What happens when an AMB destination contract call fails?
What happens when an AMB destination contract call fails? The receiver’s state change reverts; recovery depends on whether the relay records failure or reverts.
By Crypto Docket Newsroom#c2f7c43 min read

When an AMB destination contract call fails, the receiver’s changes revert while the source transaction stays committed; recovery depends on whether the relay records the failure or reverts.
What happens during an AMB destination call?
The Arbitrary Message Bridge takes a target address, encoded call data and a gas allowance, then executes that call in a destination-chain transaction. The source transaction and destination relay are separate transactions, so a destination failure does not undo the source-chain transaction.
On Gnosis Bridge, an ordinary relay through executeSignatures can record a failed call and still complete the relay transaction. For more on how the bridge moves tokens and messages between Ethereum and Gnosis Chain, see gnosis chain bridge.
The destination contract’s state changes roll back when its call reverts. The bridge stores false in messageCallStatus and stores information identifying the failed message, including its data hash, receiver and sender. The relay is marked as processed, which prevents the same ordinary relay from being executed again. The [MessageProcessor.sol](https://github.com/gnosischain/tokenbridge-contracts/blob/master/contracts/upgradeable_contracts/arbitrary_message/MessageProcessor.sol) code defines the status and failed-message fields.
Common causes include a paused receiver, invalid call arguments or too little gas for the receiver’s work. A relay can fail for different reasons, but its stored status and transaction outcome reveal whether execution was recorded or rolled back.
What does a failed call look like in practice?
Say a message targets a receiver contract and carries encoded setValue(7) data with a gas allowance. The receiver is paused and its function reverts. Its stored value remains unchanged because the failed subcall’s state changes are rolled back.
With ordinary executeSignatures, the destination transaction can still finish: the bridge records the failed status and the relay as processed. That gives the application a recorded failure to inspect, but the same message is no longer eligible for ordinary relay execution. A new attempt needs a recovery path supported by the application.
With safeExecuteSignatures, receiver failure reverts the relay transaction itself. The relay mark and other changes made in that transaction also roll back, preserving the message’s eligibility for resubmission. The BasicForeignAMB.sol code distinguishes the ordinary and safe execution paths.
Can you retry or recover a failed AMB message?
Recovery depends on which outcome the destination transaction recorded. A stored failure from ordinary execution requires an application-supported recovery process; the bridge’s failed-message data helps identify what failed, but does not by itself make the message eligible for the same ordinary relay again.
A reverted safe relay has a different trade-off. Because the transaction did not commit its relay mark, the message can be submitted again after the cause is resolved. That might mean unpausing the receiver, correcting a receiver-side condition or allowing enough gas for the call. Resubmission still requires a valid message and signatures.
In either path, gas spent on a submitted transaction is not restored by the receiver’s revert. A failed destination call also does not automatically refund a user: refunds depend on the application’s own rules and recovery support.
What should developers check after a destination failure?
Developers should first distinguish a recorded failed execution from a reverted relay. For an ordinary relay, inspect messageCallStatus and the stored failed-message details, then follow the application’s supported recovery process. For a safe relay, inspect the reverted transaction, resolve the receiver failure and resubmit the message if it remains valid.
That distinction is the key operational difference: ordinary execution preserves a record of failure but consumes ordinary relay eligibility; safe execution rolls the relay back and preserves resubmission eligibility, while spending gas on the failed attempt. Check the receiver’s pause state, arguments and gas allowance before choosing the next step.