:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
import { UsageIngestionClient } from '~/common/usage-ingestion/client'
import { UsageRecordBatch } from '~/common/usage-ingestion/usage-record-batch'
import { logger } from '~/common/utils/logger'
import { ValueMatcher } from '~/types'
const DEFAULT_FLUSH_INTERVAL_MS = 10_000
const MAX_PENDING_RECORDS = 2_000
export interface CdpBillableInvocation {
teamId: number
usageKey:
| 'cdp_billable_invocations'
| 'workflow_billable_invocations'
| 'workflow_emails_sent'
| 'workflow_push_sent'
| 'workflow_sms_sent'
/** Identity of the billed thing. A replay must produce the same value, or it bills twice. */
recordId: string
}
/**
* Deliberately independent of `app_metrics2`: it owns its own batching and flushing, so the app
* metric a call site also emits can be deleted without touching billing. The timer bounds how
* long a record waits; `shutdown()` is what keeps a graceful deploy lossless.
*/
export class CdpUsageReporterService {
private batch: UsageRecordBatch
private timer: NodeJS.Timeout | null = null
constructor(
client: UsageIngestionClient | null,
isTeamEnabled: ValueMatcher<number>,
private readonly flushIntervalMs: number = DEFAULT_FLUSH_INTERVAL_MS
) {
this.batch = new UsageRecordBatch(client, { unit: 'invocations', isTeamEnabled })
}
reportBillableInvocation(invocation: CdpBillableInvocation): void {
this.batch.add(invocation.teamId, invocation.usageKey, invocation.recordId)
if (this.batch.size >= MAX_PENDING_RECORDS) {
void this.flush()
return
}
this.scheduleFlush()
}
/** One pass, for the timer: an invocation reported mid-send waits for the next flush. */
async flush(): Promise<void> {
await this.send(() => this.batch.flush())
}
/** For a consumer that is stopping: keeps flushing until nothing is left in memory. */
async shutdown(): Promise<void> {
await this.send(() => this.batch.drain())
}
private async send(sendBatch: () => Promise<unknown>): Promise<void> {
if (this.timer) {
clearTimeout(this.timer)
this.timer = null
}
try {
await sendBatch()
} catch (error) {
logger.warn('\u26a0\ufe0f', 'failed to flush cdp usage records', { error: String(error) })
}
}
private scheduleFlush(): void {
if (this.timer || this.batch.size === 0) {
return
}
// Unreferenced so a pending flush never keeps the process alive on shutdown.
this.timer = setTimeout(() => void this.flush(), this.flushIntervalMs)
this.timer.unref()
}
}