Handling Withdrawals
Phase 1: Receiving Withdrawal Request from a User

Step 1
The user should send a withdrawal request to the server. The withdrawal request should have the currency and amount being transferred.
Step 2
The server creates a pending burn transaction for this specific withdrawal. This will lock the amount being withdrawn in the user's balance. This ensures that this amount will not be used during this withdrawal process. The transaction_id of the pending burn is required for phase 3 of this process.
Phase 2: Payment service attempts withdrawal to user's bank account

Step 1
The server makes a withdrawal request using the payment service.
This step will differ based on which payment service is used.
Step 2
Payment service attempts to make the withdrawal to the user's bank account.
Step 3
Payment service responds to server with the withdrawal's status.
Phase 3: Server updates status of pending burn

Step 1
The server updates the status of the pending burn created in phase 2 based on response from payment service.
Step 2
Once the pending burn's status is updated, the locked amount on the user's balance will be unlocked. If the withdrawal was successful, the user's locked amount will be removed entirely from the total balance. If the withdrawal was unsuccessful, the locked amount will return to the user's total balance.
Code Example
For this example, we are using a placeholder payment service sdk called PAYMENT_SERVICE_SDK_LIBRARY. The withdrawal step will differ based on the payment service used. Regardless, the overall flow of this process should be relatively the same.
Advanced
Although bank withdrawals can be handled one by one like above, we generally recommend withdrawals to be batched in intervals. You can do this by using a cron job that runs every five minutes or so. Every time the job runs, it will query all pending burns in the database, batch them by currency, and make a single withdrawal per currency to the payment service. This will lower the amount of calls being made.
Last updated