Python SDK bulk user creation loop failing with 429 despite retries

General Discussion
1

I’m writing a script to bulk-create users from a CSV file using the Genesys Cloud Python SDK. The goal is straightforward: read the CSV, map the fields to the UserPost model, and call create_user in a loop.

Here’s the relevant snippet:

from genesyscloud.api import users_api
from genesyscloud.rest import ApiError

users = users_api.UsersApi(api_client)

for row in csv_data:
 body = users_api.UserPost(
 name=row['name'],
 email=row['email'],
 division_id=default_division_id,
 phone_numbers=[{
 "number": row['phone'],
 "type": "work",
 "extension": row['ext']
 }]
 )
 
 try:
 users.create_user(body=body)
 print(f"Created {row['name']}")
 except ApiError as e:
 print(f"Failed for {row['name']}: {e.status} {e.reason}")

The docs state: “The API supports creating up to 100 users per minute.” I’ve added a simple time.sleep(0.7) between requests to stay under the limit, but I’m still hitting 429 Too Many Requests errors after about 15 users.

The error message is generic: Rate limit exceeded. It doesn’t specify which limit. Is there a different rate limit for POST /api/v2/users compared to GET endpoints? Or is the Python SDK’s internal retry logic conflicting with my manual sleep?

I’ve checked the division ID. It’s correct. The payload structure matches the schema validator. No 400s. Just 429s.

I’ve tried increasing the sleep to 1.0 second. Same result. I’ve tried using the batch endpoint. It doesn’t exist for users. Only for skills and wrap-up codes.

Am I missing a header? A specific SDK configuration for bulk operations? The Python SDK docs don’t mention bulk creation patterns. They just show single user creation.

What’s the actual rate limit for user creation? And how should I structure the loop to avoid hitting it? I don’t want to hard-code a sleep value that might break tomorrow. I need a programmatic way to handle this.

Also, is there a way to get a pre-signed URL or a batch job ID for this? I’d rather not poll for completion if I can avoid it.

Any code examples for handling this gracefully would be appreciated. I’ve seen Terraform do it, but the Python SDK feels like I’m reinventing the wheel here.

2

The 429 errors usually hit because you’re ignoring the Retry-After header in the response. Here’s a quick backoff loop that actually respects the rate limit window:

import time

for row in csv_data:
 try:
 users.create_user(body=body)
 except ApiError as e:
 if e.status == 429:
 time.sleep(float(e.headers.get('Retry-After', 2)))
 users.create_user(body=body) # retry once