chore(wolf): update session tracking state

anatomy.md/cerebrum.md/memory.md/token-ledger.json/hooks state from the
day's session covering bug-1584/1585/1587-1596 and task 12.5.

.wolf/anatomy.md

@@ -1,7 +1,7 @@
 # anatomy.md
 
-> Auto-maintained by OpenWolf. Last scanned: 2026-08-18T14:36:36.835Z
-> Files: 565 tracked | Anatomy hits: 0 | Misses: 0
+> Auto-maintained by OpenWolf. Last scanned: 2026-08-19T09:46:55.696Z
+> Files: 586 tracked | Anatomy hits: 0 | Misses: 0
 
 ## ../../../../../../tmp/
 
@@ -28,6 +28,10 @@
 - `trap_repro.sh` (~301 tok)
 - `xtest_script.txt` (~52 tok)
 
+## ../../../../../../tmp/claude-1000/-home-rshen-projects-xce-worktrees-scheduler-framework/fe2ba1fe-c8fb-4d72-986e-1c8e2406aa3a/scratchpad/
+
+- `strstr_probe.c` — define _GNU_SOURCE (~236 tok)
+
 ## ./
 
 - `.clang-format` (~226 tok)
@@ -214,7 +218,7 @@
 
 ## .claude/skills/xce-deploy-ha/
 
-- `SKILL.md` — XCE HA Cluster Deployment; now documents Step 4d (apiserver) and Step 4e (start-all.sh combined shortcut) (~5900 tok)
+- `SKILL.md` — XCE HA Cluster Deployment; now auto-picks real `xcecd` (sudo, root) vs `xcecd_virtual` per host via `/run/systemd/system` (~8200 tok)
 
 ## .claude/skills/xce-test-ha/
 
@@ -613,10 +617,10 @@
 
 - `brief.md` — Brief: scheduler-core (~2035 tok)
 - `design.md` — Technical Design: scheduler-core (~17063 tok)
-- `requirements.md` — Requirements Document (~25234 tok)
+- `requirements.md` — Requirements Document (~26234 tok)
 - `research.md` — Research & Design Decisions: scheduler-core (~4398 tok)
 - `spec.json` (~200 tok)
-- `tasks.md` — Implementation Plan (~31690 tok)
+- `tasks.md` — Implementation Plan (~32131 tok)
 
 ## .kiro/specs/scheduler-core/.solomd/
 
@@ -668,7 +672,7 @@
 - `ha-async-io.md` — HA Protocol & Async I/O (~4721 tok)
 - `persistence.md` — WAL & Persistence Durability (~3007 tok)
 - `product.md` — Product Overview (~1509 tok)
-- `roadmap.md` — Roadmap (~13249 tok)
+- `roadmap.md` — Roadmap (~13890 tok)
 - `security.md` — Security Standards (~1636 tok)
 - `structure.md` — Project Structure (~1348 tok)
 - `tech.md` — Technology Stack (~1500 tok)
@@ -859,7 +863,7 @@
 - `ha_convergence.c` — Declares char (~8389 tok)
 - `ha.c` — define _GNU_SOURCE (~125491 tok)
 - `ha.h` — ifndef HA_H (~36503 tok)
-- `job.c` — include <errno.h> (~28321 tok)
+- `job.c` — include <errno.h> (~26846 tok)
 - `join.c` — include "config.h" (~2994 tok)
 - `main.c` — include <argp.h> (~5107 tok)
 - `meson.build` (~3386 tok)
@@ -937,15 +941,29 @@
 
 - `meson.build` (~1183 tok)
 
+## src/scheduler/include/
+
+- `core.h` — if !defined(_CORE_H_) (~7848 tok)
+- `repl.h` — if !defined(_REPL_H_); task 12.5 added xce_repl_on_replay_complete()/xce_repl_check_replay_complete() (bug-1594) (~2258 tok)
+
 ## src/scheduler/src/
 
-- `core_events.c` — include <errno.h> (~9992 tok)
-- `main.c` — include <argp.h> (~3552 tok)
-- `repl_info.c` — include <stdio.h> (~7726 tok)
+- `core_clock.c` — include <stdarg.h> (~3441 tok)
+- `core_events.c` — include <errno.h> (~10786 tok)
+- `core.c` — include <errno.h> (~4310 tok)
+- `driver.c` — include <errno.h> (~9230 tok)
+- `main.c` — CLI entry (argp table); bug-1585 added --record-snapshot-interval, wired to sched.record_snapshot_interval (~4650 tok)
+- `recording_session.c` — include <stdint.h> (~2706 tok)
+- `repl_break.c` — include <stdio.h>; task 12.5 added unconditional repl_print_event_summary() in xce_repl_on_event() (bug-1595), 790/800 lines (~8861 tok)
+- `repl_info.c` — include <stdio.h>; task 12.5 rewrote info-resource for metadata + GLOBAL/SHARED-only (bug-1593), repl_job_state_name() now non-static; 856 lines, STILL OVER 800-line cap (pre-existing debt, was 883) (~8462 tok)
+- `repl_internal.h` — if !defined(_REPL_INTERNAL_H_); task 12.5 added replay_complete field + repl_job_state_name extern (~3602 tok)
+- `repl_params.c` — include <errno.h> (~4299 tok)
+- `repl.c` — include <ctype.h>; task 12.5 added xce_repl_on_replay_complete(), replay_complete guard on step/continue, quote-stripping in repl_split_tokens() (bug-1594/1596) (~6611 tok)
 - `resource.c` — include <errno.h> (~4526 tok)
-- `scheduler.c` — include <errno.h> (~15589 tok)
-- `sim_loop.c` — include <errno.h> (~9158 tok)
+- `scheduler.c` — include <errno.h> (~15773 tok)
+- `sim_loop.c` — include <errno.h>; task 12.5 added xce_repl_check_replay_complete() call after replay loop ends (bug-1594), 770/800 lines (~9386 tok)
 - `sim_picker.c` — include <errno.h> (~1374 tok)
+- `sim_verify.c` — include <stdio.h> (~4694 tok)
 - `snapshot_load.c` — include <errno.h> (~4005 tok)
 - `snapshot_wire.c` — include <errno.h> (~2611 tok)
 
@@ -955,24 +973,33 @@
 
 ## src/scheduler/testsuite/
 
-- `meson.build` — Declares vtask_entry (~14818 tok)
-- `test_core_events.c` — Declares and (~7850 tok)
-- `test_repl_breakpoints_inspection.c` — Declares and (~11461 tok)
-- `test_sim_event_loop.c` — Declares char (~8062 tok)
+- `meson.build` — Declares vtask_entry; registers sched_resval_ownership_test (bug-1585) (~15150 tok)
+- `test_clock.c` — Declares uint64_t (~5450 tok)
+- `test_core_events.c` — Declares and; real dup_val() reimpl added for link (bug-1585) (~7880 tok)
+- `test_driver_translate.c` — Declares field; full_val/delta_val fixtures now set ->which=XCE_VAL_NUMERIC (bug-1585 dup_val fix exposed the gap) (~9580 tok)
+- `test_repl_breakpoints_inspection.c` — Declares and (~16735 tok)
+- `test_repl_equivalence.c` — Declares char (~7554 tok)
+- `test_repl_execution_control.c` — Declares directly (~8798 tok)
+- `test_resval_ownership.c` — bug-1585 regression: RESVALS_FULL/RESVALS_DELTA heap-use-after-free, with a REAL (non-stubbed) kill_sched_resval to actually catch it (~5471 tok)
+- `test_sim_alt_events.c` — Declares char (~4863 tok)
+- `test_sim_event_loop.c` — Declares char (~8156 tok)
 - `test_sim_isolation.c` — Declares names (~8174 tok)
+- `test_sim_verify.c` — Declares mismatch (~9312 tok)
 
 ## src/tools/
 
-- `deploy-ha.sh` — HA deploy script; Step 3 now also conditionally generates an apiserver config+start script (opt-in on /tmp/apiserver presence) plus an always-generated start-all.sh sequencing controller→health-check→worker/scheduler/apiserver (~7800 tok)
+- `deploy-ha.sh` — auto-detects real `xcecd` vs `xcecd_virtual` (XCE_REAL_WORKER / /run/systemd/system), generates sudo-wrapped start-worker.sh for real (~8000 tok)
 - `xce-build.sh` — Declares of (~2508 tok)
 
 ## src/worker/src/
 
 - `resource_sync.c` — include "resource_sync.h" (~3926 tok)
+- `xcecd_dbus_manager.c` — include "xcecd_dbus_manager.h"; dbus_workers_update() (bug-1589 FIX): worker-list `while` loop now captures `next = worker->next` up front and every early continue/return advances to it, instead of re-testing the same worker forever or `return true`-aborting the whole scan on the first WORKER_DONE/EXITED node (~3755 tok)
 
 ## src/worker/testsuite/
 
-- `meson.build` — Worker unit test suite (component-failover-adoption, task 2.1). White-box (~872 tok)
+- `meson.build` — Worker unit test suite (component-failover-adoption, task 2.1). White-box (~1147 tok)
+- `test_dbus_workers_update.c` — Declares started (~3398 tok)
 - `test_resource_sync.c` (~1374 tok)
 
 ## src/xceadmin/

.wolf/cerebrum.md

@@ -52,6 +52,13 @@
 - **Testing a bash function that both populates caller-scope globals AND prints via a `log()`-style echo (task 8.2, `aggregate_verdict()`): never capture its output with `$(...)` command substitution if you also need the globals it set.** Command substitution forks a subshell in bash — any associative-array/variable writes the function makes are lost the instant the subshell exits, even though the function's exit code and printed text survive. Fix: call the function directly with plain `>` file redirection (`fn ... > "$tmpfile" 2>&1; rc=$?`), which does NOT fork a subshell, then read `$tmpfile` separately for text assertions while the just-set globals remain live in the current shell. This is exactly `.claude/skills/xce-test-resources/verdict-logic-selftest.sh`'s `run_agg()` helper.
 - **This codebase's established pattern for a persistent bash-tooling regression test is `.claude/skills/xce-test-ha/soak-verdict-selftest.sh`**: name it `<subject>-selftest.sh`, `source` (not execute) the target script to get its functions in scope, use a `check DESC EXPECTED ACTUAL` string-equality helper tallying a `fails` counter, end with a `SELFTEST PASS`/`SELFTEST FAIL (N failed)` line and exit 0/1. `run-resource-verify.sh` itself is written to support this — it has a `[[ "${BASH_SOURCE[0]}" == "${0}" ]]` guard at the bottom specifically so sourcing it never runs `main()`/starts a live deployment. Reused for `verdict-logic-selftest.sh` (task 8.2).
 
+- **Sandbox capability update (2026-08-19): `sudo` now works passwordless here** (`sudo -n true` succeeds) — this unblocks anything previously gated on "this session/skill cannot sudo" (e.g. real `xcecd`, `worker/src/main.c`'s `check_user()`). Still true, and load-bearing for anything assuming a real production host: this sandbox has **no systemd, no dbus, no `/run` at all** (`ls /run` → No such file or directory; PID 1 is `s6-svscan`; no `dbus-daemon`/`systemctl` even under `nix develop`). Root access and a real init/dbus stack are independent axes — don't assume sudo working means systemd-dependent things (job dispatch via `org.freedesktop.systemd1.Manager.StartTransientUnit`, anything needing `/run/dbus/system_bus_socket`) will also work. `/run/systemd/system` existing is the standard `sd_booted(3)`-style signal for "a real systemd instance is running as this host's manager" — used in `src/tools/deploy-ha.sh` to auto-pick `xcecd` (real) vs `xcecd_virtual` per host (`XCE_REAL_WORKER=0|1` overrides).
+- **A root-owned file created by a daemon under `sudo` is not automatically readable by the invoking non-root user** — confirmed live (bug-1592): real `xcecd`'s main log file comes out `root:root` mode `640` even though the SAME binary explicitly `chmod(..., 0666)`s its domain socket and its `xcecd_worker.log` sibling. Don't assume "the daemon already chmod's some of its output files permissive" generalizes to all of them — check each file's actual mode after a live run before writing a verification command that reads it as the invoking user; prefix with `sudo` where it doesn't.
+
+- **xce_resource_t (xce/types.h) has a `flags` bitmask (XCE_RESF_BUILTIN/DYNAMIC/GLOBAL/SHARED/EXTERNAL/RELEASE/DEFAULT_HOST/CONSUMABLE/LIMIT) distinguishing a genuinely cluster-wide resource (GLOBAL/SHARED, stored ONCE in `core->sharedInst`, a zarray_t indexed by the resource's `idx` -- the same pool RESVALS_FULL/RESVALS_DELTA event handling and snapshot save/load already treat as authoritative) from a host-based/per-worker-distributed resource (stored per-worker in `w->resvals`/`w->resInUse`).** Before this session, `repl_info_resource_one()` (scheduler REPL) conflated the two -- aggregated total/available across every worker's resvals for EVERY resource regardless of kind, a fabricated number for a host-based resource (cpus/mem) and redundant with `info worker`'s own correct per-host display. When adding any new resource-value display/log code, check `r->flags & (XCE_RESF_GLOBAL|XCE_RESF_SHARED)` first to decide which pool (`core->sharedInst` vs per-worker `resvals`) is the correct source of truth -- see bug-1593.
+- **`repl_info.c` (scheduler REPL state-inspection dumps) was already OVER this project's 800-line cap (883 lines) before this session even started** -- accumulated from three prior sessions' bug fixes (bug-1587/1590/1591) each adding a little without ever triggering a new sibling-file split. A task-12.5 rewrite of the resource-display functions brought it down to 856, still over. Flag this to the user rather than silently ignoring it or unilaterally doing a 4th-generation file split (repl.c -> repl_break.c/repl_info.c -> repl_params.c precedent) without being asked -- a split touches meson.build, test_sim_isolation.c's per-file scan macros, AND every white-box test TU's #include list (real, mechanical work, not worth taking on speculatively).
+- **A `repl_split_tokens()`-style whitespace tokenizer with zero quote-awareness silently breaks any downstream parser that requires a whole-string-anchored regex match** (e.g. `parse_jobid()`'s `pmatch[0].rm_so==0 && pmatch[0].rm_eo==len` check) when a user quotes an argument out of habit (common when copy-pasting a value from another command's own output). The numeric fallback path (`strtoull()`) then silently parses the leading quote character as "not a digit" and returns 0 rather than erroring -- producing a confusing wrong-answer, not a clear parse failure. Worth checking for in any REPL/CLI tokenizer this codebase adds. See bug-1596.
+
 ## Do-Not-Repeat
 - Don't run `clang-format -i` on whole existing files here — the repo isn't clang-format-clean at baseline, so a full reformat buries the real diff; recovering via `git checkout HEAD -- <file>` on a file with other uncommitted work destroys that work irrecoverably.
 - Before citing a bug as "still open" or a spec as "blocked," cross-check `.wolf/buglog.json`'s actual fix status and the spec's own tasks.md/commit history — blocked_on notes and prior narrative claims (including this file's own past entries) can be stale.

.wolf/memory.md

@@ -2,6 +2,13 @@
 
 > Chronological action log. Hooks and AI append to this file automatically.
 
+| 00:39 | Enhanced xce-deploy-ha to run the real production `xcecd` worker (not just `xcecd_virtual`) now that `sudo` works passwordless in this sandbox: deploy-ha.sh auto-detects via `/run/systemd/system` presence (real on a host, virtual here), overridable with `XCE_REAL_WORKER=0\|1`; generated `start-worker.sh`'s real branch wraps the daemon in `sudo` and preflights `sudo mkdir -p /var/run` (xcecd hardcodes db_path/domain_sock_file there, not flags). Live-smoke-tested both branches: virtual generates unchanged; real genuinely starts as root (`ps` confirmed) — but live-testing also caught a real bug (logged bug-1592): `xcecd.log` comes out `root:root` mode `640`, not chmod'd permissive like `xcecd_worker.log`/the domain socket, so verification greps need `sudo` too. SKILL.md updated throughout (Prerequisites, Step 2/4b/Clean-restart kill lines, new "Real worker" subsection, 6 new Troubleshooting rows) to reflect this and the single-instance + no-dbus/systemd-here-so-no-job-dispatch caveats | src/tools/deploy-ha.sh, .claude/skills/xce-deploy-ha/SKILL.md, .wolf/buglog.json (bug-1592) | both worker paths live-verified; real-worker log-permission gotcha found and documented | ~180k |
+| 21:57 | Session wrap: bug-1589 (dbus_workers_update linked-list stuck-node/early-return bug, worker.c, fixed+live-verified by concurrent agent), bug-1590 (repl_info.c resource value type-dispatch, fixed+2 new tests, fresh non-incremental build Ok:41/Fail:1), bug-1591 (15 raw-jobid sites -> fmt_jobid across scheduler, fixed+5 test assertions updated, Ok:41/Fail:1) all fixed, verified, and buglog updated in place | src/worker/src/xcecd_dbus_manager.c, src/scheduler/src/{repl_info.c,repl_break.c,sim_loop.c,sim_verify.c,driver.c,recording_session.c,core_events.c}, src/scheduler/testsuite/{test_repl_breakpoints_inspection.c,test_sim_alt_events.c,test_repl_execution_control.c,test_sim_event_loop.c,test_sim_verify.c}, .wolf/buglog.json | all 3 bugs CLOSED, only pre-existing bug-1560 failure remains in suite | ~900k |
+
+| 21:10 | Added roadmap backlog item "xcecd_virtual concurrent-job support (MXJ > 1)" per user request — treat as future enhancement project, dependent on bug-1589 landing first | .kiro/steering/roadmap.md | recorded in Future Specs/Backlog section | ~4k |
+| 21:40 | Logged bug-1590 (resource values not type-dispatched — string/bool resources like type/model show 0.00 in REPL) after auditing the full stack: confirmed C-side scheduler REPL (repl_info.c) and some scheduler/controller TRACE logging force everything through %.2f/Value_u_fvalue; confirmed Go apiserver (convert.go's cNameValToWorkerResval) and xcecli already handle all types correctly — not a systemic issue, scoped to scheduler C-side debug tooling | .wolf/buglog.json (bug-1590) | logged, not fixed | ~40k |
+| 21:45 | Logged bug-1591 (raw packed jobid in scheduler REPL "info queue"/"info worker" output, same class as bug-1588) — grepped src/scheduler/src/ for all raw jobid=%llu sites (~17 found across repl_info.c/repl_break.c/sim_loop.c/sim_verify.c/driver.c/core_events.c/recording_session.c), logged full inventory for future fix triage | .wolf/buglog.json (bug-1591) | logged, not fixed | ~8k |
+
 | 03:30 | resource-value-verification task 5.1: created run-resource-verify.sh (deployment step only -- invokes xce-deploy-ha's deploy-ha.sh into a fresh mktemp scratch dir, then generated start-all.sh; confirms controller/scheduler(-vvv)/apiserver(debug) all genuinely up via bounded log-tail polls; trap-based cleanup registered before start_all(), covers mid-deployment abort per Req 6.5; `--keep-running` flag skips process teardown, logs always preserved). Found+fixed a real bug live-testing the SIGTERM-mid-flight abort path: cleanup()'s original 4-separate-pkill-calls design non-deterministically left scheduler/apiserver processes running when triggered via an async signal trap (worked fine via normal exit/fail() path) -- fixed by combining into 1 pkill call (alternation) + retry-until-verified loop (bug-1575); verified reliable across 5+ repeated external-SIGTERM interrupts. Also updated SKILL.md to pin the previously-placeholder flag/env names and status. All live-validated: normal PASS+cleanup (~5s, 0 leftover procs, logs preserved), fake-apiserver-crash mid-abort (controller+scheduler killed correctly), --keep-running (processes genuinely left running, manually torn down) | .claude/skills/xce-test-resources/{run-resource-verify.sh (new),SKILL.md}, .wolf/buglog.json (bug-1575, bug-1576) | READY_FOR_REVIEW, all live tests green | ~950k |
 
 | 15:05 | Re-ran /kiro-validate-impl component-failover-adoption post-Phase-6 remediation, 3 fresh independent parallel checks (not reused from earlier this session): mechanical (lib 298/298, scheduler 40/41 known-unrelated, apiserver+test-backend all pass incl. 6 new heartbeat tests, worker regression-clean, TBD/secrets grep clean, smoke boot clean), design/boundary (frozen client.c/client_internal.h/client_failover.c/h confirmed untouched via git diff, netio.h/xcectx.c changes purely additive, dependency direction clean, 1 minor disclosed File-Structure-Plan drift same class as prior candidates.go finding), requirements/task-completion (skeptically verified tasks 6.1/6.2 genuinely meet their own observable-completion bars against real on-disk log evidence, not just prose -- exactly the check that would have caught the original task-5.2 overclaim). Final decision: GO | .kiro/specs/component-failover-adoption/{tasks.md,spec.json} (already GO from remediation turn, reconfirmed) | GO reported to user | ~1.1M |
@@ -2183,3 +2190,211 @@
 | 07:38 | Session end: 208 writes across 37 files (design.md, tasks.md, research.md, requirements.md, spec.json) | 83 reads | ~562490 tok |
 | 14:40 | Real /xce-test-resources end-user invocation (default 9-resource set, no --keep-running) | .claude/skills/xce-test-resources/run-resource-verify.sh | VERDICT resources=9 layers_checked=3 mismatches=0 unreachable=0 rejected=0 result=PASS, exit=0, ~8s wall-clock, clean teardown (proc-scan verified), logs preserved under /tmp/xce-resverify.bHJBGP/logs | ~1200 |
 | 07:41 | Session end: 208 writes across 37 files (design.md, tasks.md, research.md, requirements.md, spec.json) | 83 reads | ~562490 tok |
+
+## Session: 2026-08-18 08:09
+
+| Time | Action | File(s) | Outcome | ~Tokens |
+|------|--------|---------|---------|--------|
+| 08:30 | Surveyed task status across scheduler-framework spec tree | .kiro/specs/scheduler-{framework,core,allocation,preemption}/{spec.json,tasks.md} | scheduler-core 100% implementation-complete; allocation+preemption still requirements-only, now unblocked; compiled follow-up list from Implementation Notes | ~15k |
+| 19:47 | Edited src/scheduler/src/core_clock.c | added 1 condition(s) | ~853 |
+| 19:47 | Edited src/scheduler/include/core.h | expanded (+16 lines) | ~279 |
+| 19:47 | Edited src/scheduler/include/core.h | modified otherwise() | ~266 |
+| 19:48 | Edited src/scheduler/src/core.c | modified returns() | ~235 |
+| 19:48 | Edited src/scheduler/src/scheduler.c | modified if() | ~227 |
+| 19:48 | Edited src/scheduler/testsuite/test_clock.c | modified test_epoch_ns() | ~580 |
+| 19:48 | Edited src/scheduler/testsuite/test_clock.c | 2→4 lines | ~76 |
+| 19:49 | Session end: 7 writes across 5 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 6 reads | ~50187 tok |
+| 19:55 | Session end: 7 writes across 5 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 7 reads | ~50187 tok |
+| 19:59 | Edited src/scheduler/src/repl.c | 4→5 lines | ~26 |
+| 19:59 | Edited src/scheduler/src/repl.c | modified repl_format_ts_human() | ~372 |
+| 20:00 | Edited src/scheduler/src/repl_internal.h | expanded (+7 lines) | ~141 |
+| 20:00 | Edited src/scheduler/src/repl.c | modified repl_format_ts_human() | ~18 |
+| 20:00 | Edited src/scheduler/src/repl_info.c | modified repl_info_session() | ~150 |
+| 20:00 | Edited src/scheduler/src/repl_info.c | modified repl_info_config() | ~263 |
+| 20:01 | Edited src/scheduler/src/repl_params.c | 2→7 lines | ~61 |
+| 20:02 | Session end: 14 writes across 9 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 11 reads | ~62031 tok |
+| 20:16 | Created src/scheduler/testsuite/test_resval_ownership.c | — | ~4864 |
+| 20:17 | Edited src/scheduler/testsuite/test_resval_ownership.c | modified make_resval_delta() | ~192 |
+| 20:17 | Edited src/scheduler/testsuite/test_resval_ownership.c | 3→3 lines | ~50 |
+| 20:17 | Edited src/scheduler/testsuite/test_resval_ownership.c | 3→3 lines | ~50 |
+| 20:18 | Edited src/scheduler/testsuite/test_resval_ownership.c | modified test_resvals_delta_second_drain_does_not_use_after_free() | ~451 |
+| 20:18 | Edited src/scheduler/testsuite/test_resval_ownership.c | 5→5 lines | ~46 |
+| 20:18 | Edited src/scheduler/testsuite/test_resval_ownership.c | expanded (+8 lines) | ~278 |
+| 20:18 | Edited src/scheduler/testsuite/test_resval_ownership.c | inline fix | ~20 |
+| 20:18 | Edited src/scheduler/testsuite/meson.build | expanded (+15 lines) | ~263 |
+| 20:18 | Edited src/scheduler/testsuite/meson.build | 3→6 lines | ~73 |
+| 20:19 | Edited src/scheduler/testsuite/test_resval_ownership.c | expanded (+8 lines) | ~323 |
+| 20:19 | Edited src/scheduler/testsuite/test_resval_ownership.c | 14→14 lines | ~109 |
+| 20:20 | Edited src/scheduler/testsuite/test_resval_ownership.c | 27→31 lines | ~406 |
+| 20:20 | Edited src/scheduler/testsuite/test_resval_ownership.c | 12→13 lines | ~128 |
+| 20:21 | Edited src/scheduler/testsuite/test_resval_ownership.c | 36→41 lines | ~534 |
+| 20:21 | Edited src/scheduler/testsuite/test_resval_ownership.c | 5→5 lines | ~50 |
+| 20:21 | Edited src/scheduler/testsuite/test_resval_ownership.c | 7→7 lines | ~58 |
+| 20:21 | Edited src/scheduler/testsuite/test_resval_ownership.c | 7→7 lines | ~58 |
+| 20:23 | Edited src/scheduler/src/core_events.c | modified cause() | ~545 |
+| 20:23 | Edited src/scheduler/src/core_events.c | 1→5 lines | ~88 |
+| 20:23 | Edited src/scheduler/src/core_events.c | 1→5 lines | ~86 |
+| 20:23 | Edited src/scheduler/testsuite/test_resval_ownership.c | inline fix | ~15 |
+| 20:24 | Edited src/scheduler/testsuite/test_resval_ownership.c | inline fix | ~16 |
+| 20:28 | Edited src/scheduler/testsuite/test_driver_translate.c | modified NOTE() | ~454 |
+| 20:28 | Edited src/scheduler/testsuite/test_driver_translate.c | 5→6 lines | ~97 |
+| 20:33 | Edited src/scheduler/src/main.c | modified up() | ~159 |
+| 20:33 | Edited src/scheduler/src/main.c | 1→4 lines | ~77 |
+| 20:33 | Edited src/scheduler/src/main.c | modified 11() | ~326 |
+| 20:33 | Edited src/scheduler/src/main.c | 3→6 lines | ~49 |
+| 20:33 | Edited src/scheduler/src/main.c | added 2 condition(s) | ~358 |
+| 20:33 | Edited src/scheduler/src/main.c | modified 11() | ~128 |
+| 20:33 | Edited src/scheduler/src/main.c | modified 11() | ~131 |
+| 20:35 | Edited src/scheduler/src/repl.c | expanded (+13 lines) | ~292 |
+| 20:39 | Session end: 47 writes across 14 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 30 reads | ~162286 tok |
+| 20:45 | Deployed live 1-node controller+worker+scheduler+xceadmin cluster at /tmp/xce-bug1585-repro (localhost:19991) to reproduce bug-1585; submitted jobs via xce_test, started scheduler recording | /tmp/xce-bug1585-repro (scratch, not in repo) | scheduler crashed silently (stderr to /dev/null) during first attempt | ~40000 |
+| 20:52 | Restarted scheduler with stderr captured; live-reproduced bug-1585's TRUE root cause via ASAN: heap-use-after-free in core_events.c's RESVALS_DELTA handler (zarray_put stores rv->val by reference, driver.c's release_translated_payload frees it unconditionally -- crashes on the 2nd periodic resval report for the same worker+resource) | scheduler.stderr.log at /tmp/xce-bug1585-repro/logs | root-caused via live ASAN backtrace, not guesswork | ~8000 |
+| 20:58 | TDD: wrote test_resval_ownership.c (2 tests) with a REAL kill_sched_resval + real dup_val reimpl (existing tests all stub kill_sched_resval as no-op, which is why nobody caught this) -- RED confirmed the exact live ASAN crash at the unit level | src/scheduler/testsuite/test_resval_ownership.c, meson.build | RED confirmed | ~30000 |
+| 21:05 | Fixed core_events.c: RESVALS_FULL/RESVALS_DELTA now dup_val(rv->val) before zarray_put(), matching snapshot_load.c's existing convention. Fixed 5 other white-box test files that #include core_events.c (needed a real dup_val link stub each); fixed test_driver_translate.c's full_val/delta_val fixtures (missing ->which=XCE_VAL_NUMERIC, exposed once dup_val started actually reading it) | src/scheduler/src/core_events.c + 5 testsuite files | GREEN, full scheduler suite 41/42 (only pre-existing bug-1560) | ~20000 |
+| 21:12 | Live-reconfirmed the fix: restarted scheduler against same deployment, 22+ periodic resval cycles with zero crash (previously guaranteed crash on cycle 2); completed a full record-start/submit-3-jobs/record-stop/dump/REPL-step cycle cleanly -- job count correctly grew 6->9 after `step 12`, matching xce_test's live joblist | /tmp/xce-bug1585-repro/records2 | confirmed live end-to-end | ~10000 |
+| 21:20 | Implemented --record-snapshot-interval CLI flag (main.c, OPT_RECORD_SNAPSHOT_INTERVAL=262), mirroring --record-dir's exact pattern; wired to sched.record_snapshot_interval (field + consumption logic already existed since task 9.3, only CLI wiring was missing per tasks.md's own review note). Added repl.c banner clarification line | src/scheduler/src/main.c, repl.c | live-verified: 3 periodic SnapshotEntry checkpoints now appear mid-recording, REPL loads pre-populated (14 jobs, zero stepping needed) | ~12000 |
+| 21:25 | Investigated bug-1586 (worker periodic resvals reading 0 in xcecd_virtual): sudo -n true fails (PAM init error, confirmed non-interactive) -- fell back to code inspection. worker/meson.build: xcecd and xcecd_virtual share the IDENTICAL source list; VIRTUAL_CLUSTER only gates root/sqlite/dbus/process-exec code, never resource_sync.c/node_information.c. All 11 node_info_get_* getters read real /proc, utmp, statvfs -- no mocking. Live-reconfirmed via this session's own bug-1585 deployment: worker-1 (xcecd_virtual) reported real non-zero mem/r1m/r15m/r15s/ut/pg/it/tmp; only ls/swp/io read 0 (legitimately, idle sandbox) -- matches bug-1569's prior identical finding | src/worker/src/resource_sync.c, node_information.c, meson.build (read-only) | CONFIRMED non-bug, no code change | ~15000 |
+| 21:30 | Updated .wolf/buglog.json bug-1585 (root-caused+fixed, occurrences 1->2) and bug-1586 (investigated, confirmed non-bug, occurrences 1->2) in place; validated JSON, no new duplicate IDs, preserved concurrently-added bug-1588 untouched | .wolf/buglog.json | validated clean | ~6000 |
+| 20:43 | Session end: 47 writes across 14 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 30 reads | ~162286 tok |
+| 20:46 | Edited src/controller/job.c | modified dispatch_a_job_() | ~230 |
+| 20:46 | Edited src/controller/job.c | modified on_job_start() | ~78 |
+| 20:46 | Edited src/controller/job.c | modified on_job_finish() | ~90 |
+| 20:46 | Edited src/controller/job.c | modified on_job_fail() | ~88 |
+| 20:46 | Edited src/controller/job.c | modified on_job_stop() | ~78 |
+| 20:47 | Edited src/controller/job.c | modified on_job_resume() | ~79 |
+| 20:47 | Edited src/controller/job.c | modified on_job_postdone() | ~80 |
+| 20:47 | Edited src/controller/job.c | modified set_preason() | ~87 |
+| 20:47 | Session end: 55 writes across 15 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 35 reads | ~191476 tok |
+| 03:48 | Verified bug-1588 fix build+test (read-only): controller ASAN incremental build, 1104/1104 Unity tests pass, 0 warnings/errors in job.c | src/controller/job.c | Ok:5/Fail:0 | ~3k |
+| 03:50 | bug-1588 fixed: 8 TRACE() sites in job.c (dispatch_a_job_, on_job_start, on_job_finish, on_job_fail, on_job_stop, on_job_resume, on_job_postdone, set_preason) switched from raw `%ld` packed jobid to the codebase's existing fmt_jobid() human-readable format ("cluster-apiserver-timestamp-seq"), already used everywhere else (scheduler.c, handler.c, worker, waldump) but never adopted in job.c. Build+test independently verified clean, buglog.json bug-1588 updated in place with confirmed root cause/fix | src/controller/job.c, .wolf/buglog.json | FIXED, verified | ~55k |
+| 20:50 | Session end: 55 writes across 15 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 35 reads | ~191476 tok |
+| 20:58 | Session end: 55 writes across 15 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 35 reads | ~191476 tok |
+| 21:07 | Session end: 55 writes across 15 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 35 reads | ~191476 tok |
+| 21:09 | Session end: 55 writes across 15 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 37 reads | ~191476 tok |
+| 21:11 | Edited .kiro/steering/roadmap.md | modified sketch() | ~1072 |
+| 21:11 | Session end: 56 writes across 16 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 40 reads | ~205874 tok |
+| 21:14 | Edited src/worker/src/xcecd_dbus_manager.c | modified VIRTUAL_CLUSTER() | ~541 |
+| 21:14 | Edited src/worker/src/xcecd_dbus_manager.c | modified VIRTUAL_CLUSTER() | ~287 |
+| 21:15 | Edited src/worker/src/xcecd_dbus_manager.c | modified VIRTUAL_CLUSTER() | ~102 |
+| 21:15 | Edited src/worker/src/xcecd_dbus_manager.c | modified if() | ~87 |
+| 21:15 | Edited src/worker/src/xcecd_dbus_manager.c | 10→10 lines | ~54 |
+| 21:16 | Created src/worker/testsuite/test_dbus_workers_update.c | — | ~2930 |
+| 04:15 | Ran nix develop ./src -- xce-build.sh all asan build-asan --incremental (full verification build) | src/build-asan/* | Ok:49 Fail:2 (scheduler bug-1560 expected + controller bug-1561 pre-existing overlayfs artifact, 0 files recompiled); xceadmin pytest 258 passed 3 skipped | ~3200 |
+| 21:16 | Edited src/worker/testsuite/meson.build | modified DBUS_MANAGER() | ~314 |
+| 21:16 | Edited src/worker/testsuite/test_dbus_workers_update.c | modified main() | ~31 |
+| 21:17 | Session end: 64 writes across 18 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 46 reads | ~215953 tok |
+| 21:20 | Edited src/scheduler/src/repl_info.c | 5→10 lines | ~128 |
+| 21:20 | Edited src/scheduler/src/repl_info.c | expanded (+8 lines) | ~204 |
+| 21:20 | Edited src/scheduler/src/repl_break.c | modified if() | ~128 |
+| 21:21 | Edited src/scheduler/src/sim_loop.c | modified sim_print_decision() | ~225 |
+| 21:21 | Edited src/worker/testsuite/test_dbus_workers_update.c | added 3 condition(s) | ~509 |
+| 21:21 | Edited src/scheduler/src/sim_verify.c | expanded (+6 lines) | ~175 |
+| 21:21 | Edited src/worker/testsuite/test_dbus_workers_update.c | 128 → 160 | ~6 |
+| 21:21 | Edited src/scheduler/src/sim_verify.c | modified switch() | ~444 |
+| 21:21 | Edited src/scheduler/src/sim_verify.c | modified switch() | ~356 |
+| 21:22 | Edited src/scheduler/src/driver.c | 11→16 lines | ~172 |
+| 21:22 | Edited src/scheduler/src/recording_session.c | expanded (+6 lines) | ~163 |
+| 21:22 | Edited src/scheduler/src/core_events.c | modified if() | ~103 |
+| 21:22 | Edited src/scheduler/src/core_events.c | modified if() | ~113 |
+| 21:22 | Edited src/scheduler/src/core_events.c | modified if() | ~113 |
+| 21:23 | Edited src/scheduler/src/repl_info.c | modified for() | ~695 |
+| 21:24 | Edited src/scheduler/src/repl_info.c | added 2 condition(s) | ~692 |
+| 21:25 | Edited src/scheduler/src/repl_info.c | modified switch() | ~252 |
+| 21:27 | Edited src/scheduler/testsuite/test_sim_alt_events.c | modified test_baseline_run_uses_base_file_events() | ~198 |
+| 21:28 | Edited src/scheduler/testsuite/test_sim_alt_events.c | modified test_alt_events_run_uses_alt_file_events_not_base() | ~187 |
+| 21:28 | Edited src/scheduler/testsuite/test_repl_execution_control.c | modified test_pre_replay_pause_blocks_before_any_event_is_processed() | ~89 |
+| 21:28 | Edited src/scheduler/testsuite/test_repl_execution_control.c | expanded (+6 lines) | ~114 |
+| 21:28 | Edited src/scheduler/testsuite/test_repl_execution_control.c | modified test_step_1_processes_exactly_one_session_then_quit_exits() | ~104 |
+| 21:28 | Edited src/scheduler/testsuite/test_repl_execution_control.c | expanded (+13 lines) | ~308 |
+| 21:29 | Edited src/scheduler/testsuite/test_sim_event_loop.c | modified test_job_present_only_in_initial_snapshot_still_reaches_the_picker() | ~389 |
+| 21:29 | Edited src/scheduler/testsuite/test_sim_verify.c | modified test_log_mismatch_reports_logical_time_and_decision_detail() | ~292 |
+| 21:30 | Edited src/scheduler/testsuite/test_sim_verify.c | modified test_verify_against_altered_copy_reports_exact_mismatch() | ~330 |
+| 21:30 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | modified test_info_queue_reports_status_and_queued_jobs() | ~282 |
+| 21:30 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | modified test_event_breakpoint_pauses_repl_at_matching_event() | ~78 |
+| 21:30 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | expanded (+6 lines) | ~116 |
+| 21:30 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | modified test_decision_breakpoint_pauses_after_decision_output() | ~86 |
+| 21:30 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | expanded (+8 lines) | ~147 |
+| 21:33 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | modified test_info_worker_reports_string_and_boolean_resource_values() | ~1718 |
+| 21:33 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | 1→3 lines | ~50 |
+| 21:34 | Session end: 97 writes across 28 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 59 reads | ~288074 tok |
+| 04:36 | bug-1589 root-caused + fixed: xcecd_dbus_manager.c dbus_workers_update() worker-list traversal (continue-without-advance + return-true-abort) starved every job in xcecd_ctx.worker_list after the first; fixed by capturing next=worker->next and always advancing | src/worker/src/xcecd_dbus_manager.c, src/worker/testsuite/test_dbus_workers_update.c, src/worker/testsuite/meson.build | RED/GREEN-verified (git stash), live 5-element job array repro against fresh /tmp/xce-deploy-bug1589 cluster shows all 5 dispatch_a_job_+on_job_start (was only [0] before fix); worker suite 4/4 green; controller/scheduler suites unaffected by this fix (3 unrelated scheduler fails traced to a concurrent session mid-editing driver.c/recording_session.c, 3 unrelated controller fails in test_config_repl.c untouched by this change) | ~1200 |
+| 21:35 | Session end: 97 writes across 28 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 59 reads | ~288074 tok |
+| 21:38 | Created ../../../../../../tmp/claude-1000/-home-rshen-projects-xce-worktrees-scheduler-framework/fe2ba1fe-c8fb-4d72-986e-1c8e2406aa3a/scratchpad/strstr_probe.c | — | ~236 |
+| 21:38 | Session end: 98 writes across 29 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 59 reads | ~288327 tok |
+| 21:43 | Session end: 98 writes across 29 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 59 reads | ~288327 tok |
+| 21:45 | Session end: 98 writes across 29 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 59 reads | ~288327 tok |
+| 21:47 | Session end: 98 writes across 29 files (core_clock.c, core.h, core.c, scheduler.c, test_clock.c) | 59 reads | ~288327 tok |
+
+## Session: 2026-08-19 00:18
+
+| Time | Action | File(s) | Outcome | ~Tokens |
+|------|--------|---------|---------|--------|
+| 00:29 | Edited .claude/skills/xce-deploy-ha/SKILL.md | modified worker() | ~243 |
+| 00:30 | Edited src/tools/deploy-ha.sh | expanded (+29 lines) | ~637 |
+| 00:30 | Edited src/tools/deploy-ha.sh | modified script() | ~424 |
+| 00:31 | Edited src/tools/deploy-ha.sh | modified xcecd() | ~593 |
+| 00:32 | Edited src/tools/deploy-ha.sh | 7→6 lines | ~118 |
+| 00:33 | Edited .claude/skills/xce-deploy-ha/SKILL.md | 7→4 lines | ~68 |
+| 00:33 | Edited .claude/skills/xce-deploy-ha/SKILL.md | 7→8 lines | ~85 |
+| 00:33 | Edited .claude/skills/xce-deploy-ha/SKILL.md | 5→9 lines | ~163 |
+| 00:34 | Edited .claude/skills/xce-deploy-ha/SKILL.md | expanded (+35 lines) | ~728 |
+| 00:34 | Edited .claude/skills/xce-deploy-ha/SKILL.md | 4→5 lines | ~83 |
+| 00:34 | Edited .claude/skills/xce-deploy-ha/SKILL.md | expanded (+8 lines) | ~216 |
+| 00:35 | Edited .claude/skills/xce-deploy-ha/SKILL.md | 1→6 lines | ~547 |
+| 00:35 | Edited .claude/skills/xce-deploy-ha/SKILL.md | inline fix | ~113 |
+| 00:36 | Edited .claude/skills/xce-deploy-ha/SKILL.md | 5→9 lines | ~160 |
+| 00:38 | Edited .claude/skills/xce-deploy-ha/SKILL.md | 4→6 lines | ~102 |
+| 00:38 | Edited .claude/skills/xce-deploy-ha/SKILL.md | expanded (+7 lines) | ~135 |
+| 00:38 | Edited .claude/skills/xce-deploy-ha/SKILL.md | 1→2 lines | ~139 |
+| 00:44 | Session end: 17 writes across 2 files (SKILL.md, deploy-ha.sh) | 3 reads | ~18731 tok |
+
+## Session: 2026-08-19 02:00
+
+| Time | Action | File(s) | Outcome | ~Tokens |
+|------|--------|---------|---------|--------|
+
+## Session: 2026-08-19 02:00
+
+| Time | Action | File(s) | Outcome | ~Tokens |
+|------|--------|---------|---------|--------|
+
+## Session: 2026-08-19 02:04
+
+| Time | Action | File(s) | Outcome | ~Tokens |
+|------|--------|---------|---------|--------|
+| 02:16 | Edited .kiro/specs/scheduler-core/requirements.md | modified of() | ~1160 |
+| 02:16 | Edited .kiro/specs/scheduler-core/tasks.md | expanded (+7 lines) | ~482 |
+| 02:16 | Edited .kiro/specs/scheduler-core/tasks.md | removed 8 lines | ~12 |
+| 02:16 | Edited .kiro/specs/scheduler-core/tasks.md | expanded (+7 lines) | ~523 |
+| 02:17 | Edited src/scheduler/src/repl_internal.h | modified field() | ~322 |
+| 02:17 | Edited src/scheduler/src/repl_internal.h | expanded (+7 lines) | ~141 |
+| 02:18 | Edited src/scheduler/include/repl.h | expanded (+14 lines) | ~314 |
+| 02:18 | Edited src/scheduler/include/repl.h | added 1 condition(s) | ~103 |
+| 02:18 | Edited src/scheduler/src/repl.c | added 1 condition(s) | ~440 |
+| 02:18 | Edited src/scheduler/src/repl.c | added 2 condition(s) | ~264 |
+| 02:18 | Edited src/scheduler/src/repl.c | added 2 condition(s) | ~298 |
+| 02:18 | Edited src/scheduler/src/sim_loop.c | modified returns() | ~244 |
+| 02:18 | Edited src/scheduler/src/repl_info.c | modified bit() | ~131 |
+| 02:19 | Edited src/scheduler/src/repl_info.c | modified repl_resource_flags_str() | ~1534 |
+| 02:21 | Edited src/scheduler/src/repl_break.c | added 2 condition(s) | ~1329 |
+| 02:20 | Merged 4 live-testing REPL gaps into scheduler-core spec (requirements.md AC22 amendment + tasks.md task 12.5) per user request | .kiro/specs/scheduler-core/requirements.md, tasks.md | done | ~3200 |
+| 02:20 | Logged bug-1593..1596 (info resource host-vs-shared, continue/no-quit exit, events not echoed in step, quoted jobid parse) | .wolf/buglog.json | done | ~2800 |
+| 02:22 | Implemented task 12.5 fixes: info resource now shows metadata + GLOBAL/SHARED-only value from core->sharedInst; continue/EOF re-enters REPL via new xce_repl_on_replay_complete() instead of exiting; step/continue echo events via repl_print_event_summary(); repl_split_tokens() strips surrounding quotes | repl.c, repl.h, repl_internal.h, repl_info.c, repl_break.c, sim_loop.c | code written, build/tests dispatched to fork agent | ~9500 |
+| 02:23 | Session end: 15 writes across 8 files (requirements.md, tasks.md, repl_internal.h, repl.h, repl.c) | 8 reads | ~96637 tok |
+| 02:27 | Edited src/scheduler/testsuite/test_repl_execution_control.c | added 9 condition(s) | ~3110 |
+| 02:27 | Edited src/scheduler/testsuite/test_repl_execution_control.c | 1→3 lines | ~63 |
+| 02:27 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | 1→2 lines | ~11 |
+| 02:27 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | added 4 condition(s) | ~847 |
+| 02:28 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | modified test_event_breakpoint_pauses_repl_at_matching_event() | ~402 |
+| 02:28 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | modified test_decision_breakpoint_pauses_after_decision_output() | ~339 |
+| 02:28 | Edited src/scheduler/testsuite/test_repl_equivalence.c | 1→2 lines | ~11 |
+| 02:29 | Edited src/scheduler/testsuite/test_repl_equivalence.c | added 2 condition(s) | ~488 |
+| 02:31 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | modified test_info_resource_reports_string_value() | ~2110 |
+| 02:31 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | modified 1595() | ~810 |
+| 02:31 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | 8→12 lines | ~162 |
+
+## Session: 2026-08-19 02:38
+
+| Time | Action | File(s) | Outcome | ~Tokens |
+|------|--------|---------|---------|--------|
+| 02:46 | Edited src/scheduler/testsuite/test_repl_breakpoints_inspection.c | expanded (+8 lines) | ~210 |