fix(cleanup): bind live-paused statuses as a Postgres array, not a row constructor

Author: TheodoreSpeaks

apps/sim/background/cleanup-logs.test.ts

@@ -61,6 +61,7 @@ vi.mock('@/lib/cleanup/batch-delete', () => ({
 
 vi.mock('@/lib/execution/payloads/large-value-metadata', () => ({
   LIVE_PAUSED_REFERENCE_STATUSES: ['paused', 'partially_resumed', 'cancelling'],
+  livePausedStatusArray: { op: 'livePausedStatusArray' },
   markLargeValuesDeleted: mockMarkLargeValuesDeleted,
   pruneLargeValueMetadata: mockPruneLargeValueMetadata,
   unreferencedLargeValuePredicate: vi.fn(() => ({ op: 'unreferencedLargeValuePredicate' })),

apps/sim/background/cleanup-logs.ts

@@ -20,6 +20,7 @@ import {
 } from '@/lib/cleanup/batch-delete'
 import {
   LIVE_PAUSED_REFERENCE_STATUSES,
+  livePausedStatusArray,
   markLargeValuesDeleted,
   pruneLargeValueMetadata,
   unreferencedLargeValuePredicate,
@@ -256,7 +257,7 @@ async function cleanupLegacyLargeExecutionValues(
                       SELECT 1
                       FROM ${pausedExecutions} AS ref_pe
                       WHERE ref_pe.execution_id = ref.execution_id
-                        AND ref_pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+                        AND ref_pe.status = ANY(${livePausedStatusArray})
                     )
                   )
                 )
@@ -279,7 +280,7 @@ async function cleanupLegacyLargeExecutionValues(
                     SELECT 1
                     FROM ${pausedExecutions} AS parent_owner_pe
                     WHERE parent_owner_pe.execution_id = parent_value.owner_execution_id
-                      AND parent_owner_pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+                      AND parent_owner_pe.status = ANY(${livePausedStatusArray})
                   )
                   OR EXISTS (
                     SELECT 1
@@ -300,7 +301,7 @@ async function cleanupLegacyLargeExecutionValues(
                             SELECT 1
                             FROM ${pausedExecutions} AS parent_ref_pe
                             WHERE parent_ref_pe.execution_id = parent_ref.execution_id
-                              AND parent_ref_pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+                              AND parent_ref_pe.status = ANY(${livePausedStatusArray})
                           )
                         )
                       )
@@ -316,7 +317,7 @@ async function cleanupLegacyLargeExecutionValues(
               SELECT 1
               FROM ${pausedExecutions} AS pe
               WHERE pe.execution_id = split_part(${workspaceFiles.key}, '/', 4)
-                AND pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+                AND pe.status = ANY(${livePausedStatusArray})
             )`
           )
         )

apps/sim/lib/execution/payloads/large-value-metadata.ts

@@ -22,6 +22,19 @@ const LARGE_VALUE_METADATA_PRUNE_BATCH_SIZE = 1_000
 const LARGE_VALUE_METADATA_PRUNE_MAX_ROWS_PER_TABLE = 5_000
 export const LIVE_PAUSED_REFERENCE_STATUSES = ['paused', 'partially_resumed', 'cancelling'] as const
 
+/**
+ * `LIVE_PAUSED_REFERENCE_STATUSES` as a Postgres `text[]` for `= ANY(...)`.
+ *
+ * Interpolating the raw JS array into a `sql` template renders a row
+ * constructor — `($1, $2, $3)::text[]` — which Postgres rejects with
+ * "cannot cast type record to text[]", failing the whole statement. Building
+ * `ARRAY[$1, $2, $3]::text[]` explicitly keeps one bind param per status.
+ */
+export const livePausedStatusArray = sql`ARRAY[${sql.join(
+  LIVE_PAUSED_REFERENCE_STATUSES.map((status) => sql`${status}`),
+  sql`, `
+)}]::text[]`
+
 export interface LargeValueOwner {
   key: string
   workspaceId: string
@@ -412,7 +425,7 @@ async function pruneStaleReferences(
                 SELECT 1
                 FROM ${pausedExecutions} AS pe
                 WHERE pe.execution_id = ref.execution_id
-                  AND pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+                  AND pe.status = ANY(${livePausedStatusArray})
               )
             )
             OR ref.source NOT IN ('execution_log', 'paused_snapshot')
@@ -568,7 +581,7 @@ export function unreferencedLargeValuePredicate() {
               SELECT 1
               FROM ${pausedExecutions} AS pe
               WHERE pe.execution_id = elvr.execution_id
-                AND pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+                AND pe.status = ANY(${livePausedStatusArray})
             )
           )
         )
@@ -582,7 +595,7 @@ export function unreferencedLargeValuePredicate() {
       SELECT 1
       FROM ${pausedExecutions} AS owner_pe
       WHERE owner_pe.execution_id = ${executionLargeValues.ownerExecutionId}
-        AND owner_pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+        AND owner_pe.status = ANY(${livePausedStatusArray})
     )
     AND NOT EXISTS (
       SELECT 1
@@ -602,7 +615,7 @@ export function unreferencedLargeValuePredicate() {
             SELECT 1
             FROM ${pausedExecutions} AS parent_owner_pe
             WHERE parent_owner_pe.execution_id = parent_value.owner_execution_id
-              AND parent_owner_pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+              AND parent_owner_pe.status = ANY(${livePausedStatusArray})
           )
           OR EXISTS (
             SELECT 1
@@ -623,7 +636,7 @@ export function unreferencedLargeValuePredicate() {
                     SELECT 1
                     FROM ${pausedExecutions} AS parent_ref_pe
                     WHERE parent_ref_pe.execution_id = parent_ref.execution_id
-                      AND parent_ref_pe.status = ANY(${LIVE_PAUSED_REFERENCE_STATUSES}::text[])
+                      AND parent_ref_pe.status = ANY(${livePausedStatusArray})
                   )
                 )
               )

apps/sim/lib/execution/payloads/live-paused-status-array.test.ts

@@ -0,0 +1,39 @@
+/**
+ * @vitest-environment node
+ */
+
+// Renders against the real drizzle dialect: the bug this guards against is a
+// rendering bug, so the global drizzle-orm mock would hide it.
+import { describe, expect, it, vi } from 'vitest'
+
+vi.unmock('drizzle-orm')
+
+import { sql } from 'drizzle-orm'
+import { PgDialect } from 'drizzle-orm/pg-core'
+import {
+  LIVE_PAUSED_REFERENCE_STATUSES,
+  livePausedStatusArray,
+} from '@/lib/execution/payloads/large-value-metadata'
+
+describe('livePausedStatusArray', () => {
+  const dialect = new PgDialect()
+
+  it('renders an ARRAY literal, not a row constructor', () => {
+    const { sql: text, params } = dialect.sqlToQuery(
+      sql`select 1 where s = ANY(${livePausedStatusArray})`
+    )
+
+    expect(text).toContain('ARRAY[')
+    expect(text).toContain('::text[]')
+    // `($1, $2, $3)::text[]` is what interpolating the raw JS array produces;
+    // Postgres rejects it with "cannot cast type record to text[]".
+    expect(text).not.toMatch(/\(\$\d+(, \$\d+)+\)::text\[\]/)
+    expect(params).toEqual([...LIVE_PAUSED_REFERENCE_STATUSES])
+  })
+
+  it('binds one parameter per status', () => {
+    const { sql: text } = dialect.sqlToQuery(sql`${livePausedStatusArray}`)
+    const placeholders = text.match(/\$\d+/g) ?? []
+    expect(placeholders).toHaveLength(LIVE_PAUSED_REFERENCE_STATUSES.length)
+  })
+})