Code:
akira@positron:~/CCE$ ./dbload.py -n -v
Processing Block: 6368 of 1637812
Main Loop cannot concatenate 'str' and 'int' objects
akira@positron:~/CCE$ ./dbload.py -v
Recheck Called
Processing Block: 7188 of 1637812
Processing Block: 6368 of 1637812
Main Loop cannot concatenate 'str' and 'int' objects
akira@positron:~/CCE$ ./dbload.py -v
Recheck Called
Processing Block: 7188 of 1637812
This is due to the time-out in normal mode , you later discovered to correct with the -l option to extend the time-out to 24 hours.
"Main Loop cannot concatenate 'str' and 'int' objects" is a bug corrected in the latest commit. It was purely a cosmetic bug with sending the error string to stderr, no function involved.
Edit - After second look, the error occurred during a proper build (-n). My guess is your coin daemon is running slow or is sluggish to respond. The RPC interface in comm.py will timeout the connection
after 10 seconds. If the daemon is responding slow, it would also account for the slow build time.
Code:
Processing Block: 10192 of 1637812
Main Loop (1062, u"Duplicate entry '-1' for key 'PRIMARY'")
Main Loop (1062, u"Duplicate entry '-1' for key 'PRIMARY'")
Odd error that pops up once in a great while when restarting during the initial load. Just remove the row in the block table with the height "-1" and continue.
Or
Code:
Processing Block: 13430 of 1637812
Main Loop sequence index must be integer, not 'str'
Main Loop sequence index must be integer, not 'str'
Related to the first to question. Make sure you are using the latest version.
I'm confused now
Edit: Looks like it's solved by adding the -l argument. - This is correct
The 250 block check takes only a second or two. It is asking for the block hash from the coin daemon and checking it against the database. If you feel that is too long it can be adjusted in the configuration file.
There is also a 15 second delay when starting the loader without the new database option. This is to allow the coin daemon time to catch orphans it discovers shortly after calling the loader through block notify.
Both of these functions run once when the loader is started, when you see the count-up they are no longer being called.
As to questions about speed of the initial load, it is hard to answer without actually running it on whatever system you are using. The CCE servers have about a half dozen explorers each running and generally have a CPU usage around 0.8 - 1.5 on a six CPU system. This is running the coin daemons, loaders and web servers.
I recently built a database on a 600K + block chain and the initial load took about 10 hours. The initial load can take some time, but after the initial load is finished it runs smooth and quick.