Oracle Fusion Cloud ERP and Adobe Commerce integration

What moves between them

Orders and invoices flow from Adobe Commerce into Oracle Fusion Cloud ERP. When a customer places an order in Adobe Commerce, ml-connector receives the webhook, maps the order lines to Oracle Fusion AR line items and GL account combinations, creates a receivables invoice in Fusion, and posts the corresponding GL journal batch to record the sale. Customer and product reference data can also flow back from Fusion into Adobe Commerce to synchronize GL account assignments and customer master attributes. GL posting is one-way: ml-connector does not write back to Adobe Commerce.

How ml-connector handles it

ml-connector stores both credential sets encrypted and refreshes the Oracle Fusion OAuth token whenever a call returns 401 (approximately every hour). It accepts the customer-specific Fusion pod URL and validates API paths against that instance. On the Adobe Commerce side, ml-connector verifies incoming webhook signatures using the configured HMAC key, buffers the order event, and polls Oracle Fusion to fetch the required GL accounts and customer master records before posting the invoice. Because Oracle Fusion has no push capability, ml-connector also polls Fusion on a schedule to detect new customers or products that need to sync back to Adobe Commerce. Customer and product codes are matched across both systems to ensure invoice lines post to the correct GL account combinations in Fusion. Every record carries timestamps and a full audit trail so transactions can be replayed if a downstream call fails.

A real-world example

A mid-sized B2B distributor runs Adobe Commerce SaaS for its online ordering portal and Oracle Fusion Cloud ERP for procurement, inventory, and finance. Before the integration, orders placed in Adobe Commerce were exported as CSV, imported into Fusion via FBDI bulk file, and the finance team manually validated GL account assignments before running the final revenue posting. This process introduced delays in AR reporting and left room for manual error. With Oracle Fusion and Adobe Commerce connected, each order webhook immediately triggers an AR invoice in Fusion with the correct GL postings based on the product and customer, and the finance team runs month-end close without the import delay or the reconciliation step.

What you can do

• •Receive Adobe Commerce order and invoice webhooks and post them as accounts receivable documents in Oracle Fusion Cloud ERP.
• •Map Adobe Commerce customers and product codes to Oracle Fusion GL account combinations and customer master records.
• •Create and post sales journal batches in Fusion with the correct GL account assignments based on customer and product dimension.
• •Verify Adobe Commerce webhook signatures with HMAC SHA256 and handle OAuth 2.0 token refresh on both the Fusion and Commerce sides.
• •Sync customers and product reference data from Oracle Fusion back to Adobe Commerce to keep the product catalog and customer attributes in agreement.

Questions

How does the integration bridge the OAuth differences between Oracle Fusion and Adobe Commerce?

ml-connector stores both credential sets encrypted: the Fusion OAuth 2.0 client credentials for the pod-specific tenant, and either the PaaS OAuth 1.0a credentials or the SaaS OAuth 2.0 credentials for your Adobe Commerce instance. It refreshes the Fusion token every hour and the Commerce token on expiry, presenting the correct authorization header on each API call. The credentials are never logged or exposed.

What happens when an Adobe Commerce webhook arrives while Oracle Fusion is temporarily unavailable?

ml-connector buffers the webhook event in a queue and retries posting the order to Fusion with exponential backoff and jitter. If the posting fails after the retry limit, the event is marked for manual review, logged in the audit trail with the original payload, and can be replayed once the Fusion issue is resolved. Every transaction has a full timestamp history so no order is lost.

Does the integration support both PaaS and SaaS models of Adobe Commerce?

Yes. ml-connector accepts the appropriate credential set based on your deployment: OAuth 1.0a consumer key and secret for PaaS, or OAuth 2.0 client credentials for SaaS. The API endpoints differ slightly between the two models, and ml-connector routes requests to the correct base URL for your configuration.