynl rayswaynl committed Jun 22, 2026
  • docs(wiki): accuracy + integrity pass across the developer wiki Integrity (reader-facing, validator-invisible): - Fix 30 broken heading anchors (Page#heading 404s; validator strips #frags) - Fix 1 broken page link; strip 2 UTF-8 BOMs - Restore the Function-And-Module-Index legacy compiled-alias table that 3 pages cite as the canonical owner (10 aliases re-verified vs master) Accuracy (~190 corrections, source-verified vs origin/master then adversarially re-verified with a docs-head-baseline + false-positive screen): - Missing vehicle-array entries: RM70/RM70_ACR (US/RU/US_Camo artillery), Mi17_medevac_CDF (RU salvage) - Loop/condition errors: town-patrol `||`->`&&` (game-over exit inversion); camp scan "~1s"->`sleep 0.1`; missing "Air" town-wake entity class - Stale claims corrected: RU PARAAMMO is live (not dead code); water-avoidance loops are capped; HC has a reseat heartbeat; HC sleep->send timing - AI-Squad templates replaced by the 2026-06-14 combined-arms rebuild - Misc value/log-string/label fixes across ~50 reference pages No gameplay source changed.

    @rayswaynl rayswaynl committed Jun 21, 2026
  • codex: route patrols v2 current master

    @rayswaynl rayswaynl committed Jun 13, 2026
  • wiki: prune ui runtime quickrefs

    @rayswaynl rayswaynl committed Jun 5, 2026
  • wiki: harvest static supply ui pv scout wave

    @rayswaynl rayswaynl committed Jun 4, 2026
  • claude: restore + extend Headless-Client-Scaling-And-Topology RECOVERY: commit 41c5a20 accidentally reverted the 829e7f5 heavy-expansion (the shared worktree clobbered the file between my edit and git add). This restores the full expanded page from 829e7f5 AND adds the lost 'Improving the current HC' roadmap (with feasibility) + a concrete #1/#5 implementation note (Mode-2 server fallback + least-loaded HC selection, A2-OA-safe SQF). Published from a clean separate clone to avoid the shared-worktree clobber. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    @rayswaynl rayswaynl committed Jun 2, 2026
  • claude: HC scaling page — add 'Improving the current HC' roadmap with feasibility 9-item prioritised roadmap (reliability / effectiveness / hygiene), each with a source-cited fix sketch AND a feasibility read (effort + risk + A2-OA blocker). Constraints note: SQF mission changes need repack + private-server test (no CI); no setGroupOwner/remoteExec/params caps the design (no live transfer). Highest-value + feasible: #5 load-balanced selection + #1 server fallback. Lowest feasibility: delegating purchases/patrols/support (engine- blocked, per-category rewrite). Committed only my file. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    @rayswaynl rayswaynl committed Jun 2, 2026
  • claude: heavily expand Headless-Client-Scaling-And-Topology (source-grounded discovery) Full rewrite grounded in a 7-thread source discovery (121 findings). Adds: the THREE delegation modes (0 server / 1 player-client FPS-gated / 2 HC) with the version gate; the complete end-to-end HC path (registration handshake, random selection, create-local, report-back, cleanup); the OFFLOAD CEILING (~40-60% - town infantry + static-defence gunners delegatable; patrols, support, ALL purchases, respawn, supply-truck server-pinned); running a 2nd HC (mission.sqm has only ONE forceHeadlessClient slot - needs a 2nd); server sizing (AX-Ryzen vs CCX23) + 'what an extra CCX23 could do' (HCs same-DC / bot+DB offload / test server / CI+monitoring); HC specialisation; ideation (port Mode 1's load-balancing to Mode 2); and robustness gaps (DR-21/42, Mode 2 crudeness, delegate-ai dead path, guer-only town statics). Public- wiki safe (no IPs/hosts; provider/tier framing only). Committed only my file. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    @rayswaynl rayswaynl committed Jun 2, 2026
  • claude: HC scaling page — add server-sizing for a co-located 2nd HC A2 OA is single-threaded/32-bit: prioritise per-core GHz over core count (~4 fast cores + ~8GB for server+2 HCs; loopback sync = ~no extra bandwidth). Leads with 'do you even need a new box' (a 2nd HC is one more -client process on a box with spare fast cores). Sizing recommendation names Hetzner AX (Ryzen dedicated) as the fit + Cloud CCX as alternative. No IPs/hostnames/ account — provider+tier sizing only. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    @rayswaynl rayswaynl committed Jun 2, 2026
  • claude: add Headless-Client-Scaling-And-Topology (multi-HC mechanics + topology + specialisation) Source-grounded design note: the mission keeps a LIST of HCs and random- load-balances AI across them (Server_DelegateAITownHeadless.sqf:24-30, Init_HC.sqf, Server_HandleSpecial.sqf:128, Init_Server.sqf:110). Covers the LAN-vs-WAN placement rule, HC specialisation (pin categories/regions, host new AI on a 2nd HC, offload only at spawn time since A2 OA has no setGroupOwner), and DR-21/42 prerequisites. PUBLIC-WIKI SAFE: no IPs, hostnames, or infrastructure specifics (generic WAN/LAN wording only). Committed only my own file. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

    @rayswaynl rayswaynl committed Jun 2, 2026