05e7ac0360
Pass the CMake bit width setting to compile flags for blockchain_import and blockchain_converter. For LMDB on 32-bit, hyc has found that batch size of 100 appears to be a good default. |
||
|---|---|---|
| .. | ||
| blockchain_converter.cpp | ||
| blockchain_dump.cpp | ||
| blockchain_export.cpp | ||
| blockchain_import.cpp | ||
| blockchain_utilities.h | ||
| blocksdat_file.cpp | ||
| blocksdat_file.h | ||
| bootstrap_file.cpp | ||
| bootstrap_file.h | ||
| bootstrap_serialization.h | ||
| CMakeLists.txt | ||
| cn_deserialize.cpp | ||
| fake_core.h | ||
| README.md | ||
Monero Blockchain Utilities
Copyright (c) 2014-2016, The Monero Project
Introduction
The blockchain utilities allow one to convert an old style blockchain.bin file
to a new style database. There are two ways to upgrade an old style blockchain:
The recommended way is to run a blockchain_export, then blockchain_import.
The other way is to run blockchain_converter. In both cases, you will be left
with a new style blockchain.
For importing into the LMDB database, compile with DATABASE=lmdb
e.g.
DATABASE=lmdb make release
This is also the default compile setting on the master branch.
By default, the exporter will use the original in-memory database (blockchain.bin) as its source.
This default is to make migrating to an LMDB database easy, without having to recompile anything.
To change the source, adjust SOURCE_DB in src/blockchain_utilities/bootstrap_file.h according to the comments.
Usage:
See also each utility's "--help" option.
Export an existing in-memory database
$ blockchain_export
This loads the existing blockchain, for whichever database type it was compiled for, and exports it to $MONERO_DATA_DIR/export/blockchain.raw
Import the exported file
$ blockchain_import
This imports blocks from $MONERO_DATA_DIR/export/blockchain.raw (exported using the blockchain_export tool as described above)
into the current database.
Defaults: --batch on, --batch size 20000, --verify on
Batch size refers to number of blocks and can be adjusted for performance based on available RAM.
Verification should only be turned off if importing from a trusted blockchain.
If you encounter an error like "resizing not supported in batch mode", you can just re-run
the blockchain_import command again, and it will restart from where it left off.
## use default settings to import blockchain.raw into database
$ blockchain_import
## fast import with large batch size, verification off
$ blockchain_import --batch-size 100000 --verify off
## LMDB flags can be set by appending them to the database type:
## flags: nosync, nometasync, writemap, mapasync
$ blockchain_import --database lmdb#nosync
$ blockchain_import --database lmdb#nosync,nometasync
Blockchain converter with batching
blockchain_converter has also been updated and includes batching for faster writes. However, on lower RAM systems, this will be slower than using the exporter and importer utilities. The converter needs to keep the blockchain in memory for the duration of the conversion, like the original bitmonerod, thus leaving less memory available to the destination database to operate.
$ blockchain_converter --batch on --batch-size 20000