Posted by Pieter Wuille
Jul 27, 2012/09:59 UTC
The class of machine referred to in a 2012 post by 'grarpamp' has become useless for Bitcoin due to a disk-based overload issue linked to blk0002.dat, which was created to store more blocks. The problem resulted in forward progress processing turning into losing ground by around 20 blocks a day. Another user, who did not experience the same problem with a more advanced machine, suggested that the issue may be related to the use of a filesystem (ZFS) known for its memory-hungry performance on 32-bit architectures. The backend Bitcoin uses (bdb) performs many I/O operations and writes synchronously to disk, negating any caching from the filesystem. However, improvements are expected with changes being made from bdb to leveldb, and the bitcoin block verifier switching to an ultraprune database layout, allowing faster syncing and pruning.
Thread Summary (30 replies)
May 5 - May 14, 2020
31 messages • 30 replies
TLDR
We’ll email you summaries of the latest discussions from high signal bitcoin sources, like bitcoin-dev, lightning-dev, and Delving Bitcoin.
We'd love to hear your feedback on this project.
Give Feedback