Getting several errors when running Libra validators on a cluster

Hi all,

I’m performing some (scalability) experiments and I am currently trying to setup 4 Libra validator nodes, running on distinct machines. I’m using the code in the testnet branch of Github and generate the configuration files with the config generator. I managed to compile the code and run the four validator nodes. They seem to be able to connect to each other and coordinate the consensus correctly.

Now I am extending my current setup by running 4 clients. These clients connect to each deployed validator. I use Python with the pexpect library to spawn the client and interact with it. After the client has spawned, it creates two new accounts (with the a c command) and three seconds later, it submits a mint request to the validator (with the a m 0 1000000 command). Ten seconds later, a client transfers some coins from account 0 to account 1 with the t 0 1 100 100 command.

This transfer only succeeds for one client. I’m getting these errors on the other clients:

When minting:

[ERROR] Error minting coins: Transaction failed with mempool status: MempoolAddTransactionStatus { code: InvalidUpdate, message: "Failed to update gas price to 0" }

When transferring:

[ERROR] Failed to perform transaction: Transaction failed with vm status: Validation(SendingAccountDoesNotExist("sender address: 44715dbb8eb3167b4e209b8c747868a289076550c305dc692ceaf04d6d38ec1c"))

However, the client did report that the accounts have been created:

a ce
>> Creating/retrieving next account from wallet
Created/retrieved account #0 address 44715dbb8eb3167b4e209b8c747868a289076550c305dc692ceaf04d6d38ec1c

ea ce
>> Creating/retrieving next account from wallet
Created/retrieved account #1 address 77ea32be98275d9779150a0e4e68bd0873c773400eef95b40e2bc7e34dcdf2e2

Am I missing something here? Or could this be a bug?

Any help would be greatly appreciated! :slight_smile:

1 Like

I figured out what went wrong. I was running multiple minting clients that were using the same minting account simultaneously. Making sure that there is only one client that is started with the -m flag solved it :+1: