Describe the bug
I use the testing programs to send 500000 transactions to a group with 4 nodes, it seems that over 100 of the transactions cannot be committed successfully.
To Reproduce
Steps to reproduce the behavior:
Setup a group with 4 nodes, each with a small size of tx pool.
Constantly sending txs to the group
The tx pool of some nodes are full and txs forwarded from other nodes are rejected
If the tx pool of the leader node is empty and the txpool of less than f nodes are empty, the txs of non-leader node will not be handled, the bug will be triggered.
Expected behavior
All transactions should be commited successfully.
Environment (please complete the following information):
OS: Ubuntu 20.04
FISCO BCOS release-3.0.0-rc2
The text was updated successfully, but these errors were encountered:
After the txppool is full, the current logic will reject the txs forwarded by other nodes from P2P, resulting in some txs only exist in less than f nodes
As the stress test progresses, txs in the txpool with more than (2*f+1) nodes will be processed, causes the txpool of more than (2f+1) nodes are empty and less than (f) nodes are non-empty.
If the subsequent consensus leader does not belong to the non-empty node combination, it will not seal and generate new block, resulting in some transactions not being processed.
Even if the node txpool is full, accept txs forwarded from P2P by other nodes, ensure that txs received by some consensus node are broadcast to all other consensus nodes that may act as leaders
Describe the bug

I use the testing programs to send 500000 transactions to a group with 4 nodes, it seems that over 100 of the transactions cannot be committed successfully.
To Reproduce
Steps to reproduce the behavior:
Expected behavior
All transactions should be commited successfully.
Environment (please complete the following information):
The text was updated successfully, but these errors were encountered: