Refresh durable investigation lifecycle from authoritative actions while
preserving completed evidence. Keep resolved history reviewable and label
recorded plan facts separately from action outcomes. Follow all resource
pages during qualification and record live approval, rejection and storage
semantic-review results. Integrate current main and preserve its alert
ordering correction.
Live funded qualification found hidden tool results and misleading action
submission outcomes. Share the result-bearing transcript across stored chat
and product history, render the retained evidence, and distinguish captured
proposals from broker acceptance. Keep review usable while Patrol is paused.
Record Gemini route pricing and exact qualification limits. Integrate current
main and repeat browser proof for the incoming login flow. Approved/rejected
recovery remains unqualified without the development command agent.
Exercise the production collector through healthy, full and recovered
storage states using the existing bounded disposable lab. Preserve exact
mount configuration while keeping collector proof separate from model
diagnosis and installed-agent qualification.
The fleet sampler and coordinator had no production alert trigger and could
repeat cached values as fresh incident evidence. Preserve saved recordings
through explicit read-only lookups, propagate read failures and report the
former live incident count as unmeasured. Keep historical status and duration
units explicit without rewriting archived observations.
Integrate main's alert dispatch wording and startup replay qualification.
Canonical incident listing and real-model outcome qualification remain open.
Preserve both monitoring contracts and combine the diagnostic history
correction with current host continuity and delivery evidence changes.
Verify the combined backend, frontend and browser behaviour before
landing the existing diagnostic work.
Keep alert and inventory evidence together after container removal and
restart without rewriting retained events or transferring approval and
operator authority. Resolve exact source identities in the shared store
and preserve event replay idempotency across old and current records.
Verify actual retained homelab events, registered tool reads, lifecycle
callbacks, tenant isolation, race behaviour and responsive evidence views.
Assistant incident reads used an unconnected cached-metric recorder while
resource history already retained operational events. Read the shared
organization-scoped timeline with original provenance, bounded results
and explicit coverage and failure semantics. Keep legacy archive reads
resource-bound.
Record the separately reproduced alert identity split at the shared
write boundary. This read-path correction does not qualify complete
incident diagnosis or recovery.
Docker can report tmpfs mounts only in HostConfig.Tmpfs. Preserve those
entries in shared inventory and retain type, options and canonical write
access in diagnostic queries. Configured size is not measured free space.
Record the failed ordinary storage diagnosis and independently verified
recovery without claiming autonomous or installed-collector qualification.
Name command transport explicitly in shared query results and preserve
unobserved connection state in topology and Assistant inventory context.
Keep parent-node transport distinct from a direct guest connection so
monitoring evidence cannot imply command access or collection downtime.
Existing execution policy and approval checks remain authoritative.
Bind the ordinary read-only storage assessment to the corrected runtime and
persisted tool evidence. Keep useful capacity observations separate from
unsupported causal and temporal claims so completion is not counted as a
qualified diagnosis or autonomous outcome.
Native container config reads confused missing session discovery with a
missing resource and could reuse stale placement. Read identity and
capability from canonical inventory while preserving explicit query
restrictions and existing action authority.
Keep unavailable providers, unsupported adapters and empty observations
distinct from resource absence, with the tool error bit preserved.
Missing block I/O and container image sizes could become false evidence
for diagnosis. Preserve per-direction counter presence and measured zero
through collection, resource conversion and browser rendering. Separate
new observed history from ambiguous retained disk series without deleting
old rows or changing public metric names.
Keep partial host rates distinct and persist a newly enabled Disk I/O
column across the first preference reload.
Concurrent calls with the same tool name could collapse pending activity or
remove a sibling approval. Deep transcript reconciliation also mutated shared
tool objects when status rows disappeared, replacing earlier evidence with a
later result. Match supplied invocation IDs strictly and key immutable message
rows without reconciling their nested evidence.
Add regressions for both causes and record desktop, intermediate and narrow
browser replay of all 23 captured tool results. Record the healthy and dependency
model qualification limits separately from renderer and fixture-cleanup proof.
Route labels classify ASCII digits and hexadecimal UUID bytes. Scan those
bytes directly while preserving numeric precedence, Unicode names and
invalid UTF-8 handling. This reduces the shared normaliser overhead exposed
by paired landing benchmarks without changing the benchmark gate.
Refs #1928
Contract-Neutral: ASCII route-label optimization preserves label values, identifier precedence, Unicode and invalid UTF-8 behavior, and agent lifecycle authority.
Preserve the real Assistant result separately from passing access routing
and disposable fault oracles. Correct tool evidence still produced
unsupported temporal claims, so customer outcome readiness remains open.
Refs #1928
Validate the checked-in dependency and three service restart fault contracts
against disposable Docker resources before using them to assess model output.
Verify baseline, injected fault, refused duplicate injection, explicit fixture
recovery, and two-pass cleanup with unchanged pre-existing inventory.
These opt-in tests make no Pulse or model request. Fixture recovery is teardown
and does not count as an approval, rejection, execution or customer outcome.
Record exact live, owning-package and source-bound proof in the redesign plan.
Resolve monitored topology before checking command connections so unavailable
inspection retains the known resource and parent node. Prevent known targets
from falling through to a colliding agent ID, and preserve the single-agent
requirement when no target is supplied.
Use one failed tool envelope for diagnostic reads and file mutations. Missing
connections neither prove an installation problem nor count as successful
writes. Hypervisor lifecycle authority remains on its canonical action path.
Verify disconnected and unknown targets, collision isolation, all affected
tool handlers, token/WebSocket scope boundaries, and the linked Patrol and
Assistant failure journey at desktop and narrow widths.
The 500-node dashboard query triggered repeated ordinal string conversions
in the SQLite driver while matching numbered parameters. Use alphabetic
named bindings to preserve current values and shared query branches without
that allocation cost.
Cover large cached scopes with changed resource families, identities,
metric filters and windows, including IDs that resemble SQL syntax.
Refs #1928
A completed probe remained discoverable after its first caller returned,
allowing the next collection to reuse stale filesystem measurements.
Remove it and publish completion within one registry critical section.
Keep in-flight sharing, cancellation and timeout behaviour intact.
The controlled regression fails before this change. Twenty full package
runs and three race runs pass on the worker.
Refs #1928
Add a disposable service-storage fault with an independent filesystem
oracle, bounded tmpfs writes, identity checks and verified recovery.
Exercise overwrite and symlink refusal without contacting a model.
Align the published schema with supported summary-term groups and validate
the complete catalogue in CI. Record the exact proof and remaining model
and missing-access qualification limits in the customer-journey plan.
Canonical tier reconciliation rebuilt SQL and probed absent preferred tiers
for each fallback point, regressing batch reads and allocation costs. Reuse
bounded query shapes with current bindings and snapshot-scoped absence
checks, then append consecutive points directly to their output series.
Preserve coverage and ordering semantics and verify fresh bindings after
new preferred observations arrive. Integrate current main test additions.
The runtime and three restart scenarios use health_process_stop, but the
published schema rejected it. Accept that implemented injector and check
actual catalogue fault types against the schema to prevent recurrence.
Record the remaining missing-access and storage qualification gaps.
Integrate the latest alert, delivery and action-result changes with the
Patrol evidence conversation. Replace the conflicted browser receipt
with current source-bound qualification and fix shared warning-card
wrapping exposed by the intermediate-width check.
Real diagnostic and autonomous action outcome qualification stays open.
Remove contextless evaluator and assessment passes, signal-count budgets,
and post-finding prompt replacement. Keep evidence tools available until
explicit run limits and retain incomplete assessments and provider errors
alongside accepted decisions. Failed file reads now preserve error status
through the model, telemetry and saved Assistant history.
Full chat, AI and tools packages, focused Patrol API and race tests pass.
Real read-only and scripted browser checks preserve failed reads and linked
uncertainty. Real-model and verified action outcome qualification remain open.
Execute plain retained reconciliation in one current SQLite snapshot and
reuse bounded compiled statements. Preserve per-series chronology without
a metric sort while keeping display aggregation ordering explicit.
Exact-base worker comparisons cover the prior PR benchmark failures. Full
metrics/database and focused concurrent race checks pass. Final CI and
real diagnostic outcome qualification remain open.
Tool-call totals do not establish diagnostic sufficiency. Preserve seed-only
and failed-read conclusions, remove count-based completion instructions from
evidence, and retain configured limits and authority checks.
Keep findings grouped under alerts selectable in the shared review panel so
their investigations and access limits remain available to Assistant.
Keep canonical disk risk, source freshness and retained history intact when
Assistant and Patrol gather evidence. Proposal acceptance validates an action
contract and must not rewrite uncertain conclusions as established root cause.
Preserve complete subscription tool batches without exposing routing envelopes
as answers. Keep wide answer tables readable and keyboard-scrollable on mobile.
Optimize retained tier reconciliation without discarding gaps or newer samples.
Record failed real-model diagnoses and outstanding autonomous qualification
separately from passing data-path and interface checks.
Keep overlapping success, failure and loading completion owned by the latest started read. Retain caller and queue-action controls, subsystem contracts and exact-source scripted Chromium evidence. Installed delivery and release qualification remain separate.
Change-source: pulse-maintainer
Non-empty aggregate tiers hid recent raw samples and missing metric series.
Unify single and batch reads with indexed overlap resolution before
downsampling, and preserve recorded extrema through subsequent rollups.
Use canonical Proxmox storage coordinates in summaries and node history.
Discovery routing does not establish an installed Agent or agent history.
Record live evidence freshness and the remaining diagnosis qualification gaps.
A provider policy refusal was classified as a connection failure, while
summary tools mixed high-utilisation heuristics with unchecked health claims.
Preserve explicit refusals before tool recovery and provide retained metrics
with source scope, observation timestamps and bucket extrema for diagnosis.
Record the live qualification limits and the shared temporal tier-query gap.
Real homelab investigations lost metrics behind mismatched resource IDs,
queried only post-restart memory, and filtered every physical disk out.
Resolve canonical metrics targets, use retained history, preserve CPU
topology and chronological samples, and separate disk operation and format.
Keep unattached resource shortcuts blocked while allowing named resources
to be resolved through query search. Report incomplete readiness honestly,
including saved evaluations, without granting Patrol suitability.
Real homelab evaluation exposed missing node sensor observations and loss of
recent tool evidence before the model reached its conclusion. Project both
canonical sensor sources and retain observations while context fits.
Add source and selection regressions plus a multi-turn retention proof.
Record the live diagnosis result and the unresolved Patrol provider block.
Explain actions previously opened a blank conversation and discarded richer
finding context. Dispatch the selected explanation through shared chat
handling, retain evidence and drafts, and cancel pending work on tenant
switches. Keep unrelated workflow starters out of scoped conversations.
Record the remaining real-model and customer-outcome qualification gap
without treating scripted browser responses as proof of product value.
Installed PBS qualification must distinguish accurate metric history from API load and persistence write traffic. File growth and whole-device counters alone cannot attribute write cost to Pulse, so an unqualified measurement could lead to incorrect regression or wear claims.
Document matched baseline and repaired measurement windows, counter limitations and backup I/O separation without changing runtime behaviour or claiming installed recovery.
Validation: git diff --check passed; documentation-only change. Installed artifact qualification remains dependent on an authorised environment.
Change-source: pulse-maintainer
Aggregate historical row counts across unlike resources do not establish whether the integrated PBS source-time repair works on an installed artifact. Release qualification needs common observation windows and independent source timestamps to distinguish valid unchanged-value samples from restamped cached data.
Document an operator-run check covering mixed polling, outage, recovery and whole-process restart, with exact artifact identity and per-metric SQL measurements. This provides acceptance criteria, not installed proof or grounds to close#1882.
Validation: documented SQL previously exercised against synthetic SQLite healthy-cadence, duplicate-identity and empty-window cases; git diff --check passes. Documentation only; no production behaviour changed.
Change-source: pulse-maintainer
rcourtman