Troubleshooting Centre

Business Central 429 Too Many Requests: Diagnosis and Resolution

How should waiting, retries, webhooks and queues be designed when Business Central reaches API limits?

Codariva ·

If a Dynamics 365 Business Central API call returns 429 Too Many Requests , a request limit has been exceeded. This alone does not indicate a problem with a user password or product data. Microsoft's official documentation recommends a controlled retry strategy when request volumes are high.

Common triggers

  • Querying the entire stock list again on every screen.
  • Several integrations generating heavy request traffic at the same time.
  • Polling too frequently instead of tracking changes through events.
  • Resending a failed operation repeatedly without waiting.

A practical solution

1. Measure: Record which endpoint and integration are increasing call volume. Correlate operations belonging to the same customer.

2. Add a cooldown: Use progressive delays or exponential backoff in line with relevant server responses and documentation. Adding jitter to retries can reduce the risk of all clients retrying at once.

3. Use a queue: Put critical work orders or orders into a traceable job queue rather than leaving them only in memory. Retried operations must not create duplicate orders.

4. Process changes: Consider Business Central webhooks for suitable resources; do not download a large dataset on every screen.

5. Reconcile again: Reconcile periodically to identify missed events and missing orders.

Example pseudocode

istek_yap()
429 gelirse:
    uygun_bekleme_suresini_belirle()
    kuyruga_al_ve_bekle()
    islem_kimligini_koruyarak_tekrar_dene()
sinirli_tekrar_sayisi_asildiysa:
    insan_incelemesine_gonder()

Success criterion

The goal is an integration that runs at peak times without losing orders, beyond fixing a single request. Monitor the ratio of failed to successful calls and queue delay separately.

When should retries stop?

Limit both the number of attempts and total waiting time for each job. A job must not disappear when the limit is reached; keep its failed or pending record and the person responsible for review visible. Adding unlimited parallel workers to speed up a queue increases load.

Example scenario: An overnight catalogue transfer overlaps with morning dealer orders. First narrow the catalogue query and reduce total concurrency for that environment. The order queue can be prioritised, but validation and duplicate prevention rules must still apply.

Acceptance check Success condition
Retry limit reached The job moves to human review; no infinite loop
Service restarted Pending jobs resume from durable storage
Submission response lost The existing reference is checked before a new record is created

For a single long-running request, use the 504 timeout guide.

When should an administrator help?

If a single controlled read still consistently returns 429 after reducing the request rate, review other applications and queues using the same environment with an administrator. Do not try to bypass the limit by assigning users arbitrary new identities. Contact product support with a sanitised error ID, time range, total concurrency and queue delay.

Sources and verification

Example flows are architectural recommendations, not claims of completed client projects or vendor partnerships. Technical sources checked: 17 October 2026. Installation version, licence and permissions must be verified separately.

Adapt it to your business workflow

ERP integration planning guide. Explore the scope of custom software development for further reading; we can assess your current process together.

Let's discuss your needs →