On Windows 11 24H2/25H2 (observed on builds 26200 and 26220, session
161b838c) the MS_Power_Management_Event_Provider is still registered in the
WMI repository but activating the extrinsic push query fails with
WBEM_E_NOT_FOUND - the legacy provider no longer loads on exactly the OS
builds new devices enroll with, so the power watcher degraded fleet-wide.
Arm chain: try the push event first (still alive on older builds), then
fall back to the intrinsic __InstanceModificationEvent WITHIN 30 on
Win32_Battery. The WITHIN sampling runs inside WinMgmt, not in the agent;
the handler path is unchanged (payload ignored, re-probe + diff), worst-case
detection latency ~35s incl. debounce. collector_degraded now fires only
when BOTH attempts fail; the arm log line names the active mode.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Event-driven AC/battery observation during enrollment (no polling): a WMI
Win32_PowerManagementEvent (EventCode 10) push subscription re-probes
GetSystemPowerStatus and diffs snapshots through a pure tracker into
power_state_change events - transition ac_to_battery (Warning, immediate) /
battery_to_ac (Info) / threshold_crossed at 50/30/15% (per-level lifetime
latch, lowest-crossed-only on jumps, arm-time baseline covers sessions
starting low, 5s trailing-edge debounce, emission cap 20). Never armed on
devices without a battery; arm failure surfaces as collector_degraded.
Kill-switch Collectors.EnablePowerStateWatcher (ConfigVersion 38).
power_state_change is classified non-activity on both sides
(SignalActivityClassifier + IsPeriodicOrStallEvent) so battery drain can
neither keep idle collectors alive nor heal a Stalled session.
New analyze rules, both with on_event:power_state_change interim triggers
and deliberately single-required-condition (event_data conditions scan
events independently - no same-instance join):
- ANALYZE-DEV-009 (high): thresholdPercent equals 15
- ANALYZE-DEV-010 (warning): transition equals ac_to_battery
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The JSON editors serialize the FORM object, but what users paste is usually a
rule EXPORT (e.g. received from another tenant). The create/edit whitelists
rebuilt collector parameters and evaluateOn from form fields the export does
not carry, so a pasted gather rule silently lost every collector parameter and
a pasted analyze rule silently reverted to the enrollment-end default.
Both rules pages get a single shape-tolerant mapper in types.ts:
- gatherRuleToForm unflattens the nested parameters object (authoritative when
present, so edit-merge leftovers cannot leak), maps nullable rule fields to
form defaults, imports a missing emitMode as 'always' (the rule's actual
behavior) instead of the empty-form on_change default, and now also backs
startEditing so edit and paste share one mapping.
- analyze jsonToForm maps evaluateOn onto the eval* checkboxes (an explicit
evaluateOn wins over stale form flags in edit-mode merges); terminal-only
exports fall back to the enrollment-end default.
- tags and the folded effective markSessionAsFailed now survive the import:
both forms carry tags into the create/save payloads, analyze creates carry
markSessionAsFailedDefault. notify/notifyChannelIds stay excluded — channel
ids are tenant-specific.
9 new vitest cases pin both shapes, the merge precedence, and the round-trips;
full web suite 824/824, tsc and eslint clean.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Custom rules are tenant-partitioned and functionally collision-free, but three
surfaces still keyed on the bare non-unique ruleId string:
- RuleStats global aggregate (global_{ruleId}) summed same-ID custom rules of
unrelated tenants (title/severity last-writer-wins). The global row is now
catalog-only: analyze gates on IsBuiltIn/IsCommunity, gather gates on the
reserved-namespace pattern (agent events carry no flag; custom rules can
never occupy the reserved pattern).
- The 'never ship a built-in in the CUSTOM namespace' commitment from
RuleIdPolicy was documented but unenforced: combine.js now fails on
duplicate or non-reserved built-in IDs, BuiltInRuleCatalogPolicyTests pins
the same invariants on the embedded catalog, and the GitHub reseed rejects
non-reserved IDs before writing to the global partition.
- DeleteRuleStatesForRuleIdAcrossTenantsAsync (cross-partition delete by bare
RowKey) now refuses IDs outside the reserved namespace with failed=-1 so
callers abort the follow-up global catalog delete.
Verified: prod global partitions (45 analyze + 11 gather) are fully inside the
reserved namespace, so the guards touch no existing data. New OKF doc
docs/rules/rule-id-namespace.md captures the identity/collision model.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The async queue correlation stored the event with Sequence 0, sorting it to
the top of the timeline (and tripping the time-jump badge) even though it
arrives last. Follow the BuildSessionTimeoutEvent/BuildAgentEmergencyBreakEvent
pattern: max+1 so it sorts last; WhiteGlove Part 1 anchors at the Part-1
terminal markers so a late device_setup report can never cross the web
timeline's sequence-based Part-1/Part-2 split. Also upsert the EventType
index the queue path never wrote, making the event visible to cross-session
search.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The IME processes the ESP app list in multiple passes (device-ESP evaluation
ending Skipped, real install up to an hour later), and AppInstallSummary
folded first observation -> last terminal across them: a 23-second app late
in the queue reported 4002 s (session 44862cd3, "Set TimeZone").
Backend ingest: new max-folded LastAttemptStartedAt anchors DurationSeconds
(CompletedAt - last attempt start; legacy rows without the anchor keep the
span fallback), replay-guarded additive InstallPassCount, attempt-scoped
DownloadDurationSeconds pairing, and a weaker-terminal guard (completed with
state Skipped/Postponed never overrides Installed/Error, in-batch and in the
cross-batch reconcile). StartedAt stays the earliest-wins window/version-
ordering key. AppsDashboardProjection carries the new columns.
AppVersionRegressionRadar: cutover gate — only attempt-anchored rows enter
the duration population on BOTH comparison sides, so mixed-semantics medians
can neither mask real regressions nor falsely re-arm active episodes.
Time attribution: per-pass segment pairing replaces the first-start ->
last-terminal hull. An app's seconds = sum of its ACTIVE segments (failed
attempts and evaluation passes count; inter-pass idle gaps don't), occupancy
merges the segments, hull endpoints remain for display and what-if bounds.
Agent: a terminal->active transition re-arms BOTH timing endpoints, so event
payload durationSeconds also describes the attempt.
Web/PDF/MCP: labels and tooltips say "install time (final attempt)", the
apps sessions table gets a xN multi-pass chip, the time-series chart carries
a cutover footnote, MCP tool descriptions document the semantics split
between get_app_install_metrics and get_time_attribution.
No backfill: pre-cutover rows keep span values (documented in the public
changelog). OKF doc: docs/backend/app-install-duration-attempt.md.
Suites: Functions 4383, Agent 2407, MCP 564, Web 814 - all green, build 0/0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Enrollment durations are heavily right-skewed; a handful of overnight/WhiteGlove
outliers made the Fleet Health and Dashboard average cards read like hours
(541 min) and answer no real question. Both cards now lead with the median
(robust 'typical enrollment' answer) and keep the tail visible via a P90
subtext. AvgDurationMinutes stays in both DTOs for API compatibility.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
bbc7a1d8 stamped LastIngestAt into IncrementSessionEventCountAsync's merge
entity intending it as Sessions-only ("deliberately not mirrored"), but that
same entity flows through SyncSessionIndexAsync into the index merge — the
field was already landing in SessionsIndex on every ingest batch, just
undeclared. The fail-soft drift guard has been warning ~1,200x/h in prod
since the 08-20 deploy, and a StartedAt-shift full rebuild would silently
drop the column, making index-served session reads lose lastIngestAt that
Sessions-served reads carry.
Declare it properly instead of suppressing: ConditionallyProjected in
SessionIndexFieldManifest, projected in BuildSessionIndexEntity (read side
MapToSessionSummary already handled it), sample value pinned in the
bidirectional manifest tests.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The 60-min RealmJoinTimeout deadline arms at RJ detection (the RJ agent MSI
install during DeviceSetup) and used to cut off monitoring mid-install: report
55e6afd61c9d (session 46a96078) lost the tail of a large package catalog whose
last observed package completed 3 seconds before the deadline, with the first
deployment only starting 16 minutes into the window.
When the deadline now fires with LastDeploymentPhase 100/101 and deployment
activity (phase change or package observation, persisted restart-safe in
RealmJoinFacts.LastActivityUtc; the detection-time phase capture deliberately
does not count) inside the last 60 min, the handler re-arms to
lastActivity + 60 min instead of timing out — Info event
realmjoin_timeout_extended — clamped at an absolute cap of detection + 4 h.
realmjoin_timeout carries a reason payload (hard_timeout / inactivity /
absolute_cap); a fire of a deadline incarnation superseded by a re-arm
dead-ends via the OccurredAtUtc == DueAtUtc scheduler contract. The idle case
(no activity ever) still times out at 60 min, and phase-200/210 activity never
extends. Packages are only registry-visible at completion, so the inactivity
window keeps the full original 60 min.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Agent stamps X-Send-Time-Utc (device clock, ISO-8601 round-trip) per upload
ATTEMPT; ingest parses it (year>=2000 guard, deliberately no upper clamp) and
server-stamps SentAt onto every event of the request. SentAt - OccurredUtc is
pure spool delay, ReceivedAt - SentAt is network + device-vs-server clock
offset - previously indistinguishable and only separable via median statistics.
clock_skew condition: batches carrying SentAt measure the frame directly
(spool-immune by construction - no spread cap, IME/clamped exclusions
irrelevant, IME-only sessions become measurable; evidence sentAtBatchCount);
legacy batches keep the per-event median path unchanged. CmTraceSkewTripwire
deliberately untouched: its differential median math cancels clock offset as
common mode.
Full persistence chain: Store x2 + Map, fields= projection (sentAt), MCP
get_session_events field list + rule-authoring-guide. New OKF doc
backend/send-time-frame-separation.md.
Suites: Functions 4370 (+16), Agent 2407 (+1), MCP vitest 564 + tsc 0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The labeled provenance rows duplicated values already visible in the
raw JSON dump and inflated the metadata block for every IME event. The
section is now collapsed by default behind a small "time provenance"
toggle sitting inline on the Created row (whose value it explains) -
zero extra height when collapsed, and the base label column returns to
its original width, so the default detail view is pixel-identical to
before P13. Expanding reveals the same rows, indented under Created.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The per-line self-anchoring provenance (sourceLocalTs, applied offset +
origin, observational writer offset, derived/rejected markers) reaches
the browser but was buried unlabeled in the raw JSON dump. Sequence
stays THE sort order - without exception (now pinned by a regression
test); the corrected UTC is display-only, and a backwards step of the
displayed time within sequence order is marked as information instead
of confusing the reader.
- lib/timeProvenance.ts: readTimeProvenance (camelCase/snake_case,
tolerant of truncated payloads) + classifyTimeJump with cause
precedence derived > rejected-source > era-offset; era-offset
requires applied offsets on BOTH rows - it never guesses.
- TimeJumpBadge on the row AFTER the jump, computed against the
previous RENDERED row (filters/phase boundaries respected); sky for
known causes, amber otherwise; tooltip carries both times, the delta
and the "true write order (sequence)" sentence. Threshold 5 min,
field-calibrated on 20 sessions: 12/12 normal sessions show zero
badges, the era-mixed regression session shows exactly its 6.
- Clamped chip (rare) + TIME PROVENANCE detail section: only the
applied offset says "(applied)"; the writer offset is explicitly
observational (sticky after era flip-backs); origin chips with
forward-compatible fallback for unknown values.
- EnrollmentEvent gains timestampClamped/originalTimestamp (wire
already carried them); formatUtcOffset joins lib/formatting.
Verified: 814 web tests (19 new), tsc/eslint/build clean; Playwright
against e9753578 (badges match the pre-measurement), 9f475bb2 (no
badge on line-anchored rows, detail section correct, raw mode shows
exactly the one predicted jump).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Operator side: CmTraceSkewTripwire piggybacks on the terminal counter
reconcile's partition scan (zero extra reads) and compares median
ReceivedAt-OccurredUtc deltas of IME-derived vs other events. A clean
15-min-grid divergence (residual < 2 min, >=20 samples / >=3 distinct
batches per side) emits the CmTraceTimeSkewRegression ops event
(Agent/Warning, goal: never fires). Bias-dominated sessions are
suppressed; measuredWriterOffsetMinutes is never consulted (sticky
after era flip-backs). Kill switch: CmTraceSkewTripwireDisabled.
Customer side: new clock_skew analyze condition source (skewMetric
clock_jump | sustained_offset, threshold seconds, gt/gte magnitude)
behind built-in ANALYZE-DEV-008 (300 s = the Kerberos/Entra 5-min skew
boundary). Batch-median frames with spool-spread filter, plateau step
detection (a gradual backlog-drain ramp cannot fake a jump) and
end-state persistence; IME-derived events are excluded so a CMTrace
anchoring regression can never fire the customer rule. Evidence carries
field=skewSummary/value for {{skewSummary}} interpolation.
Full mirroring chain: schema enum + skewMetric property, shared
manifest + combine.js regeneration, MCP authoring guide + semantic
lint, web SOURCES/form/card, ops-event dual-register, docs. P5 is
subsumed by the tripwire. PercentileMath is now the single nearest-rank
implementation (Apps/PlatformMetrics delegate).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Field forensics from the 1412 verification run (session 067797d8): the
label/anchor crossing recurred on a provably genuine binary, and the
new anchor-timestamp tripwire delivered — the "+02:00 measured for
IntuneManagementExtension.log" anchor (local 19:51:26.948) does not
exist in ANY log file on the device, and IME.log never produced its
expected first measurement during 24 s of growing passes beforehand.
Mechanism still unexplained; per-line anchoring keeps resolution
immune (all 58 line-anchored events of the session were correct), but
the observational layer must convict the culprit next time:
- CalibrateFrom now receives the iteration's ACTUAL file name next to
the calibration key. A mismatch — the smoking gun the two field
occurrences imply but static analysis says is impossible — logs a
Warning with both names plus the anchor, and refuses the anchor, so
even the unexplained path can no longer poison the measurement store.
- The first calibrator rejection per file logs at Debug (anchor local
time + clock). The 24 s of silent non-measurement is exactly what
made the phenomenon undiagnosable from the agent log.
232 IME tracker tests green against a verified-clean build.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Why the reader-zone assumption and the per-file offset application both
failed, the per-line self-anchoring rule with its strict freshness
contract (LastCheckedUtc on every poll look, 30 s window, no freshness
across restarts), the provenance vocabulary that keeps every event
re-derivable, and the documented limits: the bounded grid-replay edge,
snapshot readers staying on the marked fallback, and the still-open
e9753578 label/anchor crossing with its tripwire.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
IntegrityCheckFailed emergency reports used to land in App Insights
only — session e9753578's agent reported a hash mismatch and no product
surface showed it, which cost a day of mis-attributed root-causing.
They now materialize as an AgentBinaryIntegrityMismatch ops event
(Warning, Agent category, dual-registered in the web alert-rule
catalog per feedback_ops_event_types_dual_register).
With suppression for the KNOWN, accepted release race the operator
correctly called out: around every release, agents compare against
hash snapshots that do not belong to their own binary — an older agent
sees the new release's hash on its next config fetch, and a freshly
installed latest agent can receive the PREVIOUS release's hashes from
the 5-minute AdminConfiguration cache (e9753578: genuine 1410 exe
checked against the cached 1409 hash — proven by the 1409 archive
logging "verified OK" with exactly that hash). Both self-heal via the
forced self-update and are suppressed at materialization time against
the CURRENT config: reported running hash == published exe, or
reported version != latest. What remains — an agent claiming the
latest version whose binary is NOT the published one — is the real
signal: tamper, stale blob, or a build outside the release pipeline.
Missing config fails open to recording; visibility beats suppression.
Ops event only, deliberately no timeline event: the report is about
the binary, not the enrollment, and may precede session registration.
Backend suite 4320 green, web tsc clean.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The era-aware successor to the reverted per-file offset application
(04b1a7c6). At a 100 ms poll, every line of a provably fresh pass was
written essentially "now", so its own distance to the agent clock,
rounded to the 15-minute offset grid, IS its writer's UTC offset — no
cross-line state, which is what interleaved writer eras break. The line
timestamp still contributes sub-poll precision and ordering; the grid
absorbs poll latency. Origin: LineAnchored → "line-anchored".
Freshness is stricter than "the file grew": every poll look stamps
LastCheckedUtc (empty first sights included, so AgentExecutor.log
anchors from its very first content), a pass anchors only when the gap
since the last look is ≤ 30 s (FreshLineMaxAge — far above the poll,
far below the 13-minute zone where a line's AGE rounds onto the grid),
and a restart-restored bookmark deliberately carries no freshness at
all — downtime backlog must never anchor. Everything not provably
fresh keeps the uniform reader-zone fallback: wrong in absolute terms,
self-consistent for durations, and marked as such (the revert's core
lesson — partially corrected is strictly worse than uniformly wrong).
Accepted, bounded edge (characterization test): a freshly WRITTEN line
carrying a replayed timestamp aged exactly N×15 min ± 2 min anchors to
≈now — an anchored line lands at now ± 2 min by construction, and
sub-24h replays are treated as current by the historic-replay guard.
The per-file measurement stays observational (Info log +
measuredWriterOffsetMinutes) and now logs the anchor's local timestamp
as a tripwire: session e9753578 logged "+02:00 measured for
IntuneManagementExtension.log" although that file provably held no
+2-era line — a label/anchor pairing the committed code could not be
shown to produce (iteration-local anchor, single sequential call site,
green repro of the exact field scenario, no +2-capable chunk in the
window). The mechanism remains open; the resolution path is immune to
it by design, and CmTraceCalibratorFileIsolationTests pins the file
isolation so a real regression can never ship silently.
Tests: 13 new — resolves-to-T across five writer beliefs, interleaved
eras inside one chunk, first-sight/empty-first-sight/poll-gap/restart
freshness boundaries (including the nastiest backlog: exactly 30 min
old, on-grid), the grid-replay edge, and the committed two-writer-era
fixture replayed at production poll cadence across its write silences.
Agent suite 2406 green against a verified-clean build.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Two corrections to 3c25172d, both found by asking why the repo suddenly had two
fixture locations.
Location: committed test data lives under tests/fixtures/. I had put this one
inside the agent test project without checking, creating a second parallel home.
It does NOT belong in tests/fixtures/enrollment-sessions/ either — that is a
governed corpus of anonymized DecisionSignal JSONL for the Replay Harness, with a
schema, category prefixes and a paired-scenario-test rule that raw log text cannot
satisfy. So: a sibling corpus, tests/fixtures/cmtrace-logs/, with its own README
stating why it is separate.
Anonymization: the fixture carried three real Intune policy/script GUIDs from a
customer tenant. Point 1 of the existing fixture policy requires tenant-scoped
identifiers to be replaced, and I committed without applying it — the check
belonged BEFORE the commit, not after. Replaced with
policy-anon-NNNN-0000-0000-0000-000000000000, consistent with the tenant-anon /
device-anon / user-anon convention already in use. The all-zero GUID is a literal
in IME's own output and stays; the session ID stays as our diagnostic anchor and
is already referenced from the commit history and tasks/todo.md.
The new README carries the commit policy for this corpus, including the rules that
were implicit before: verbatim content (the parser is tested against real
formatting), a header stating what the fixture pins, versioned filenames so a
pinned test cannot silently change meaning, and .cmtrace rather than .log because
*.log is globally gitignored.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Captured from ImeLogs/AgentExecutor.log of the session that exposed the regression
(agent 2.0.1410; device booted on the Pacific OOBE default, agent set W. Europe at
12:13:22Z).
One file, two writer eras: 371 lines rendered at local hour 05 by a writer still
on PDT (UTC-7) and 195 at hour 14 by one already on CEST (UTC+2) - nine hours
apart, interleaved, and no line carries a bias suffix, so nothing in the file
itself says which is which. Exactly the case the per-file calibration could not
express, and the test that was missing.
Analysing the archive also turned up a SECOND, independent defect the revert
commit did not know about: the agent log reports "IntuneManagementExtension.log
writer offset measured 02:00:00" while that file contains ONLY hour-05 lines
(2022 of them, zero at hour 14). A +2 measurement is impossible from its own
content, so the calibration anchor leaked across file boundaries - files are
walked in sorted order and AgentExecutor.log precedes it. Two-eras-per-file and
anchor-leak are separate bugs; both must be fixed.
Extension is .cmtrace.txt, not .log: *.log is globally gitignored, and a fixture
has no reason to claim that extension.
Fixture only, nothing consumes it yet - the era-aware calibrator is next.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Field regression, session e9753578 (agent 2.0.1410, PST-defaulted device in DE).
Five script_started events resolved 9 h early while their script_completed
counterparts resolved correctly, inflating every platform-script duration to
~32,400 s. Reverts the application introduced in e9dba11b.
Root cause: calibration keeps ONE offset per log file, but a single file can hold
lines from two writer eras. IME restarting across a timezone change leaves older
lines in the old zone and newer ones in the new zone, in the same file. The anchor
is the newest line, so its offset was also applied to the older lines, for which
it does not hold. The class doc even claimed "self-healing on writer restart" —
that reasoning only ever covered FUTURE lines, never the ones already on disk.
The more expensive mistake was the second one: correcting ABSOLUTE timestamps
without checking what depends on their DIFFERENCES. Applying the reader zone
uniformly is wrong in absolute terms but self-consistent — both ends of a duration
are wrong by the same amount, so the duration is right. Correcting only some
timestamps produced a mixed frame inside one session, which is strictly worse than
being uniformly wrong. This is the same asymmetry the original analysis described
for AppInstallSummaries (68 min reported against 8 min real) and then reproduced.
What stays, deliberately:
- Bias-carrying lines still resolve via the writer-declared offset; that path was
never implicated.
- The measurement keeps running. It costs nothing, feeds the Info log, and is the
field data the era-aware fix has to be built on — the first anlauf got the era
question wrong from assumption rather than observation.
- Provenance now separates measured from applied (measuredWriterOffsetMinutes vs
sourceOffsetMinutes). The broken event was tagged "calibrated" while carrying an
offset that did not hold for its line; that confusion must not be expressible.
- The fallback path reports the offset it actually applied. An event stating no
offset cannot be recomputed later, which defeats the point of provenance.
Provenance earned its keep here: 57 events measured -420 (the PDT writer) against
5 at +120 made the diagnosis a two-minute query instead of median statistics over
a session.
Next: era-aware calibration keyed on the IME-STARTED marker, which we already
track, plus a test over a log file containing two writer eras — the case that was
missing.
Agent 2393 green against a verified-clean build.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Every IME-sourced event now states what the writer wrote, what offset was applied,
and how that offset was obtained:
sourceLocalTs raw local time, verbatim, NO zone suffix
sourceOffsetMinutes the offset subtracted
sourceOffsetOrigin bias | calibrated | reader-zone-fallback
Why: until now the only record of a conversion was the converted value, so a wrong
offset was indistinguishable from a right one. Proving the field defect needed the
diagnostics archive — and of the 5,386 sessions that went through a timezone
change, essentially none had one, so the analysis had to fall back to median
statistics across thousands of sessions to establish what one stored value makes
obvious per session. It also makes a stored timestamp recomputable after the fact,
without an agent release.
No zone suffix on sourceLocalTs on purpose: the value genuinely carries none, and
formatting one in would re-introduce the assumption this whole change removes.
sourceOffsetMinutes is normalised to one sense (local = UTC + offset) across both
origins. The wire bias uses the opposite GetTimeZoneInformation convention
(UTC = local + bias), so a "+480" line is reported as -480; without that the field
would flip sign depending on where the offset came from and be useless to compare.
Added centrally in TagDerivedTimestamp, which all 14 emit sites already call, so
this does not scatter through the adapter. DataJson only — no table schema change,
so no Store/Map/seed chain and no backend deploy needed to start collecting it.
Agent 2393 / backend 4313 green against a verified-clean build.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
ResolveOccurredAt rejected a source timestamp when ageHours < -1.0. That bound
could not fire for the single most common timezone error: with a one-hour offset
mistake the real value is parseLag - 3600 s, i.e. just ABOVE -1.0. It sat exactly
on the boundary of the case it was meant to catch, which is why every skewed
session sailed past it.
It was asymmetric too — a past-shifted error had to exceed 24 h before anything
noticed, while the measured field errors reached -17 h.
The bound is now derived rather than guessed: writer and reader share one system
clock, and the tracker only sees a line after it was written, so a correctly
converted source timestamp is always at or behind "now". Anything ahead means the
conversion was wrong. MaxFutureSourceSkew is 2 minutes of jitter tolerance —
deliberately far below the hour a timezone mistake produces.
The rejection log now names which bound tripped and by how much, instead of
printing a bare ageHours that reads the same for both directions.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This is the behaviour change the previous two commits prepared. ImeLogTracker no
longer substitutes its own timezone belief for the IME process's; it measures the
writer's offset from a freshly appended line and converts with that.
Freshness is stricter than "the file grew". GetSafePosition returns 0 for an
unknown file, so the first pass reads the whole file and its newest line may be
hours old — anchoring there would measure the line's age, not the writer's offset.
LogFilePositionTracker.HasSeen(), captured BEFORE GetSafePosition creates the
entry, gates calibration on growth against a size we had already observed.
Calibration runs at the END of a qualifying pass, so that pass still uses the
previously measured offset — at most one poll (100 ms) old, and an offset only
moves when a process restarts. Calibrating first would mean buffering the pass,
which is not viable: the first pass over AppWorkload.log can be hundreds of MB.
The cost is a warm-up of one growing pass, during which lines fall back to the
reader zone.
Bias-carrying lines are never used as anchors: they already state the writer's
offset, so they need no measurement and must not overwrite one.
CalibrateFrom logs at Info when the measured offset disagrees with this process's
own zone, naming the correction applied. That disagreement is precisely the
condition that silently corrupted every IME-derived timestamp until now; agreement
stays at Debug so the normal case does not add noise.
Tests are deliberately machine-independent: the assertion is always "a line
written at instant T resolves to T", asserted across writer beliefs of +2, +1, 0,
-7 and +10. That one property covers both field cases — the writer disagreeing
with us (the bug) and the writer agreeing with us (the majority of sessions, where
two stale beliefs cancel and which a fix to the agent's own cache would have
broken). Plus: no calibration on first sight, following a writer restart into a
new zone, and a declared bias honoured without being measured.
CheckLogFilesAsync is internal as a test seam — calibration depends on the pass
structure, which only a real read cycle exercises.
Agent 2390 / backend 4313 green against a verified-clean build.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Behaviour-neutral: 2381 agent and 4313 backend tests pass unchanged.
The parser resolved a bias-less line with DateTimeStyles.AssumeLocal followed by
ToUniversalTime(), i.e. with the READING process's cached zone. That silently
assumes reader and writer agree. They do not — each caches its zone at process
start and neither follows a later tzutil or Windows auto-timezone change — and the
assumption was buried inside a shared helper where no call site could see it.
CmTraceLogEntry now exposes what the line actually contains: LocalTimestamp
(Kind=Unspecified, because the value is meaningless without knowing the writer's
zone) and BiasMinutes. TimestampUtc is populated only when the writer declared its
bias, which is the one case where the offset is genuinely known. Timestamp is
removed rather than deprecated, so the compiler names every site that has to
decide — it immediately turned up ProcessLogMessageForTest, which grep had missed.
The defective path still exists, but now as one named function,
ResolveUtcAssumingReaderZone(), whose documentation states when it is wrong. All
three consumers call it and so keep their current behaviour exactly:
- ImeLogTracker routes every use through ResolveEntryUtc(), the single point where
the measured offset replaces the fallback in the follow-up change;
- StallProbeCollector and LogParserCollector read file snapshots rather than
tailing, so they cannot measure an offset and stay on the fallback for now.
The parser also no longer invents a timestamp. An unparseable time used to be
stamped DateTime.UtcNow inside the parser, making a fabricated value
indistinguishable from a real one downstream; it now reports HasTimestamp=false
and leaves the fallback to the caller, which alone knows what is safe in context
(StallProbeCollector still treats such a line as fresh — calling it stale would
hide progress).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A CMTrace line carries local time and no offset — IME 1.104 writes
DateTime.Now.TimeOfDay in both of its trace listeners. Converting it with the
agent's own TimeZoneInfo.Local assumes both processes agree on the zone. They do
not: each caches its zone at process start and neither follows a later tzutil or a
Windows auto-timezone change, so the error is Offset_writer - Offset_agent and is
zero only by coincidence.
Measured over 11,068 sessions: 26 carry a real zone-offset error, +1 h (GMT to
W. Europe, session bfff9b1c) through -17 h (a writer still on the Pacific OOBE
default while the agent had already moved to E. Australia). Note the sign — this
is why "make the agent's own clock honest" is the wrong fix: in the -17 h cases
the agent was correct and the writer was stale, and clearing the agent's cache
would have broken the majority of sessions where both are stale and the two errors
cancel.
The tracker tails at a 100 ms poll, so a line that appeared since the previous
pass was written essentially "now". That makes the writer's offset measurable:
round(lineLocal - agentUtcNow) onto the 15-minute grid every real UTC offset sits
on. Per source file, because IntuneManagementExtension.log and AppWorkload.log are
written by different processes; re-measured on every fresh anchor, so a writer
restart is picked up within one poll.
The freshness requirement is the load-bearing part and cannot be replaced by the
residual guard: a backlog line exactly 30 minutes old is itself a clean multiple
of the grid, rounds with zero residual, and would calibrate the offset half an
hour wrong. RejectsBacklogLine_EvenWhenItRoundsCleanly pins that the calibrator
cannot detect this on its own — only the caller's freshness rule prevents it.
Additive only; no call site uses it yet.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Sequence is a persisted, strictly monotonic per-session counter assigned at emit
time (EventSequenceCounter: locked, saved before return, survives restarts). It is
immune to every clock problem. Timestamp carries the device clock frame plus, for
CMTrace-derived events, the writing process's timezone belief.
EvaluateAppInstallDurationCondition and EvaluateEventCorrelationCondition both
ordered by Timestamp and expressed "A happened before B" as a timestamp
comparison. Both now order by Sequence and compare Sequence for ordering.
Timestamp deliberately stays where an actual ELAPSED DURATION is meant:
durationSeconds, the evidence fields, and TimeWindowSeconds.
Scope, honestly: this is hardening, not a live bug fix. BuiltInAnalyzeRules.cs
contains zero event_correlation conditions, so the cross-frame join can only be
reached by a self-authored rule. And the app-install duration evaluator pairs
app_install_started with app_install_completed — both IME-sourced, one frame — so
its arithmetic was already correct. (The measured 68-vs-8-minute inflation comes
from EventIngestProcessor.AppInstall, which folds DO-collector download_progress
together with IME app_install_completed; that is fixed at the root in the agent.)
The TimeWindowSeconds check stays exposed: the window IS a duration, so there is
no alternative to a timestamp there. A +1 h skew on B makes any window below
3600 s fail to match. Only the agent-side calibration closes that.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
LastEventAt is the maximum AGENT-supplied event timestamp and therefore lives in
the device clock frame. The silence sweep compared it against a server-derived
cutoff, mixing two frames.
Measured over 11,068 sessions: 26 carry a genuine CMTrace timezone skew on their
IME-derived events, from -17 h to +1 h (a further 27 carry plain device-clock
offsets). At -17 h a live agent looks silent the moment it starts and is marked
Stalled with a customer-visible "Agent silent for 1020min"; at +1 h the sweep is
blind for an hour.
Adds SessionSummary.LastIngestAt, stamped from the server clock in
IncrementSessionEventCountAsync — reaching that method IS contact from the agent,
so it is written unconditionally rather than gated on latestEventTimestamp.
The sweep's server-side filter now spans both frames (LastIngestAt lt X or
LastEventAt lt X): filtering on LastIngestAt alone would silently drop every row
written before the field existed, because Table Storage does not match a filter
against a missing property. The authoritative decision moves into
ServerFrameLastContact() (LastIngestAt ?? LastEventAt ?? StartedAt) and is applied
in both stages — before the Stalled marker and in the stage-2 alive guard.
Deliberately not mirrored into SessionsIndex: the sweep reads Sessions, and the
index manifest is pinned bidirectionally by tests.
shared-manifests.json + .generated.ts regenerated — SharedManifestParityTests is
the drift brake and correctly went red on the new field.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Section headings now reserve keep-with-next space (charts reserve their full
height), so a title never strands at a page bottom while its table starts on
the next page. Table cells wrap up to 4 lines instead of single-line ellipsis,
and the exit-code column carries the full error description (the 24-char slice
is gone) - that text is what the app owner acts on. Failure-code column widths
rebalanced (Code 42 / Description 62 / Exit 60 / Count 16).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The pipeline's AutomaticDecompression (GZip|Deflate) removes the response
Content-Length header on virtually every backend JSON response, so
net_total_bytes_down in agent_metrics_snapshot read 0 on healthy sessions.
When the header is absent, buffer the content (LoadIntoBufferAsync, 16 MB cap)
and record the DECOMPRESSED payload size — HttpClient buffers responses anyway
(default ResponseContentRead), so no extra memory cost. Fail-soft: a buffering
error keeps 0 and leaves the failure for the caller to observe.
AutomaticDecompression itself stays untouched (pinned by
MtlsHttpClientFactoryTests).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
IME 1.104 removed the [DO TEL] log line, leaving the agent's DO collector as
the only do_telemetry source — and its gate only fires when a 3s poll itself
observes the completed download. Fast, interrupted or between-poll-completed
downloads never emit one, so DoDownloadMode stayed -1 and DoAggregator dropped
the whole row: DO bytes read as "practically null" on 1.104 fleets even though
download_progress events carried the full do* set on every poll.
- Shared ApplyDoFields for both writers: download_progress folds byte counters
and file size via Math.Max (monotonic, replay-idempotent) and sets the mode
only when >= 0; do_telemetry stays authoritative (last-write-wins) but never
regresses DoDownloadMode to the -1 unset sentinel.
- DownloadBytes now prefers actually-transferred bytes over doFileSize (the
file size used to inflate the transfer measure).
Intended effect after deploy: DO aggregates count previously-invisible
incomplete downloads, so app counts/bytes rise. No backfill of historic rows.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- New GA-only tool get_agent_efficiency_metrics: near-pass-through of the
server-side per-version aggregation (120s timeout, default 500 sessions,
optional tenantId). Added to GA_FULL + PLATFORM_ONLY catalog snapshots.
- platformWindowEcho decides truncated on the backend's sessionsScanned
(sessions actually covered) and only falls back to the has-snapshots count
for old backends — the previous comparison read false even when the cap was
hit exactly, whenever some scanned sessions carried no snapshots.
- get_platform_metrics: 90s timeout (backend fans out one storage query per
session); spool block, crashRate, deliveryLatency, computedAt,
computeDurationMs and fromCache are no longer dropped on the MCP side;
description clarifies totalBytesDown is the agent's own HTTP traffic.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- New GET /api/global/metrics/agent-efficiency (GlobalReadOrAdmin, ?tenantId=):
per-agent-version percentiles (CPU, working set, private bytes, threads,
handles, spool, API latency/requests), crash rates per version, top offender
sessions per dimension, 5-min cache keyed (days, limit, tenant). First
backend consumption of agent_thread_count/agent_handle_count.
- PlatformAgentMetricsResponse.SessionsScanned: sessions the scan actually
covered — the honest basis for the caller's truncated flag (Sessions.Count
only counts sessions WITH snapshots).
- Per-session fetch no longer drains the full Events partition: new
GetSessionEventsByTypesAsync (multi-type OData filter + column projection +
real maxResults enforcement); the 20 delivery-latency sample sessions keep
the full fetch so latency deltas stay computed over all events.
- Shared AgentMetricsAggregation (crash tally, payload getters, bounded
fan-out, metrics event types) so the two services cannot drift.
- GetAllSessionsPageAsync gains an optional select projection (default null,
existing callers unchanged), pinned by an equivalence test.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Adds an "Export PDF" button to the app detail page that renders the full
90/30/7-day install report (KPIs, both charts, failure codes, device model
correlation, version breakdown, detection-rule/regression callouts) as a
branded A4 PDF, drawn data-driven with jsPDF — no DOM capture, so the output
is light-palette and theme-independent. jsPDF stays out of the route bundle
via on-click dynamic import (verified: separate async chunk).
Sample messages and the affected-sessions list are deliberately excluded:
the PDF is meant to be emailed outside the org.
Tracks app_report_pdf_exported in Application Insights (windowDays, scope,
durationMs, pageCount, success/errorName — chunk-load failures included).
Generic chrome (header/footer/tables/KPI cells/callouts) lives in
utils/pdf/pdfPrimitives.ts for reuse by future report exports.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The five stuck sits-d Cloud PCs carried SkipUserStatusPage=True — verified
across three agent processes (10+ registry reads, True every time) and six
hours of live Shell-Core watching with zero 62407: Windows never rendered a
user ESP page, so no exit and no AccountSetup provisioning registry can ever
exist on this flow. Arm B of ShouldTransitionToAwaitingHello was open from
the bootstrap EspConfigDetected onward, but every evaluation site hangs off
a signal the flow cannot produce, and the AdvisoryCompletion backstop arms
exclusively from an EspExiting signal — an open gate nobody knocks on, six
hours to the max-lifetime watchdog. The earlier lost-exit-through-reboot
diagnosis is retired by the same evidence.
HandleImeUserSessionCompletedV1's completion attempt now accepts the
observed skip in place of IsPostAccountSetupFinalExit (anchor + genuine IME
user-session completion + real-user desktop stay mandatory; Device
Preparation excluded). The knock deliberately lives on the IME edge, not on
desktop arrival: Fix 10's AccountSetup bounce-back would undo a desktop-side
promotion two signals later, while the IME phase line always precedes the
user-session-complete line in log order, so a bounced promotion is
re-knocked by the very next re-emission — restart re-parse included.
Docs and comments no longer claim an ESP-exit recovery that does not exist
(stray duplicate summary in ShellCoreTracker, ImeLogHost restore-callback
comment); the OKF doc is reframed on the evidence and names the residual
SkipUser=false exit-in-downtime gap honestly.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Round-four review, confirmed — and the cause sits a layer below where I had been
fixing. The IsBackfill return added last round only guarded the coordinator's own
completion logic. Upstream of it, ShellCoreTracker still raised
FinalizingSetupPhaseTriggered, the coordinator still forwarded EspExited to the
adapter before that return, and the adapter discards the args entirely
(OnEspExited(...) => EmitEspExiting()). So a replayed exit still arrived at the
reducer as an ordinary, unmarked EspExiting.
There it is decisive: HandleEspExitingV1 passes espFinalExitInFlight: true for
every arriving exit, and IsPostAccountSetupFinalExit orders exits by ingest
ordinal rather than timestamp — so a historic exit replayed today reads as
post-AccountSetup by construction. With state restored from the snapshot
(AccountSetup entered, a genuine IME user session, desktop arrived) arm C of
ShouldTransitionToAwaitingHello opens on the Device→Account handoff. The reducer
is itself a completion gate and cannot tell the two apart, so the fix has to be
upstream.
The replay is therefore narrowed to what is provably safe: the Hello-wizard start
(62404). That fact is conservative — it vetoes a premature "Hello is disabled"
skip and can never by itself complete a session, so replaying it can only make the
agent wait longer. It is also the observation session 772fe502 wrote the replay
for. A 62407 is no longer raised at all, neither exit nor failure: the exit
because Windows writes the identical description for the intermediate and the
final one and nothing downstream can order it, the failure because re-injecting it
as fresh can fail a session that recovered on retry (ANALYZE-ESP-006). Skipped
records are reported once as an agent_trace with counts and bounds — silently
dropping evidence is how this class of bug survives.
The cross-layer proof is now a pair, as the review asked:
ClassicEspExitingOnRestoredStateTests pins that the reducer opens on any arriving
exit (including one with an hour-old source timestamp), and
ShellCoreTrackerReplayScopeTests pins that the replay never produces one.
Dead machinery removed with it: EspExitedEventArgs.IsBackfill, allowEspExit, the
newest-exit selection, the IsBackfill return. The method is now called
BackfillRecentHelloWizardStart.
One pre-existing integration test drove EspExited through HandleBackfillRecord
because the replay is easier to call than the event-log watcher, and its class doc
claimed that was the production path — it never was, the replay had no caller when
that test was written. Switched to the live path and the doc corrected.
Agent V2.Core 2361, DecisionCore 538, Backend 4287, all green.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Round-three review, confirmed. The previous round stopped a replayed exit from
arming the DEFERRED re-check but left the immediate synthesis attempt right below
it untouched. My justification for that — "pre-existing behaviour" — was simply
wrong: the immediate attempt on a replayed exit never existed, because the
backfill had no caller. I introduced it, and by moving the replay behind the IME
state restore I made the dangerous state reachable by design: if the newest 62407
in the window is only the intermediate DeviceSetup→AccountSetup exit while the
restored app states are already terminal, that attempt opens the strong
AccountSetup gate before the final exit has even happened. A test was pinning the
faulty assumption as expected behaviour.
A replayed exit now returns before touching completion at all — no deferred edge,
no immediate attempt. Nothing available to it can establish its position in time:
Shell-Core writes the identical description for both exits, and every probe that
could order it reads state as it is now rather than at the event's time. The
replay keeps the job it was written for in session 772fe502 — feeding EspExiting,
FinalizingSetup and the Hello wizard to the reducer for the window in which no
agent process was observing. Completion is the exclusive business of exits this
agent watched live.
With the synthesis attempt gone, the replay no longer needs restored app states,
so it moves back into EspAndHelloHost.Start(); that also keeps this host's startup
independent of the IME host's. ImeLogHost.onStateRestored goes back to doing just
the re-check, which is still needed for a live exit observed before the restore
finishes.
The test asserting the faulty assumption is inverted, plus coverage that the
replay is not neutered (EspExiting still forwards) and that a live exit following
a replayed one completes normally.
Agent V2.Core 2364, DecisionCore 533, Backend 4287, all green.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Round-two review, both findings confirmed against the code.
A replayed exit was being confirmed as post-AccountSetup by reading the registry
at replay time. That read reflects NOW, not the event's time, so an agent that was
down across the DeviceSetup→AccountSetup transition would confirm the stale
intermediate exit and, once the apps settled, synthesise a completion while the
AccountSetup page was still up. Windows offers no way to tell the two apart from
the event itself — Shell-Core writes the identical description
CommercialOOBE_ESPProgress_Page_Exiting for both. So the deferred edge is now
armed by LIVE exits only (EspExitedEventArgs.IsBackfill): only for those is the
continuous-observation argument valid. A replayed exit keeps the single immediate
synthesis attempt it always had.
For that attempt to be worth anything it must see real data, so the replay moved
out of EspAndHelloHost.Start() into ReplayEspExitBackfill(), invoked from
ImeLogHost.onStateRestored — after LoadState() has restored the package states.
That is what makes "the agent came back long after the ESP finished" recoverable.
The five-minute window for non-reboot restarts was unsound: the scheduled task
carries a BootTrigger only, with no restart-on-failure, so after an
exception_crash the agent does not return until the next boot — hours later, long
after Windows wrote the final exit. The window is now measured rather than
guessed: back to the last moment the previous run is known to have been alive,
taking the wider of the snapshot mtime (rewritten every decision step, available
for every exit type) and LastBootUtc. The five-minute default is the floor,
ClampLookbackMinutes the ceiling, and a future timestamp from clock skew is
ignored instead of producing a negative window.
Also switched the backfill tests to the real Shell-Core wording — the invented
string matched the regex but hid the fact that both exits look identical.
9 new tests. Agent V2.Core 2362, DecisionCore 533, Backend 4287, all green.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Follow-up to the review hardening: the replay ran on EVERY agent start with a
5-minute window, first run included. On a first run there is no gap to recover —
no earlier agent process existed — and a 62407 already sitting in the Shell-Core
log belongs to a phase transition this agent was never meant to observe. Replaying
it pushes an EspExiting into the reducer that the pre-fix agent never saw, i.e. a
happy-path behaviour change for no gain.
ResolveEspExitBackfillLookbackMinutes now keys the window to the previous exit
type: first_run (or unknown) disables the replay entirely, clean/hard_kill/
exception_crash keep the 5-minute default because the process is back within
seconds, and reboot_kill widens to "since LastBootUtc" because that gap spans the
forced restart until the post-reboot logon. The result is still clamped to
BackfillLookbackMaxMinutes, so policy and backstop together cannot build an
unbounded event-log query.
With this the happy path is byte-identical to pre-fix behaviour: on a first run
nothing replays, no edge is remembered, and the deferred re-check returns on a
volatile bool read.
No other backfill is affected — the Hello, MDM-reboot, Windows-Update and
ModernDeployment trackers each own a private backfill invoked from their own
Start(), none of which this change touches.
13 policy tests including clock skew across the reboot, a missing boot timestamp
and the clamp backstop. Agent V2.Core 2353, DecisionCore 533, Backend 4287.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Three real regressions, all confirmed against the code before touching it — two of
them introduced by the recovery itself.
The remembered exit edge was unqualified. Shell-Core raises 62407 at every ESP
phase transition, so OnEspExited also latched the DeviceSetup→AccountSetup one;
the deferred re-check would then open the strong AccountSetup gate as soon as the
last user app settled, while the AccountSetup page was still up and its other
subcategories still running. IsConfirmedPostAccountSetupExit() now demands
positive evidence (AccountSetup activity observed, or SkipUser == true) and is
deliberately stricter than the existing IsIntermediateDeviceEspExit() forward
guard: an unknown SkipUser counts as not confirmed. Erring strict costs the
pre-fix stall; erring loose costs a premature Succeeded. The immediate on-edge
synthesis keeps its previous behaviour verbatim.
The replay picked the oldest exit, not the final one. The reader walks
oldest-first and the exit branch is fire-once, so the widened downtime window —
which routinely holds both the intermediate and the final exit — handed over the
intermediate one and the fire-once guard swallowed the exit that matters.
ReplayBackfillRecords buffers the batch and lets only the newest exit-matching
record through; ESP failures and the Hello-wizard rail keep replaying
chronologically.
The replay ran before the IME state restore. ImeLogTracker.Start() restores the
persisted package states via LoadState() without raising OnAppStateChanged, and
ImeLogHost starts after EspAndHelloHost, so the replay evaluated an empty app
list — leaving the sharpest case unfixed: apps already terminal before the
reboot, only the final exit lost to the downtime. ImeLogHost gained an
onStateRestored callback wired to the re-check, and the host order is now pinned
by DefaultComponentFactoryOrderingTests so a reshuffle fails loudly.
11 new tests. Agent V2.Core 2340, DecisionCore 533, Backend 4287, all green.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Seven sits-d Cloud PCs enrolled the same day with identical configuration
(~138 required apps at the user-ESP gate, DeviceGuard/DmaGuard device-assigned).
The two that finished AccountSetup before the coalesced reboot landed succeeded;
the five the reboot killed in Stage=EspAccountSetup never derived a completion
again and ended Incomplete on healthy devices. Two defects behind that:
ShellCoreTracker.BackfillRecentEspExitEvents() has existed since session
772fe502 but no caller ever invoked it, so the Shell-Core watcher only saw
records written after Start(). Every agent restart silently dropped an ESP exit
(62407) or Hello-wizard start (62404) from its downtime. EspAndHelloHost.Start()
now calls it, and the fixed 5-minute window is sized to the actual downtime when
the previous run died to a reboot (LastBootUtc, clamped to 360 min).
The user-apps-settled synthesis ran only on the EspExited edge. When the ESP page
exits with apps still in flight the single attempt misses and nothing re-checks —
8110e262 and a89aac2d both reached 138/138 apps, 0 failed, and still never
completed. The exit edge is now recorded and the synthesis re-evaluated from
ImeLogTracker.OnAppStateChanged. No new completion path: the gate conditions are
unchanged, this only gives the existing conservative check a second look. The
fire-once claim moves to Interlocked (edge and re-check run on different threads)
and the re-check does not re-emit app_install_starved.
Rules: ANALYZE-ESP-005 gains on_event:session_stalled so a hung session produces
a finding at the 60-minute stall instead of staying finding-less for hours
(verified against 8110e262 — fired, confidence 95). New ANALYZE-APP-016 turns the
already-emitted app_install_starved events into a customer finding and, unlike
ANALYZE-APP-015, fires even on a succeeded session: 3b9291aa was recorded
Succeeded with 99 required apps that never started installing.
ANALYZE-APP-015 and ANALYZE-ESP-001 were considered for the stall trigger and
deliberately left alone — APP-015 is the documented worked example of the
terminal-precondition trap, and ESP-001's phase_duration gate on a still-open
phase is unverified mid-run.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The first CI run overwrote all four customer-facing scripts on the download
host and the legacy mirror with their own byte values as decimal text
(Install-AutopilotMonitor.ps1 went from 27331 to 95679 bytes).
PowerShell unrolls an array returned from a function into the output stream, so
the caller receives Object[] rather than byte[]. Invoke-RestMethod serialises an
Object[] body as space-separated decimals. Every hash check stayed green because
[byte[]] parameter coercion repairs the type everywhere except on the wire.
build-agent.yml's Send-Blob was never affected: it read the bytes inside the
function. Proven with a local HttpListener: the old shape puts 50 bytes starting
54,48,32,51,53 ("60 35 13 1...") on the wire where 17 bytes starting 60,35,13,10
("<#\r\n.SYN") are expected.
Fixes:
- ',' on both byte-array returns, plus an explicit [byte[]] at the upload site.
- Read each blob back from storage right after its PUT and abort before the
mirror and the remaining files. Verifying only at the end left every blob
corrupted.
- The version-bump guard tolerates an unparseable published copy with a warning
instead of throwing: that is the state a republish exists to repair, and a
guard that blocks the repair is worse than no guard.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Byte comparison made the guard fire on comment fixes, typos in comments and
reflow -- changes no consumer of the version can act on, so forcing a bump for
them is friction without a payoff.
The guard now compares a code fingerprint: the PowerShell token stream with
comments, newlines and line continuations removed. A behaviour change under an
unchanged $ScriptVersion still fails; comment and formatting changes publish
with a warning. That is also the rule the source file's own comment on
$ScriptVersion already states ("bump on meaningful changes").
Tokenising rather than regex-stripping keeps a '#' inside a string literal from
being mistaken for a comment.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The scripts behind download.autopilotmonitor.com/agent/ rotated only on a
publish_as_stable run of build-agent.yml, so a bootstrap fix sat unpublished
until the next agent cutover -- while the WDP bootstrap MSI downloads the
published copy on every enrolling device.
Publish-BootstrapScripts.ps1 becomes the single owner of the chain: dev render
with anchor check, $ScriptVersion bump guard, uploads (eu + fail-soft legacy
mirror, no-cache), the two version oracles, and verification through the alias.
publish-scripts.yml calls it on every push to main touching the sources; the
publish_as_stable branch of build-agent.yml calls the same code, so a release
still lands a consistent set.
Blobs are published in a deterministic CRLF/UTF-8-without-BOM form. Git
normalises line endings on checkout, so raw working-copy bytes would differ per
environment and make every byte comparison -- guard and verify alike -- depend
on where the workflow ran. The form reproduces the currently published bytes
exactly.
The Pester (PS 5.1) and ASCII gates move into the reusable
bootstrap-script-gates.yml and now block the upload; pester.yml keeps the
pull-request view and drops its push trigger. AdminConfiguration
.LatestBootstrapV2ScriptVersion gets exactly one writer.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Codex findings on d8c84f41/a369ff99: (1) MarkUploaded now recomputes
_pendingBytes from the pending list when accounting drifts negative or
stays non-zero on an empty list — a silent negative would disarm the
spool-pressure trigger. (2) New test covers the DevicePrepCompletion
handler''s AwaitingDesktop fallback (stale/replayed timer without a
desktop fact): synthetic Hello timeout, park, and completion on the
later real desktop arrival. Third finding (HashingCappedReadStream
maxBytes:0) refuted — the 0=uncapped contract is implemented in the
same class and exercised by the heal-path tests.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
99/100 spool_pressure_detected events in the field were file-size-only
triggers on a near-empty backlog (append-only spool.jsonl grows past 5 MB
in healthy long sessions). TelemetrySpool now tracks PendingBytes (exact
serialized size of the not-yet-uploaded tail, maintained on enqueue /
mark-uploaded / restart rehydrate) and the collector thresholds on that
instead of SpoolFileSizeBytes, which stays as an informational metric.
Snapshot gains spool_pending_bytes; pressure event data carries
pendingBytes + pendingBytesThreshold.
Also: startup recovery scan streams via File.ReadLines instead of
materializing the whole spool with File.ReadAllLines (corrupt-tail
break semantics unchanged).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Hero gets a quiet checkmark chip row (user-driven, pre-provisioning,
self-deploying, Device Preparation, Windows 365, Entra ID & Hybrid join);
about page deployment claim now includes the MSI LOB channel for WDP.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver Kieselbach