improvement(cleanup): chunk-index labels, clarify upper-bound failure counter

Addresses Greptile review feedback:
- Disambiguate downstream logs when a plan splits into multiple workspace chunks (e.g. 'free/1', 'free/2')
- Document that deleteRowsById's failed counter is an upper bound (chunk rolls back to 0 deletes on error)

Author: waleedlatif1

apps/sim/lib/billing/cleanup-dispatcher.ts

@@ -213,8 +213,14 @@ async function buildCleanupChunks(jobType: CleanupJobType): Promise<CleanupJobPa
       .filter((row) => planByWorkspaceId.get(row.id) === plan)
       .map((row) => row.id)
     if (workspaceIds.length === 0) continue
-    for (const ws of chunkArray(workspaceIds, WORKSPACES_PER_CLEANUP_CHUNK)) {
-      chunks.push({ plan, workspaceIds: ws, retentionHours, label: plan })
+    const planChunks = chunkArray(workspaceIds, WORKSPACES_PER_CLEANUP_CHUNK)
+    for (const [idx, ws] of planChunks.entries()) {
+      chunks.push({
+        plan,
+        workspaceIds: ws,
+        retentionHours,
+        label: planChunks.length > 1 ? `${plan}/${idx + 1}` : plan,
+      })
     }
   }
 

apps/sim/lib/cleanup/batch-delete.ts

@@ -240,9 +240,12 @@ export async function deleteRowsById(
         .returning({ id: idCol })
       result.deleted += deleted.length
     } catch (error) {
+      // Upper bound: Postgres rolls back the chunk on error, so actual deletes = 0,
+      // but we can't tell which IDs in the chunk would have matched. The next cron
+      // run picks up whatever's still expired, so this only inflates the metric.
       result.failed += chunkIds.length
       logger.error(
-        `[${tableName}] Delete chunk ${chunkIdx + 1}/${chunks.length} failed (${chunkIds.length} rows):`,
+        `[${tableName}] Delete chunk ${chunkIdx + 1}/${chunks.length} failed (up to ${chunkIds.length} rows):`,
         { error }
       )
     }