Shutdown and startup times are long on scalenet
On Scalenet, we have lots of bigger blocks. It is a common observation for me that after processing big blocks, the first shutdown and the first startup are slow.
This seems to have something to do with leveldb updates, as indicated by -debug=leveldb
activity in debug.log file.
Steps to reproduce:
- Sync Scalenet to tip
- Shutdown and startup a couple of times, to start with a clean state
- Produce a big block (elsewhere) and make it sync until there is no more processing going on
- Shutdown
- Startup
At 4. Shutdown
, there will be a big amount of leveldb activity in debug.log. It will slow the shutdown. At 5. Startup
there will be a lot of leveldb activity that will not slow down the startup: it will occur in the background after the startup.
This has been observed also by Mr Culianu on his system.