MEDIUM: fireWebhook awaited synchronously in session-scoring — sequential endpoint timeouts stall scoring job

[walidboulanouar]

Problem

packages/jobs/src/session-scoring.ts line 139: await fireWebhook(...) is called synchronously. Internally deliverToEndpoint iterates endpoints in a for-loop with up to 10s timeout each. An org with N slow/timing-out webhook endpoints holds the scoring job open for up to N × 10s.

With Trigger.dev retry backoff < webhook timeout, retries pile up before the first completes.

Fix

Option A — run deliveries concurrently:

await Promise.all(endpoints.map(ep => deliverToEndpoint(ep, ...)));

Option B — move fireWebhook to a fire-and-forget child task:

await webhookDeliveryTask.trigger({ orgId, event, data });
// returns immediately; scoring is not blocked on delivery

Option B is preferred as it completely decouples scoring latency from delivery infrastructure.

[walidboulanouar]

Verified fixed in commit 8d624b2 — session-scoring fireWebhook is fire-and-forget, decoupled from scoring latency (session-scoring.ts:197-204).