Skip to content
CSA Loom — the Microsoft Fabric experience for Azure tenants where Fabric isn't yet available: lakehouses, warehouses, notebooks, semantic models, Activator rules, Data Agents, across Commercial, GCC, GCC-High, and DoD IL5

health-check — parity with health/observability check (Foundry Health Checks / Azure Monitor)

Source UI: - Palantir Foundry Health Checks / Data Health — https://www.palantir.com/docs/foundry/health-checks/overview, https://www.palantir.com/docs/foundry/health-checks/checks-reference, https://www.palantir.com/docs/foundry/observability/data-health, https://www.palantir.com/docs/foundry/monitoring-views/overview, https://www.palantir.com/docs/foundry/health-checks/notifications - Azure Monitor log-search alert rule editor — https://learn.microsoft.com/azure/azure-monitor/alerts/alerts-create-log-alert-rule, https://learn.microsoft.com/azure/azure-monitor/alerts/alerts-types#log-alerts, https://learn.microsoft.com/azure/azure-monitor/alerts/action-groups

Azure-native default backend (no Microsoft Fabric / no Reflex on the default path, per .claude/rules/no-fabric-dependency.md): Azure Monitor scheduledQueryRules over Log Analytics / ADX, action groups (email/SMS/webhook/Logic App) for notification, alert history (Alerts / alertsmanagement) for run/fired history, and listResourceHealth / fetchMetrics for the status dashboard. Fabric Reflex remains opt-in only via LOOM_ACTIVATOR_BACKEND=fabric.

Last verified: 2026-07-01 against current code (apps/fiab-console/lib/editors/palantir-editors.tsx health-check editor + app/api/items/health-check/[id]/{rule,rule/[ruleId],run,history}). The editor is now a 4-tab shell (Checks · Status · History · Settings — no Notifications tab yet); the status dashboard (G), fired-alert history (H), and per-rule lifecycle (K) have shipped and flip ❌→✅ below.

Real feature inventory

A. Check definition library (Foundry checks-reference — ~30 check types in 5 families)

  1. Status checks — Schedule status, Build status, Job status, Sync status (each: Severity, Escalate, Notes, Issues toggle; Sync adds a destination).
  2. Time checks — Data freshness (column + freshness range), Sync freshness, Build duration, Sync duration, Time since last updated (+ ignore-empty-transactions, custom schedule), Time since sync last updated. All support Median deviation (MAD → approx σ over N recent builds).
  3. Size checks — Dataset file count, Dataset partition (auto logic), Row count (operators Between/≥/≤/=, compare-to-last-successful mode), Transaction file count, Transaction file size.
  4. Content checks — Allowed column values, Approximate unique %, Column regex, Approximate column relation, Date range, Null %, Numeric mean, Numeric median, Numeric range, Primary key.
  5. Schema checks — Column (present + type), Column count, Schema (EXACT_MATCH_ORDERED / _UNORDERED / COLUMN_ADDITIONS_ALLOWED / _STRICT). Each check exposes: Severity (Moderate / Critical), Escalate toggle, Notes, Issues (auto-create-issue-on-fail) toggle, optional Median deviation and Schedule (Automatic / Custom).

B. Target / scope picker

  • Foundry checks bind to a dataset / pipeline / schedule / job / sync destination.
  • Azure Monitor binds a rule to a target resource scope (single resource, Log Analytics workspace, resource group, or subscription) + a table / KQL query, plus Split by alert dimensions.

C. Alert logic / threshold editor (Azure Monitor)

  • Measurement (table/result count vs metric measurement), Operator (>, <, between, dynamic upper/lower), Static or Dynamic threshold (+ sensitivity High/Med/Low), Aggregation granularity / Period (window), Frequency of evaluation (1 min – 24 h), number of breaches / consecutive breaches, KQL preview with sample data.

D. Schedule

  • Evaluation frequency + look-back window; Foundry "Automatic vs Custom Schedule"; Azure "Enable upon creation", stateful / auto-resolve alerts.

E. Severity & escalation

  • Azure severity 0–4; Foundry Moderate/Critical + Escalate. Warning→Critical escalation ladder.

F. Actions / notifications (action groups)

  • Email, SMS, Voice, Push, Webhook, Logic App, Azure Function, ITSM; Foundry adds Slack / PagerDuty / email + in-platform watcher notifications. Custom email subject, custom webhook JSON payload, test notification.

G. Run / status dashboard

  • Foundry Data Health app: per-dataset health status (green/yellow/red), Data-Lineage coloring, historical health timeline, filter/sort by status. Azure: Alerts blade — fired/resolved counts by severity, alert timeline.

H. Run history / fired-alert log

  • Per-check history of pass/fail evaluations; Azure fired-alert history (fired time, resolved time, severity, fired value).

I. Issues / incident management

  • Auto-create + auto-close Issues on fail/recover; integrate to external incident systems; watcher assignment & severity.

J. Monitoring Views (scale)

  • Collections of checks/rules that auto-apply to resources matching a scope as new resources are added (recommended over ad-hoc check groups).

K. Per-rule lifecycle actions

  • Enable / disable, edit, delete, test-fire / trigger, view in portal, duplicate.

L. Remediation

  • Suggested fix on breach; link to runbook / Logic App auto-remediation; re-run pipeline.

Loom coverage (built ✅ / honest-gate ⚠️ / MISSING ❌)

# Capability Status Backend today
A Check-type library (5 families, ~30 types) ❌ — only freshness / rowcount / custom KQL buildQuery() in [id]/rule/route.ts
B Target / scope picker ❌ — free-text "Table" input only none
C Threshold/operator + dynamic threshold + preview ⚠️ — static threshold value inputs built (thresholdMinutes/minRows); operator hard-coded, no dynamic threshold, no KQL preview ScheduledQueryRuleInput
D Schedule (freq + window) ✅ — evalFreq + windowSize dropdowns createMonitorActivatorRule({evaluationFrequency, windowSize})
E Severity & escalation ⚠️ — severity Dropdown (0–4) wired + rendered as badge; escalation chains missing createMonitorActivatorRule({severity})
F Action groups / channels + test-fire ❌ — single email input only; upsertActionGroup/sendActionGroupTestNotification exist but UI never calls them (clients unused)
G Status dashboard ✅ — Status tab, 4 stat tiles (Total/Healthy/Firing/Disabled) from rule state + fired alerts getActivatorHistory + listScheduledQueryRules
H Run / fired-alert history ✅ — History tab Table of fired/resolved events (time, severity, state) getActivatorHistory ([id]/history)
I Issues / incident management ❌ none
J Monitoring Views (scoped groups) ❌ listScheduledQueryRules exists
K Per-rule enable/disable/delete/test ✅ — row Run/Enable/Disable/Delete buttons wired triggerMonitorActivatorRule ([id]/run) + enable/disable/deleteMonitorActivatorRule ([id]/rule/[ruleId])
L Remediation ❌ Logic App receiver in action groups

Current grade: C (functional but incomplete) — a real 4-tab shell with an end-to-end check lifecycle (create rule, per-rule Run/Enable/Disable/Delete, a derived status dashboard, and fired-alert history), all wired to real Azure Monitor scheduledQueryRules. The "ignored clients" critique is now accurate only for action groups (upsertActionGroup / sendActionGroupTestNotification), listResourceHealth, and fetchMetrics. Still missing: the Foundry check-type library, scope picker, operator/dynamic-threshold/KQL-preview, action-group channel management + test-fire, a Notifications tab, and incident / Monitoring-Views / remediation.

Build plan (prioritized)

P0 — make it a real monitor, not a form

  1. Tabbed editor shell (ItemEditorChrome + Fluent TabList): Checks · Status · History · Notifications · Settings. Replaces the single scroll form.
  2. Status dashboard tab — TileGrid of stat tiles (Healthy / Warning / Critical / Disabled counts) from listAlertHistory + per-rule current state; severity-colored Badges; EmptyState when no checks. Backend: listAlertHistory + listScheduledQueryRules.
  3. Per-rule lifecycle actions — row Menu (Enable, Disable, Delete, Test-fire, Open in portal) wired to patchScheduledQueryRule / deleteScheduledQueryRule / triggerMonitorActivatorRule. New routes PATCH/DELETE on /api/items/health-check/[id]/rule/[ruleId].
  4. Run / fired-alert history tab — DataGrid of fired/resolved events (time, severity, fired value, state) from getActivatorHistory / listAlertHistory.

P1 — check library + notifications + severity

  1. Check-type gallery — TileGrid of check categories (Status / Time / Size / Content / Schema) each opening a typed wizard Dialog; build at least Time (freshness, time-since-updated, build/sync duration) + Size (row count, file count) + Content (null %, regex, allowed values, numeric range) against KQL templates. Extend buildQuery() per type.
  2. Severity + escalation controls — Dropdown severity 0–4 (Moderate/Critical mapped), Escalate Switch, Notes Textarea, "auto-resolve / stateful" Switch. Wire severity + autoMitigate into ScheduledQueryRuleInput.
  3. Notifications tab (action groups) — manage channels: Email / SMS / Webhook / Logic App rows via ActionGroupInput; Test button → sendActionGroupTestNotification; custom email subject. Backend: upsertActionGroup / listActionGroups.
  4. Threshold/operator editor — operator Dropdown (>, <, between, =), static/dynamic threshold toggle + sensitivity, KQL preview running the built query via queryLogs against the look-back window.

P2 — scale + remediation + issues

  1. Monitoring Views — scope-based group (resource-type / tag filter) that bulk-creates rules; persisted on state.views; list + apply.
  2. Issues / incident panel — auto-open an audit-log incident on fail; show open/closed issues; assign watcher. Backend: existing audit-log + Cosmos state.issues.
  3. Remediation — per-check optional Logic App receiver as auto-remediation; "Re-run pipeline" link to the bound data-pipeline item.
  4. Target/scope picker — Dropdown of Loom items / Log Analytics tables / ADX clusters as the rule scope, replacing the free-text Table input.

Per web3-ui.md / loom_design_standards: all surfaces use Loom tokens, TileGrid for the gallery & status tiles, EmptyState for empty panes, Fluent icons per section, elevated cards, severity Badges — no freeform config beyond the KQL expression box (allowed, 1:1 with Azure's query editor).