Files
fusion/packages/core/src/team-analytics.ts
gsxdsm c15c78feeb feat: migrate storage from SQLite to PostgreSQL (#1793)
# Migrate storage from SQLite to PostgreSQL — full dashboard cutover

Migrates Fusion's storage layer to the embedded PostgreSQL
`AsyncDataLayer` (the default backend) and **completes the
satellite-store + feature cutover** so every dashboard and Command
Center surface works in PG mode.

## Status — every surface works in embedded-PG mode

Verified live against a running embedded-Postgres dashboard (all
**200**, zero 5xx) and gate-tested (**23 files / 99 tests** on embedded
PG, plus engine-core 294 and ci-shape 63 in the blocking merge gate;
core/engine/cli/dashboard typecheck clean).

| Area | Surfaces | State |
|---|---|---|
| Satellite stores | workflows, todos, insights, research, missions,
goals, mailbox | ✅ |
| Views | artifacts, documents, evals | ✅ |
| Command Center | activity, productivity, team, tokens, tools,
**workflows**, **github**, **signals**, **plugin-activations**, **live**
(all 10) | ✅ |
| Run execution | insight generation, research run execution | ✅
(store-path; AI step needs a provider) |
| Live updates | SSE push for mission/research/insight events | ✅ |
| Workflow editing | create / update / delete / select (+ id counter) |
✅ |
| Engine | mission autopilot, incident-signal ingestion, regression
storm-guard, agent wake-on-message | ✅ |
| Core | tasks, agents, secrets, automations, memory, chat, usage, PRs,
git | ✅ |

## Approach

Each satellite store gets an `Async<Store>` wrapper exposing the sync
store's method names over the existing `async-*-store.ts` helpers;
`get<Store>Store()` returns a `Sync | Async` union; consumers `await`
(harmless on sync), and engine/CLI paths that can't convert use
`instanceof Sync` graceful fallback. Analytics aggregators branch on
`"ping" in dbOrLayer` to run schema-qualified raw SQL over `project.*`
(snake_case) in PG. Executors/orchestrators/autopilot are
await-converted to drive the union store; the async store wrappers
extend `EventEmitter` so SSE live-push fires in both backends.

Not-yet-ported capabilities degrade gracefully (never 500) and are
individually called out in commits.

## Sync with main

The branch is kept continuously merged with `main` (currently through
FN-7845, 2026-07-12); the earlier "final rebase deferred" note no longer
applies. Use **Create a merge commit** (or squash) to land it — GitHub's
rebase-merge cannot replay a merge-maintained branch.

## Residual Review Findings

Multi-agent code review of the PostgreSQL satellite-store ports (U1–U5)
applied 3 safe fixes (see `fix(review): apply autofix feedback`). The
following are **real but gated** — recorded here as follow-up work
rather than auto-applied. All are SQLite→PostgreSQL
**concurrency/atomicity regressions**: the sync stores were immune only
by SQLite's single-writer, single-threaded-handler execution; the async
ports open multi-await read-modify-write windows. **Reachability is low
today** because the execution engines that generate concurrent same-run
mutations (insight run executor, research orchestrator/dispatcher) are
`instanceof`-gated to sync mode in PG. No process-crash class survived
(all engine fallbacks correctly guard the sync store).

- **[P1] Research `appendResearchEvent` dual-write is non-atomic**
(`packages/core/src/async-research-store.ts`, corroborated: adversarial
+ reliability). The `research_run_events` insert (own transaction) and
the `run.events` jsonb update are separate writes — a crash between
them, or two concurrent appends, splits the table count from the jsonb
array. **Fix:** perform the seq-insert and the jsonb update in one
`layer.transactionImmediate`.
- **[P1] Research run terminal-reversion via stale full-row persist**
(`async-research-store.ts` `persistResearchRun`/`updateResearchStatus`).
Concurrent `PATCH /runs/:id/status` + `POST /runs/:id/events` can revert
a terminal run to `running` by overwriting the whole row, bypassing the
transition guard. **Fix:** scoped column `UPDATE`s with a `WHERE status
…` guard, or optimistic version column.
- **[P2] `updateResearchRun`/`updateInsightRun` read-then-write TOCTOU**
— concurrent PATCHes last-writer-wins on the lifecycle merge. **Fix:**
`SELECT … FOR UPDATE` / enclosing transaction.
- **[P2] `upsertRun`/`createRunOrThrowConflict` check-then-create race**
(`async-insight-store.ts`) — two callers can each create an "active"
run. **Fix:** partial unique index on `(projectId, trigger) WHERE status
IN ('pending','running')`.
- **[P3] `createResearchRetryRun` return-value divergence** — sync
returns the pre-update `queued` snapshot; async returns the reloaded
`retry_waiting` run (persisted state is identical). Pick one side for
cross-backend parity.
- **[P2/perf] Mission `getMissionWithHierarchy`/`getMissionHealth` N+1
fan-out** — O(milestones×slices) sequential round-trips hold one pool
slot per request; can starve the pool for large hierarchies. **Fix:**
batched/joined reads.
- **Testing gaps:** no PG-mode concurrency tests (interleaved
status/event mutations), no sync↔async parity assertion for the
lifecycle-error codes, and no mission status/health rollup parity test
vs the sync `MissionStore`.

~~Out of scope (deferred): AI run *execution* (insight/research) +
mission autopilot + live SSE mission events remain sync-gated/degraded
in PG mode.~~ **Since ported** — insight/research run execution, mission
autopilot, and SSE live push all run on the async layer now, which also
makes the concurrency findings above genuinely reachable; they remain
open follow-ups.







---

## Update — 2026-07-12: production-readiness hardening & live acceptance

Everything below landed on this branch since the description above was
written:

**Production blockers from review — fixed**
- `recoverStaleTransitionPending` ported to the async layer (backend
moves write + clear the crash-safe marker; startup/maintenance sweeps no
longer throw).
- Lost-update class fixed: `atomicWriteTaskJson`/`WithAudit` write
changed columns only (full-row upserts silently resurrected stale fields
across concurrent store instances — the "task stuck unplanned forever"
bug).
- First-boot **auto-migration**: booting the PG backend over a project
with a legacy `fusion.db` migrates it automatically (loud failure,
SQLite kept as backup), and the dashboard shows a one-time **"your data
was migrated" banner** with the backup paths and a Need-help Discord
link.
- `pg_dump`/`pg_restore` discovered from common install locations for
embedded-mode backups.
- The PG suite is part of the blocking merge gate (`test:pg-gate`).

**Multi-project isolation (PR #2007, merged into this branch)**
- `project_id` partition key on tasks / archived tasks / config,
`taskProjectScope` threaded through every scan/claim/count, per-project
config rows, layer bound to the project at startup.
- Review P1 follow-up: the shared cold-storage `archive.archived_tasks`
table is also partitioned and all archived-board reads/counts/searches
are scoped.
- Schema drift self-heal generalized to schema-qualified columns so
existing databases upgrade in place.

**Other changes**
- Node settings sync **removed** in PG mode (409
`settings-sync-disabled-postgres`) — nodes share state by connecting to
the same database; auth sync kept (per-machine file).
- Perf (review findings): `listTasks` pushes column filter + ORDER BY +
LIMIT/OFFSET into SQL; `getConversation` capped to the most recent 200
messages.
- Fixed a false "operator action required" pause-abort log fired on
every successfully auto-merged task.

**Live acceptance — PASSED (2026-07-12)**
A sandboxed instance (isolated HOME, embedded PG, real Opus executor)
ran a task through the complete cycle: create → triage (AI spec) →
execute → in-review → AI squash-merge landed on the project's `main` →
done. A write+read sweep of every data surface (settings, comments,
documents, attachments + artifact bridge + artifact edit, chat with real
generation, goals, missions, agent mail, secrets, workflows, memory, CC
analytics) was green on embedded PG.

**Known remaining work**
- The per-project `config` PK re-key has no upgrade path for
pre-isolation embedded-PG databases (needs a real `DROP
CONSTRAINT`/re-key migration; fresh databases are fine).
- `pg_dump`/`pg_restore` binaries are not yet bundled in release
artifacts (PATH/common-location discovery only).
- The satellite-store concurrency findings listed above.

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Phil Larson <hello@phillarson.xyz>
Co-authored-by: fusion-merge <fusion-merge@local>
2026-07-13 19:07:58 -07:00

447 lines
15 KiB
TypeScript

import { sql } from "drizzle-orm";
import type { Database } from "./db.js";
import type { AsyncDataLayer } from "./postgres/data-layer.js";
import { costFor, type CostResult, type ModelPricingOverrides } from "./model-pricing.js";
import type { TokenTotals } from "./token-analytics.js";
export interface TeamAnalyticsQuery {
/** ISO-8601 lower bound (inclusive). */
from?: string;
/** ISO-8601 upper bound (inclusive). */
to?: string;
/** Epoch ms "now" used only for pricing-staleness. */
now?: number;
/** User-managed pricing overrides that take precedence over the built-in baseline. */
pricingOverrides?: ModelPricingOverrides;
}
export interface TeamMetricTotals {
tokens: TokenTotals;
cost: CostResult;
filesChanged: number;
tasksCompleted: number;
tasksInProgress: number;
tasksInReview: number;
}
export interface TeamAgentSummary extends TeamMetricTotals {
agentId: string;
agentName: string | null;
role: string | null;
state: string | null;
}
export interface TeamAnalytics {
from: string | null;
to: string | null;
totals: TeamMetricTotals;
agents: TeamAgentSummary[];
}
interface AgentRow {
id: string;
name: string | null;
role: string | null;
state: string | null;
}
interface TaskTokenRow {
agentId: string;
inputTokens: number | null;
outputTokens: number | null;
cachedTokens: number | null;
cacheWriteTokens: number | null;
totalTokens: number | null;
modelProvider: string | null;
modelId: string | null;
tokenUsageModelProvider: string | null;
tokenUsageModelId: string | null;
}
interface CountByAgentRow {
agentId: string;
count: number;
}
interface ModifiedFilesRow {
agentId: string;
modifiedFiles: string | null;
}
function emptyTokenTotals(): TokenTotals {
return {
inputTokens: 0,
outputTokens: 0,
cachedTokens: 0,
cacheWriteTokens: 0,
totalTokens: 0,
nTasks: 0,
};
}
interface CostAccumulator {
usd: number;
anyPriced: boolean;
anyUnavailable: boolean;
anyStale: boolean;
}
function emptyCostAccumulator(): CostAccumulator {
return { usd: 0, anyPriced: false, anyUnavailable: false, anyStale: false };
}
function finalizeCost(acc: CostAccumulator): CostResult {
return {
usd: acc.anyPriced ? acc.usd : null,
unavailable: acc.anyUnavailable,
stale: acc.anyStale,
};
}
function addTokenRow(totals: TokenTotals, row: TaskTokenRow): void {
totals.inputTokens += row.inputTokens ?? 0;
totals.outputTokens += row.outputTokens ?? 0;
totals.cachedTokens += row.cachedTokens ?? 0;
totals.cacheWriteTokens += row.cacheWriteTokens ?? 0;
totals.totalTokens +=
row.totalTokens ??
(row.inputTokens ?? 0) +
(row.outputTokens ?? 0) +
(row.cachedTokens ?? 0) +
(row.cacheWriteTokens ?? 0);
totals.nTasks += 1;
}
function addRowCost(
acc: CostAccumulator,
row: TaskTokenRow,
now?: number,
pricingOverrides?: ModelPricingOverrides,
): void {
const result = costFor(
{
inputTokens: row.inputTokens ?? 0,
outputTokens: row.outputTokens ?? 0,
cachedTokens: row.cachedTokens ?? 0,
cacheWriteTokens: row.cacheWriteTokens ?? 0,
},
{
/*
* FNXC:CommandCenter 2026-07-10-08:25:
* Current runtime usage rows persist the actually-used model in tokenUsageModelProvider/tokenUsageModelId while task-level modelProvider/modelId can stay empty. Team cost analytics must price the usage snapshot first so estimated cost survives model-resolution hierarchy and catalog drift instead of reverting to the unavailable sentinel.
*/
provider: row.tokenUsageModelProvider ?? row.modelProvider,
model: row.tokenUsageModelId ?? row.modelId,
},
now,
pricingOverrides,
);
if (result.stale) acc.anyStale = true;
if (result.unavailable || result.usd === null) {
acc.anyUnavailable = true;
} else {
acc.usd += result.usd;
acc.anyPriced = true;
}
}
function emptyMetricTotals(): TeamMetricTotals {
return {
tokens: emptyTokenTotals(),
cost: { usd: null, unavailable: false, stale: false },
filesChanged: 0,
tasksCompleted: 0,
tasksInProgress: 0,
tasksInReview: 0,
};
}
function countModifiedFiles(value: string | null): number {
if (!value) return 0;
let files: unknown;
try {
files = JSON.parse(value);
} catch {
return 0;
}
if (!Array.isArray(files)) return 0;
let count = 0;
for (const file of files) {
if (typeof file === "string" && file.length > 0) count += 1;
}
return count;
}
function addRangeClauses(column: string, clauses: string[], params: string[], query: TeamAnalyticsQuery): void {
if (query.from !== undefined) {
clauses.push(`${column} >= ?`);
params.push(query.from);
}
if (query.to !== undefined) {
clauses.push(`${column} <= ?`);
params.push(query.to);
}
}
function makeSummary(agentId: string, agent?: AgentRow): TeamAgentSummary {
return {
agentId,
agentName: agent?.name ?? null,
role: agent?.role ?? null,
state: agent?.state ?? null,
...emptyMetricTotals(),
};
}
/**
* Aggregate store-derived per-agent Command Center metrics over a date range.
*
* FNXC:CommandCenter 2026-06-18-16:57:
* Team analytics derives per-agent tokens/cost, files changed, and tasks completed from the tasks+agents tables only; no new schema, no GitHub-issue data (that is FN-6653). Keep the aggregator pure/read-only and project-scoped by accepting the already-scoped Database handle from the HTTP layer.
*/
export async function aggregateTeamAnalytics(
dbOrLayer: Database | AsyncDataLayer,
query: TeamAnalyticsQuery = {},
): Promise<TeamAnalytics> {
// FNXC:PostgresCommandCenterAnalytics 2026-06-27-10:00:
// Backend (PostgreSQL) path. Fetch agents + the four task-derived row sets
// from schema-qualified project.* tables (snake_case columns; the async
// connection has no `project` on search_path), then run the identical pure
// per-agent aggregation as the sync branch via buildTeamAnalytics.
if ("ping" in dbOrLayer) {
return aggregateTeamAnalyticsAsync(dbOrLayer, query);
}
const db = dbOrLayer as Database;
const agents = db
.prepare(`SELECT id, name, role, state FROM agents ORDER BY id`)
.all() as AgentRow[];
const tokenClauses = ["assignedAgentId IS NOT NULL", "tokenUsageLastUsedAt IS NOT NULL"];
const tokenParams: string[] = [];
addRangeClauses("tokenUsageLastUsedAt", tokenClauses, tokenParams, query);
const tokenRows = db
.prepare(
`SELECT
assignedAgentId AS agentId,
tokenUsageInputTokens AS inputTokens,
tokenUsageOutputTokens AS outputTokens,
tokenUsageCachedTokens AS cachedTokens,
tokenUsageCacheWriteTokens AS cacheWriteTokens,
tokenUsageTotalTokens AS totalTokens,
modelProvider,
modelId,
tokenUsageModelProvider,
tokenUsageModelId
FROM tasks
WHERE ${tokenClauses.join(" AND ")}`,
)
.all(...tokenParams) as TaskTokenRow[];
const completedClauses = ["assignedAgentId IS NOT NULL", `"column" = 'done'`, "columnMovedAt IS NOT NULL"];
const completedParams: string[] = [];
addRangeClauses("columnMovedAt", completedClauses, completedParams, query);
const completedRows = db
.prepare(
`SELECT assignedAgentId AS agentId, COUNT(*) AS count
FROM tasks
WHERE ${completedClauses.join(" AND ")}
GROUP BY assignedAgentId`,
)
.all(...completedParams) as CountByAgentRow[];
const currentRows = db
.prepare(
`SELECT assignedAgentId AS agentId, "column" AS columnName, COUNT(*) AS count
FROM tasks
WHERE assignedAgentId IS NOT NULL AND "column" IN ('in-progress', 'in-review')
GROUP BY assignedAgentId, "column"`,
)
.all() as Array<CountByAgentRow & { columnName: string }>;
const filesClauses = ["assignedAgentId IS NOT NULL", "modifiedFiles IS NOT NULL", "modifiedFiles NOT IN ('', '[]')"];
const filesParams: string[] = [];
addRangeClauses("updatedAt", filesClauses, filesParams, query);
const fileRows = db
.prepare(
`SELECT assignedAgentId AS agentId, modifiedFiles
FROM tasks
WHERE ${filesClauses.join(" AND ")}`,
)
.all(...filesParams) as ModifiedFilesRow[];
return buildTeamAnalytics(agents, tokenRows, completedRows, currentRows, fileRows, query);
}
/**
* FNXC:PostgresCommandCenterAnalytics 2026-06-27-10:00:
* PostgreSQL fetch path for {@link aggregateTeamAnalytics}. Modified-files is
* jsonb (postgres-js returns it parsed), so countModifiedFiles receives a
* re-stringified value to keep the legacy JSON.parse path identical.
*/
async function aggregateTeamAnalyticsAsync(
layer: AsyncDataLayer,
query: TeamAnalyticsQuery,
): Promise<TeamAnalytics> {
const agents = (await layer.db.execute(
sql`SELECT id, name, role, state FROM project.agents ORDER BY id`,
)) as unknown as AgentRow[];
const tokFrom = query.from !== undefined ? sql`AND token_usage_last_used_at >= ${query.from}` : sql``;
const tokTo = query.to !== undefined ? sql`AND token_usage_last_used_at <= ${query.to}` : sql``;
const tokenRowsRaw = (await layer.db.execute(
sql`SELECT
assigned_agent_id AS "agentId",
token_usage_input_tokens AS "inputTokens",
token_usage_output_tokens AS "outputTokens",
token_usage_cached_tokens AS "cachedTokens",
token_usage_cache_write_tokens AS "cacheWriteTokens",
token_usage_total_tokens AS "totalTokens",
model_provider AS "modelProvider",
model_id AS "modelId",
token_usage_model_provider AS "tokenUsageModelProvider",
token_usage_model_id AS "tokenUsageModelId"
FROM project.tasks
WHERE assigned_agent_id IS NOT NULL AND token_usage_last_used_at IS NOT NULL ${tokFrom} ${tokTo}`,
)) as Array<Record<string, unknown>>;
const tokenRows: TaskTokenRow[] = tokenRowsRaw.map((r) => ({
agentId: String(r.agentId),
inputTokens: r.inputTokens == null ? null : Number(r.inputTokens),
outputTokens: r.outputTokens == null ? null : Number(r.outputTokens),
cachedTokens: r.cachedTokens == null ? null : Number(r.cachedTokens),
cacheWriteTokens: r.cacheWriteTokens == null ? null : Number(r.cacheWriteTokens),
totalTokens: r.totalTokens == null ? null : Number(r.totalTokens),
modelProvider: (r.modelProvider as string | null) ?? null,
modelId: (r.modelId as string | null) ?? null,
tokenUsageModelProvider: (r.tokenUsageModelProvider as string | null) ?? null,
tokenUsageModelId: (r.tokenUsageModelId as string | null) ?? null,
}));
const compFrom = query.from !== undefined ? sql`AND column_moved_at >= ${query.from}` : sql``;
const compTo = query.to !== undefined ? sql`AND column_moved_at <= ${query.to}` : sql``;
const completedRows = (await layer.db.execute(
sql`SELECT assigned_agent_id AS "agentId", count(*)::int AS count
FROM project.tasks
WHERE assigned_agent_id IS NOT NULL AND "column" = 'done' AND column_moved_at IS NOT NULL ${compFrom} ${compTo}
GROUP BY assigned_agent_id`,
)) as unknown as CountByAgentRow[];
const currentRows = (await layer.db.execute(
sql`SELECT assigned_agent_id AS "agentId", "column" AS "columnName", count(*)::int AS count
FROM project.tasks
WHERE assigned_agent_id IS NOT NULL AND "column" IN ('in-progress', 'in-review')
GROUP BY assigned_agent_id, "column"`,
)) as unknown as Array<CountByAgentRow & { columnName: string }>;
const filesFrom = query.from !== undefined ? sql`AND updated_at >= ${query.from}` : sql``;
const filesTo = query.to !== undefined ? sql`AND updated_at <= ${query.to}` : sql``;
const fileRowsRaw = (await layer.db.execute(
sql`SELECT assigned_agent_id AS "agentId", modified_files AS "modifiedFiles"
FROM project.tasks
WHERE assigned_agent_id IS NOT NULL
AND modified_files IS NOT NULL
AND jsonb_typeof(modified_files) = 'array'
AND jsonb_array_length(modified_files) > 0
${filesFrom} ${filesTo}`,
)) as Array<{ agentId: string; modifiedFiles: unknown }>;
const fileRows: ModifiedFilesRow[] = fileRowsRaw.map((r) => ({
agentId: String(r.agentId),
modifiedFiles: r.modifiedFiles == null ? null : JSON.stringify(r.modifiedFiles),
}));
return buildTeamAnalytics(agents, tokenRows, completedRows, currentRows, fileRows, query);
}
/**
* FNXC:PostgresCommandCenterAnalytics 2026-06-27-10:00:
* Pure per-agent aggregation shared by the sync (SQLite) and async (PostgreSQL)
* paths of {@link aggregateTeamAnalytics}. No I/O — takes already-fetched rows.
*/
function buildTeamAnalytics(
agents: AgentRow[],
tokenRows: TaskTokenRow[],
completedRows: CountByAgentRow[],
currentRows: Array<CountByAgentRow & { columnName: string }>,
fileRows: ModifiedFilesRow[],
query: TeamAnalyticsQuery,
): TeamAnalytics {
const summaries = new Map<string, TeamAgentSummary>();
const costAccumulators = new Map<string, CostAccumulator>();
const totalTokens = emptyTokenTotals();
const totalCost = emptyCostAccumulator();
const pricingOverrides = query.pricingOverrides;
for (const agent of agents) {
summaries.set(agent.id, makeSummary(agent.id, agent));
costAccumulators.set(agent.id, emptyCostAccumulator());
}
const ensureSummary = (agentId: string): TeamAgentSummary => {
const existing = summaries.get(agentId);
if (existing) return existing;
const created = makeSummary(agentId);
summaries.set(agentId, created);
costAccumulators.set(agentId, emptyCostAccumulator());
return created;
};
for (const row of tokenRows) {
const summary = ensureSummary(row.agentId);
const agentCost = costAccumulators.get(row.agentId) ?? emptyCostAccumulator();
costAccumulators.set(row.agentId, agentCost);
addTokenRow(summary.tokens, row);
addTokenRow(totalTokens, row);
addRowCost(agentCost, row, query.now, pricingOverrides);
addRowCost(totalCost, row, query.now, pricingOverrides);
}
for (const row of completedRows) {
ensureSummary(row.agentId).tasksCompleted = row.count;
}
for (const row of currentRows) {
const summary = ensureSummary(row.agentId);
if (row.columnName === "in-progress") summary.tasksInProgress = row.count;
if (row.columnName === "in-review") summary.tasksInReview = row.count;
}
for (const row of fileRows) {
ensureSummary(row.agentId).filesChanged += countModifiedFiles(row.modifiedFiles);
}
for (const [agentId, summary] of summaries) {
summary.cost = finalizeCost(costAccumulators.get(agentId) ?? emptyCostAccumulator());
}
let filesChanged = 0;
let tasksCompleted = 0;
let tasksInProgress = 0;
let tasksInReview = 0;
for (const summary of summaries.values()) {
filesChanged += summary.filesChanged;
tasksCompleted += summary.tasksCompleted;
tasksInProgress += summary.tasksInProgress;
tasksInReview += summary.tasksInReview;
}
const sortedAgents = [...summaries.values()].sort((a, b) => {
const tokenCmp = b.tokens.totalTokens - a.tokens.totalTokens;
if (tokenCmp !== 0) return tokenCmp;
return a.agentId.localeCompare(b.agentId);
});
return {
from: query.from ?? null,
to: query.to ?? null,
totals: {
tokens: totalTokens,
cost: finalizeCost(totalCost),
filesChanged,
tasksCompleted,
tasksInProgress,
tasksInReview,
},
agents: sortedAgents,
};
}