Closed
Description
Description
Forking happens when a network under stress test especially when got spam transaction for less than an hour. Need to find a solution to make the network at least still synchronized each other under spam transaction.
Some suspect forking issues:
- Push Block process time became too long because waiting for
DB Executor BeginTX
(havesync Mutex
).DB Executor
is used everywhere API, P2P & Core Service - Receiving & broadcasting transaction too high it makes the number of goroutines increased significantly
Steps to reproduce
- Run alpha network with 100 nodes wait until synchronized
- Spam with around 10
send money
transactions per seconds
Expected behavior
The network still synchronized under spam transactions.