Documentation Index

Fetch the complete documentation index at: https://help.elaine.io/llms.txt

Use this file to discover all available pages before exploring further.

API errors, retries, and troubleshooting

Prev Next

 Reliable Elaine integrations distinguish permanent request errors from temporary failures, preserve correlation information, and retry only when the operation can be repeated safely.

Error and retry matrix

                                                                                                                                                                                             
ConditionTypical meaningRecommended action
Authentication or authorization failureThe identity, token, credential, network restriction, or permission is invalid.Do not retry unchanged. Correct access configuration and rotate compromised credentials.
Validation or business-rule rejectionThe request format or data does not meet the operation's requirements.Correct the payload. Retrying the same request will normally fail again.
Rate or concurrency limitThe integration is sending more parallel work than the configured capacity accepts.Reduce concurrency and retry with exponential backoff and jitter.
Timeout or temporary service failureThe outcome may be unknown or the dependency may be temporarily unavailable.Check status or use the correlation identifier before retrying. Apply bounded backoff.
Partial bulk failureSome items succeeded while others failed.Reconcile item-level results and repeat only safe failed items.
SIEIS file rejectionThe file, schema, name, encoding, configuration, or contained records are invalid.Review processing status and logs, correct the cause, and submit a clearly identified replacement file.

Prevent duplicates

 A timeout does not prove that Elaine did not accept the operation. Use supported idempotency controls or a stable external identifier, query the resulting status where possible, and keep a durable record of every submission and response before repeating a request.

Troubleshooting checklist

     
  1. Record the timestamp, environment, interface, operation, correlation identifier, and sanitized error response.
  2.  
  3. Confirm endpoint, authentication, permissions, IP or VPN route, and certificate validity.
  4.  
  5. Validate payload syntax, required fields, encoding, size, and operation-specific rules.
  6.  
  7. Check configured concurrency, timeout, retry, and batch-abort behavior.
  8.  
  9. Determine whether the request failed, was accepted, or completed partially.
  10.  
  11. Reproduce with the smallest non-sensitive payload.
  12.  
  13. Escalate with correlation data and timing information, never with production secrets.

Monitor the integration

     
  • Request rate, latency, and failure rate.
  •  
  • Rate-limit responses, retry volume, and oldest pending work.
  •  
  • Transactional acceptance, queue, delivery, and bounce status.
  •  
  • SIEIS files awaiting processing, rejected records, and delayed exports.
  •  
  • Event and webhook delivery age, repeated delivery, and terminal failure.