Author: Mike Hearn 2013-03-12 09:10:15
Published on: 2013-03-12T09:10:15+00:00
The first part of the context discusses issues with lock exhaustion in BDB on some 0.7 nodes, despite having processed 1mb sized blocks on the testnet. This is causing transactions to stack up in the memory pool again, and there are no bounds on the memory pool size, which could lead to nodes dying with OOM failures. Additionally, users have no way of finding out the current highest fees/bids for block space or changing the fee on sent transactions, which means Bitcoin will soon start breaking for most users who don't fully understand the system.As a result, the author suggests two alternatives - blocking or down-prioritizing SatoshiDice transactions at the network level, which could help avoid mempool exhaustion, or performing a crash hard fork as soon as possible, potentially lifting the 1mb limit at the same time. In another email, Pieter Wuille addresses a problem with lock entries in the BDB database that caused 0.7 nodes to get stuck around block 225430, while 0.8 nodes did not face this issue. The author advises miners to revert to 0.7 until the majority of mining power is back on the old chain, and merchants using 0.8 to stop processing transactions until both sides have switched to the same chain again. Upgrading to 0.8 or manually setting the number of lock objects higher in the database were suggested as immediate solutions.
Updated on: 2023-06-06T10:44:43.366313+00:00