Steps to Reproduce:
1) prepare pool with 25 nodes and 100l txns
2) setup load - 2000 clients, 15 txns/sec write-only
python3 perf_processes.py -g pool_transactions_genesis -m t -c 200 -t 22 -n 1 -k nym python3 perf_processes.py -g pool_transactions_genesis -m t -c 200 -t 66 -n 1 -k schema python3 perf_processes.py -g pool_transactions_genesis -m t -c 200 -t 66 -n 1 -k attrib python3 perf_processes.py -g pool_transactions_genesis -m t -c 200 -t 132 -n 1 -k cred_def python3 perf_processes.py -g pool_transactions_genesis -m t -c 200 -t 132 -n 1 -k cred_def
Actual results:
After ~12 hours pool performance started to degrade and 11 view_changes were happened one by one (no veiw_change happened before)
Additional info:
Logs are available on LogProcessor:logs/cyclic_view_change_14jul2018
Historical states of nodes are available in status_nodeX.log files