Could use a hand troubleshooting this specific failure mode when running load tests against the Predictive Routing Campaign API.
We are executing a stress test using JMeter 5.6 to validate our capacity planning for a new contact center expansion in the US1 environment. The test plan simulates the creation and activation of multiple predictive outbound campaigns simultaneously. We are targeting an API throughput of 50 requests per second, which should be well within the documented limits for a standard admin role, but we are hitting a wall.
The issue manifests when we attempt to update the campaign status from ‘draft’ to ‘active’ via PATCH /api/v2/outbound/campaigns/{campaignId}. Under low concurrency (10 users), the requests succeed with 200 OK. However, once we scale to 50 concurrent threads, approximately 30% of the requests fail with a 400 Bad Request error. The error payload is not very descriptive, which makes debugging difficult.
{
“message”: “Invalid campaign configuration”,
“code”: “invalid.request”,
“status”: 400,
“details”: “The campaign settings do not meet the requirements for activation.”
}
This is confusing because the JSON body sent in the failing requests is identical to the successful ones. We have verified that the contactListId, campaignRules, and wrapUpCode are all valid and present. We are also ensuring that the campaigns are not overlapping in their dialing windows to avoid logical conflicts.
Given my background in API rate limiting, I initially suspected a 429 Too Many Requests response, but the 400 status suggests a validation error on the server side. Could there be a hidden dependency or a race condition in the Architect flow validation when multiple campaigns are activated in rapid succession?
We are using the Genesys Cloud OAuth2 client credentials flow for authentication, and the tokens are refreshed every 50 minutes. Is there a specific concurrency limit on the outbound campaign activation endpoint that is not documented? Any insights into what ‘Invalid campaign configuration’ might refer to in a high-concurrency context would be appreciated.