Details
-
Bug
-
Status: Complete
-
Medium
-
Resolution: Done
-
None
-
None
-
None
-
Build Info:
indy-node 1.0.111
indy-anoncreds 1.0.25
indy-plenum 1.0.97
sovrin 1.0.23
OS/Platform: Ubuntu 16.04.2 LTS
Setup: performance pool (7 nodes, 2 clients)
-
12
Description
Steps to Reproduce:
1. Connect to pool.
2. Using two clients, run load script several times:
Client 1 (all tests are running one by one):
python3 load_test.py --clients-list load_test_clients.list -t NYM -c 10 -r 100 --at-once
python3 load_test.py --clients-list load_test_clients.list -t NYM -c 1 -r 500 --at-once
python3 load_test.py --clients-list load_test_clients.list -t NYM -c 1 -r 500 --at-once
python3 load_test.py --clients-list load_test_clients.list -t NYM -c 10 -r 300 --at-once
python3 load_test_inv.py --clients-list load_test_clients.list -t NYM -c 5 -r 100 --at-once
Client 2 in parallel with load_test_inv.py on client 1:
python3 load_test.py --clients-list load_test_clients.list -t NYM -c 10 -r 100 --at-once
=>test was failed because of InvalidMessageExceedingSizeException)
Invalid test from client 1 and valid test from client 2 are stopped, new test on client 2:
python3 load_test.py --clients-list load_test_clients.list -t NYM -c 10 -r 100 --at-once
Actual Results:
Pool is broken after these actions. Following messages repeatedly appear in the log:
2017-08-24 10:42:37,450 | DEBUG | replica.py ( 311) | last_ordered_3pc | Node1:1 set last ordered as (0, 0) 2017-08-24 10:42:37,450 | DEBUG | replica.py ( 953) | __is_next_pre_prepare | Node1:2 missing PRE-PREPAREs between 10 and 8 2017-08-24 10:42:37,450 | DEBUG | replica.py ( 495) | _setup_for_non_master | Node1:2 Setting last ordered for non-master as (0, 0) 2017-08-24 10:42:37,450 | DEBUG | replica.py ( 311) | last_ordered_3pc | Node1:2 set last ordered as (0, 0) 2017-08-24 10:42:37,463 | DEBUG | replica.py ( 953) | __is_next_pre_prepare | Node1:1 missing PRE-PREPAREs between 6 and 4 2017-08-24 10:42:37,464 | DEBUG | replica.py ( 495) | _setup_for_non_master | Node1:1 Setting last ordered for non-master as (0, 0)
Expected Result:
Pool should work.
Additional Information:
Count of transactions is equal for ledgers on all nodes:
ubuntu@californiaPerf1:/home/sovrin/.sovrin$ sudo read_ledger --type=domain --count 1406
Attachments
Issue Links
- relates to
-
INDY-804 Pool doesn't come back to consensus
-
- Complete
-