fix(embeddings): classify aggregated batch failures

Author: waleedlatif1

apps/sim/lib/embeddings/api-error.ts

@@ -20,3 +20,18 @@ export class EmbeddingAPIError extends Error {
     this.isBYOK = isBYOK
   }
 }
+
+/** Finds an embedding failure through bounded aggregate/cause wrappers. */
+export function getEmbeddingAPIError(error: unknown): EmbeddingAPIError | null {
+  const pending = [error]
+  const seen = new Set<unknown>()
+  while (pending.length > 0 && seen.size < 32) {
+    const current = pending.pop()
+    if (!(current instanceof Error) || seen.has(current)) continue
+    seen.add(current)
+    if (current instanceof EmbeddingAPIError) return current
+    if (current.cause !== undefined) pending.push(current.cause)
+    if (current instanceof AggregateError) pending.push(...current.errors.slice(0, 32))
+  }
+  return null
+}

apps/sim/lib/knowledge/connectors/connector-error.test.ts

@@ -10,6 +10,42 @@ import {
 import { ConnectorDirectoryError } from '@/connectors/source-error'
 
 describe('connector failure diagnostics', () => {
+  it('finds embedding failures through concurrent batches and nested cause wrappers', () => {
+    const error = new Error('private outer wrapper', {
+      cause: new AggregateError([
+        new EmbeddingAPIError('private upstream message', 503),
+        new Error('private inner wrapper', {
+          cause: new AggregateError([new EmbeddingAPIError('private upstream message', 503)]),
+        }),
+        new Error('private sibling'),
+      ]),
+    })
+    expect(getConnectorFailureDiagnostic(error)).toEqual({
+      category: 'embedding',
+      status: 503,
+      message: 'Embedding service request failed (HTTP 503).',
+    })
+  })
+
+  it('terminates cyclic aggregate wrappers and still finds an embedding sibling', () => {
+    const error = new AggregateError([])
+    error.errors.push(new EmbeddingAPIError('private upstream message', 429), error)
+    error.cause = error
+    expect(getConnectorFailureDiagnostic(error)).toMatchObject({
+      category: 'embedding',
+      status: 429,
+    })
+    const cycle = new AggregateError([])
+    cycle.errors.push(cycle)
+    expect(getConnectorFailureDiagnostic(cycle)).toBeNull()
+  })
+
+  it('bounds traversal of deeply nested aggregate wrappers', () => {
+    let error: Error = new EmbeddingAPIError('private upstream message', 503)
+    for (let i = 0; i < 40; i++) error = new AggregateError([error])
+    expect(getConnectorFailureDiagnostic(error)).toBeNull()
+  })
+
   it.each([401, 403, 404, 429, 502, 503])(
     'does not attribute a wrapped embedding HTTP %s to the source',
     (status) => {

apps/sim/lib/knowledge/connectors/connector-error.ts

@@ -1,6 +1,6 @@
 import { findCause, getPostgresErrorCode } from '@sim/utils/errors'
 import { DrizzleQueryError } from 'drizzle-orm/errors'
-import { EmbeddingAPIError } from '@/lib/embeddings/api-error'
+import { getEmbeddingAPIError } from '@/lib/embeddings/api-error'
 import {
   ConnectorDirectoryError,
   ConnectorSourceError,
@@ -66,6 +66,19 @@ function classifyFailure(error: unknown): ConnectorFailureDiagnostic | null {
   if (databaseError) {
     return { category: 'database', message: 'Database request failed without a driver error code.' }
   }
+  const embeddingError = getEmbeddingAPIError(error)
+  if (
+    embeddingError &&
+    Number.isInteger(embeddingError.status) &&
+    embeddingError.status >= 400 &&
+    embeddingError.status <= 599
+  ) {
+    return {
+      category: 'embedding',
+      status: embeddingError.status,
+      message: `Embedding service request failed (HTTP ${embeddingError.status}).`,
+    }
+  }
   const httpError = findCause(
     error,
     (value): value is Error & { status: number } =>
@@ -78,13 +91,6 @@ function classifyFailure(error: unknown): ConnectorFailureDiagnostic | null {
   )
   if (!httpError) return null
   const { status } = httpError
-  if (httpError instanceof EmbeddingAPIError) {
-    return {
-      category: 'embedding',
-      status,
-      message: `Embedding service request failed (HTTP ${status}).`,
-    }
-  }
   const category = httpError instanceof ConnectorSourceError ? httpError.category : undefined
   if (category === 'authorization' || (!category && (status === 401 || status === 403))) {
     return {