writeContract returns “ghost tx hash”; subsequent writes hang

[Nielss01]

Check existing issues

• I checked there isn't already an issue for the bug I encountered.

Viem Version

2.37.5

Current Behavior

After the first reverted transaction, the following write functions — which should succeed — return transaction hashes that do not exist on-chain.

These transactions appear to hang indefinitely (remain pending) and fail to broadcast new transactions. Eventually, waitForTransactionReceipt throws a timeout error on the ghost hash.

Restarting the server temporarily resolves the issue.
After the first 30 seconds, nothing else happens in the video — it’s mainly to show that it keeps hanging for quite a while.

viem.repro.mp4

Expected Behavior

• When a writeContract call reverts, it should throw an error without returning a transaction hash.
• Subsequent writeContract calls should continue to function normally and broadcast valid transactions.

In my demo video, the pause action should work, because the token state is currently set to unpaused.

Steps To Reproduce

1. Start the NestJS server.

2. Call POST http://localhost:3000/token/0x914dba91dd57d68c9e2826c9e2007ec6bb3d79ec/unpause
   → This correctly returns a revert error, since the token is already Unpaused.

3. Then call POST http://localhost:3000/token/0x5b782861155c3e18081b0c77edc2a4de591547ff/pause
   → This call logs a transaction hash, but that hash cannot be found on the Polygon Amoy block explorer.

After step 2, all subsequent write calls remain pending or fail to broadcast properly until the process is restarted.

Link to Minimal Reproducible Example

https://github.com/Nielss01/viem-write-txhash-error

Anything else?

Summary

When a writeContract call (e.g. unpause) fails on-chain (reverts), I still receive a transaction hash. However, that hash does not exist on-chain (cannot be found on the explorer).

After this “ghost” transaction, all subsequent writes from the same walletClient:

• no longer seem to broadcast real transactions, or
• remain stuck in a “pending” state,
• and waitForTransactionReceipt times out waiting for the previous ghost hash.

The issue seems to occur likely in writeContract itself, nonce handling, or the underlying transport.

[jxom]

This is more of a Node Provider/RPC intricacy than a Viem one.

When a writeContract call (e.g. unpause) fails on-chain (reverts), I still receive a transaction hash. However, that hash does not exist on-chain (cannot be found on the explorer).

This may be expected. The Node will place this transaction in a mempool (pool of transactions yet to be broadcast onchain) and give you a transaction hash. Subsequent writes would be failing as it would likely be waiting for the "failing" transaction to be included on the block.

[Nielss01]

I first changed the RPC to a paid provider (Infura), but that didn’t fix the issue.

Then I tried creating a new walletClient for each request, while keeping the NonceManager as a global instance — this also didn’t work.
At that point, I started getting the following error:

[Nest] 22316  - 22/10/2025, 10:53:40   ERROR [ExceptionsHandler] Nonce provided for the transaction is lower than the current nonce of the account.
Try increasing the nonce or find the latest nonce with getTransactionCount.

After that, I removed the createNonceManager from the account setup, and everything started working again — even when I made the walletClient global again and reimported it.

So it seems like the issue is related to the NonceManager, which doesn’t seem to behave as expected in this setup.

I don’t have deep knowledge about nonce handling, so this is just a speculation.