Sentry quota exhausted within first week of billing cycle

Summary

Sentry Stats & Usage (14-day window, flowfuse-frontend) shows two categories burning through quota:

Config lives in frontend/src/services/error-tracking.js.

Root Cause

Spans — two compounding issues:

1. tracesSampleRate: 0.5 in production is too high
2. Brokers/index.vue polls GET /api/v1/teams/:id/brokers/:id every 5s. A user with the page open for a workday generates ~3,000 spans from this endpoint alone

Session Replays — replaysOnErrorSampleRate: 0.25 fires on every error hit and is the primary driver (~200 replays/day)

Fix

Spans — reduce sample rate + exclude broker polling:

tracesSampleRate: window.sentryConfig.production ? 0.05 : 0.5,

new BrowserTracing({
    routingInstrumentation: vueRouterInstrumentation(router),
    shouldCreateSpanForRequest: (url) => {
        if (/\/brokers\/[^/]+$/.test(url)) return false
        return true
    }
})

Session Replays — reduce error-linked replay rate:

replaysSessionSampleRate: window.sentryConfig.production ? 0.01 : 0.1,
replaysOnErrorSampleRate: 0.1,

Expected Impact

Notes

• Errors (351 accepted / 14d) are fine, no changes needed
• We use PostHog for session recording — Sentry replays are only valuable when tied to an error, so keeping replaysOnErrorSampleRate at 0.1 rather than 0 preserves that value