We’re running into consistent timeout failures with our supplier onboarding automation script using Vault Loader for bulk document uploads. The script handles around 150-200 supplier qualification documents per batch, but we’re seeing timeouts after processing about 80-90 documents.
The error we’re getting:
Vault API Error: Request timeout after 300s
Batch processing halted at document 87/183
Connection reset by peer
We’ve tried adjusting the batch size from 200 down to 100, but still hitting the same wall. The timeout seems to occur regardless of document size - we’re uploading mostly PDFs ranging from 500KB to 5MB. No retry mechanism is currently in place, so failed batches require complete manual re-upload.
Has anyone successfully configured Vault Loader for high-volume supplier document imports? What are the recommended batch size limits and timeout configurations for 23R2?
Thanks both. We’re running synchronous calls currently. The async approach makes sense but I’m concerned about tracking which documents actually made it through versus which failed. How do you handle the audit trail when uploads are async?
I’ve seen this exact issue. The 300s timeout is actually a combination of API gateway timeout and your script’s connection timeout. For 23R2, Veeva recommends keeping batches under 50 documents when using Vault Loader API for supplier documents. The issue isn’t just document count - it’s the metadata validation overhead for supplier-specific fields that adds processing time.
Are you processing synchronously or asynchronously? We switched to async mode and it solved our timeout problems. Also, implement exponential backoff retry logic - don’t just fail the entire batch. Start with 2-second delays and double up to 32 seconds max between retries.
We implemented a job tracking table in our middleware database. Each document gets a unique job ID, and we poll the Vault job status endpoint every 30 seconds. For supplier onboarding, you need that traceability anyway for compliance. Our batch size is 25 documents max, and we haven’t had timeouts since switching. The key is tuning both batch size AND implementing proper retry with state tracking.
Check your network latency too. We had similar issues and discovered our middleware server was routing through a secondary datacenter. Reduced our average response time from 180ms to 45ms just by fixing routing. That extra overhead per document adds up fast in bulk operations.
One thing to check - are you closing connections properly between batches? We found connection pool exhaustion was causing cascading timeouts. Also verify your API burst limits in 23R2. Supplier management module has stricter rate limits than standard document control.