Details
-
Bug
-
Status: Complete
-
High
-
Resolution: Done
-
None
-
None
-
None
-
Unset
-
EV 18.17 Service Pack
Description
Environment:
indy-nodeĀ 1.5.551
Steps to Reproduce:
1. Restore pool from backup with large ledger (1,340,000 txns), keep one node clear (Node24 in attached logs).
2. Start the pool exclude clear node.
3. Initialize clear node and start it.
=> Domain ledger of clear node is 30txns.
4. Wait several hours.
Actual Results:
Clear node still has 30 txns in domain ledger, following messages appear in logs of the rest nodes:
2018-08-15 09:29:56,416|INFO|ledger_manager.py|Node1 received catchup request: CATCHUP_REQ{'seqNoStart': 1061257, 'ledgerId': 1, 'catchupTill': 1340508, 'seqNoEnd': 1117110} from Node24 2018-08-15 09:29:56,416|INFO|ledger_manager.py|Node1 generating consistency proof: 1117110 from 1340508 2018-08-15 09:30:12,778|WARNING|prepare_batch.py|Too many split steps were done 9. Batches were not created 2018-08-15 09:30:12,779|WARNING|prepare_batch.py|Too many split steps were done 9. Batches were not created ...... several hundreds of the same messages ...... 2018-08-15 09:30:12,812|WARNING|prepare_batch.py|Too many split steps were done 9. Batches were not created 2018-08-15 09:30:12,812|WARNING|prepare_batch.py|Too many split steps were done 9. Batches were not created 2018-08-15 09:30:12,812|WARNING|prepare_batch.py|Too many split steps were done 9. Batches were not created 2018-08-15 09:30:12,812|ERROR|batched.py|Node1 cannot create batch(es) for Node24
Expected Results:
Clear node should complete catch up.
Additional Information:
- Pool restore was performed not quite right. Nodes 15, 18 and 25 were not started with the rest ones. 15 and 18 were started later (after steps to reproduce), 25 was not started at all.
- The same behavior for 730K txns in the ledger, when all nodes were restored correctly.
Logs: s3://qanodelogs/indy-1595
To get logs, run following command on log processor machine:
aws s3 cp --recursive s3://qanodelogs/indy-1595/ /home/ev/logs/indy-1595/
Attachments
Issue Links
- relates to
-
INDY-1657 Node can't write after catch up of large ledger.
-
- Complete
-