docs(headless): record K4 resource envelope

This commit is contained in:
Erik 2026-07-27 11:21:44 +02:00
parent 776482da82
commit b2ab0956f2
7 changed files with 208 additions and 97 deletions

View file

@ -1,6 +1,7 @@
# Modern runtime Slice K — Linux headless and multi-session host
**Status:** K0K3 COMPLETE; K4 ACTIVE 2026-07-27
**Status:** K0K3 COMPLETE; K4 IMPLEMENTED, CONNECTED TERMINAL GATE ACTIVE
2026-07-27
**Parent:** `2026-07-24-modern-runtime-architecture.md`, Slice K
**Authorization:** the user approved Slices FL on 2026-07-24
**Purpose:** run efficient automated clients on Linux without a display, GPU,
@ -341,16 +342,37 @@ disposes only after the final host acknowledgement.
### K4 — 30-session stress and closeout
K4 telemetry checkpoint `cb512fd0` is complete. One process-level
absolute-deadline sampler now reports CPU, private/resident/managed memory,
GC, threads, handles, Linux descriptors/sockets, session/ownership debt,
K4 implementation is complete through `776482da`. One process-level
absolute-deadline sampler reports CPU, private/resident/managed memory, GC,
threads, handles, Linux descriptors/sockets, session/ownership debt,
shared-content leases, scheduler wakeups, and deadline lateness without adding
per-session workers or timers. The first native Linux two-account run exposed
and corrected both pre-connect deadline arming and a sub-millisecond
`Task.Delay` busy loop. Windows and native Ubuntu Headless tests pass 56/56;
the complete Release suite passes 8,773 tests / 5 skips. The user requested
that the active two-client observer process remain running, so the exact fixed-
binary connected baseline and numeric ceilings remain next. Evidence:
`Task.Delay` busy loop.
The automated 1/5/10/30-root gate opens one exact process-content owner,
isolates equal GUIDs, portal/reconnect generations, mutable Runtime/physics
state, and policy faults, and converges every root and content lease. Its
30-root workload advances 480,000 frames per root (two simulated hours),
14.4 million turns total, with recurring movement, use, combat, casting, chat,
portal, death, and reconnect. Five additional 30-root runs cancel at
randomized connect/entry/activity/portal/reconnect/teardown edges.
The committed `k4-linux-30-session` envelope is diagnostic-only. It scales
private/resident memory, managed live/heap memory, CPU, handles, descriptors,
and sockets by configured session count while bounding threads, scheduler
wakeups/catch-ups, and deadline lateness process-wide. It names every
violation and requires exact Runtime/content convergence in the final
`disposed` sample; it never changes simulation, networking, scheduling, or
teardown.
Windows and native Ubuntu Headless tests pass 67/67; the complete Release
suite passes 8,784 tests / 5 skips. The exact `776482da` native Linux process
has both available accounts in-world at Rynthid, zero faults/reconnect debt,
bounded wakeups and CPU, and every sampled dimension inside the committed
ceiling. The user requested that it remain running for observation; its
graceful `disposed` sample is the sole remaining K4/Slice-K closeout gate.
Evidence:
`docs/research/2026-07-27-slice-k4-resource-telemetry-checkpoint.md`.
- Measure a 1-session warm baseline, then 5, 10, and 30 sessions.
@ -443,3 +465,12 @@ git revert b6547ff38c375328417855200f41c8af6436f246
git revert 9569dadb578b299b24e30aec85eee86ec0c2b3fb
git revert 12b500d38362e8aae0a23a4f9712b86367006f76
```
K4 rollback, newest first:
```text
git revert 776482da8225d88a7c6128022d137d26f9266a87
git revert bd236ce553718cfb1248703f4a2b17b35733b11c
git revert 97c174fbb2d72fb94e2fe07448bb257989ed8941
git revert cb512fd0916fbf49285c9b64df1630c5a69a0d20
```