Challenges with Batch Submission Post-Revert and Blob Transaction Failures

[XsollaZK]

Description:

System Versions:

zksync-era commit: bfedac03b
zkstack CLI tool: v0.1.0-bfedac03

Issue #1: Extended Batch Submission Gap

Our ZK Stack server node ceased batch submissions for over three days due to insufficient testnet ETH for gas payments.

Resolution Steps Taken:

Utilized the batch-reverter tool to revert the chain state to batch 1970.
Executed a transaction to finalize the revert process: Etherscan Transaction

The on-chain transaction was successful, indicating that the off-chain infrastructure should submit the correct data.

Steps Executed:

./block_reverter rollback-db
./block_reverter print-suggested-values for our operator
./block_reverter send-eth-transaction
./block_reverter clear-failed-transactions

Question: Are there additional steps required to revert batches effectively?

Issue #2: Blob Transaction Submission Failures Post-Revert

After restarting the server, it attempted to send blob transactions with incorrect future nonces, repeatedly retrying with increasing fees, and eventually crashing due to exceeding the max priority fee per gas limit.

This behavior mirrors the issue detailed here: GitHub Issue #3373

Question: What causes this behavior, and how can we prevent it?

Issue #3: Persistent Batch Submission Failures After Forced Revert

We backed up the database and attempted to resolve the problem by removing old batches and corresponding transactions. This allowed the server to execute batch submissions with fresh timestamps. However, all transactions attempting to commit batches are reverting without an error message.

Numerous transactions sent by our server have reverted without state changes on the blockchain. We have since restored the database from the backup.

Question: Why do blob submissions revert even after the chain rollback? Is there a method to forcefully regenerate batches?