36,000 transactions were constructed on the nex chain within 30s, and the transaction completion time was 1h20min.

[moonpyt]

System information

Geth version: geth version 0.2.2-stable-0ee43076
CL client & version: e.g. lighthouse/nimbus/prysm@v1.0.0
OS & Version: linux
Commit hash : 0ee4307

Steps to reproduce the behaviour

36,000 transactions were constructed on the nex chain within 30s, and the transaction completion time was 1h20min.
But it took only 20 minutes to process 36,000 transactions

1. I tried to send 6000 transactions in the 30s, and the result was that 6000 transactions were processed in 1 minute
2. I found that the processing speed slowed down after 6,000 transactions. When I sent another 12,000 transactions, it took 20 minutes to process and obviously saw that the consumption slowed down after 6000 transactions were processed

Backtrace

[backtrace]

The following is a log I constructed of 36,000 transactions with a completion time of 1h20min.
The block at the beginning of transaction processing is about 282890, and the block after 1h20min processing is about 283327

node (3).log

[moonpyt]

@AnnaShaleva Could you please look at the above questions

[txhsl]

I forgot to, but let me indicate, two performance tests used different reannounce parameters.

Last time we tested in #194, we set txpool.reannouncetimethreshold to 1 minute, and the reannounceInterval is 1 minute as well.

But after we updated and merged the branch, the default txpool.reannouncetimethreshold is now 140 seconds, and the reannounceInterval is 140 seconds as well.