[BUG] Sequence-sender sends next sequence immediately after restart, ignoring LastBatchVirtualizationTimeMaxWaitPeriod configuration

[akshat-zeeve]

Bug Report

Sequence-sender sends next sequence immediately after restart, ignoring LastBatchVirtualizationTimeMaxWaitPeriod configuration.

Description

We found a bug in the sequence-sender of the cdk-node service. The issue is with the timing of when the next sequences are sent. The configuration is set to send sequences every 3 hours, but when we restart the cdk-node, it sends the next sequence right away instead of waiting for the 3-hour period.

To Reproduce

Steps to reproduce the behavior:

1. Wait for a batch to be created and then restart sequence-sender

Actual behavior

Upon restarting the cdk-node, next sequences are sent immediately, bypassing the configured 3-hour interval.

Expected behavior

Sequences should wait for the configured time period (3 hours) before being sent, even after restarting the cdk-node.

Configuration

[SequenceSender]
MaxPendingTx = 5
SequencesTxFileName = "/tmp/sequencesender.json"
LastBatchVirtualizationTimeMaxWaitPeriod = "3h"
        [SequenceSender.EthTxManager]
                StoragePath = "/tmp/sequencesender-ethtxmanager.sqlite"

[ToniRamirezM]

On startup, the variable latestVirtualTime is not initialized, which causes the condition to send data to L1 to be erroneously met during its first check.

To resolve this issue, the variable should be set to the timestamp of the latest virtualized sequence. Achieving this requires syncing the virtualization events. While the l1-sync-lib on the aggregator already syncs this information, its database is not accessible from the sequence sender. Moreover, replicating all that data solely for this purpose seems excessive.

As an alternative, we can retrieve the latest virtualization event directly from L1 by querying a reasonable range of L1 blocks.