Take note of ubuntu performance performance when configuring a Atomic instance
Take note of postgresql documentation for database specific operations
The following items will be covered here
- All own identified issues experienced running Atomic API
Troubleshooting rollback issues
These are issues we have encountered due to various forks causing rollbacks in Atomic API
We have experienced an incident where the rollback kept on failing and retrying, causing the Atomic Indexer to run out of sync
Example_1
Typical error in the .pm2/logs/eosio-contract-api-filler-out-0.log would be:
2023-07-21T07:31:39.226Z [PID:4145502] [info] : Executed rollback query 29 / 274
2023-07-21T07:31:39.226Z [PID:4145502] [warn] : Fork rollback taking longer than expected. Executing query... {"operation":"delete","table":"contract_traces","values":null,"condition":{"str":"\"global_sequence\" = $1 AND \"account\" = $2","values":["82233095234","atomicassets"]}}
2023-07-21T07:40:45.070Z [PID:406251] [info] : Executed rollback query 1 / 274
2023-07-21T07:40:45.070Z [PID:406251] [warn] : Fork rollback taking longer than expected. Executing query... {"operation":"delete","table":"contract_traces","values":null,"condition":{"str":"\"global_sequence\" = $1 AND \"account\" = $2","values":["82233095412","atomicassets"]}}
2023-07-21T07:45:50.141Z [PID:406251] [info] : Executed rollback query 2 / 274
The database table public.contract_traces were queried and noticed I could not retrieve information for account 'atomicassets' when ordered by 'global_sequence' and limited the results to 100 records
It would just timeout and indicated a performance issue
I immediately thought this could could contribute to the constant rollback failures