IRC meeting summary for 2016-06-23
- perceived validation slowdowns
perceived validation slowdowns
Gmaxwell noticed very slow reindexing while testing with default dbcache, which got confirmed by sipa who saw similar behavior.
Sipa noticed very slow chainstate writes, however that might be because of his disk setup. We still need to figure out what exactly is behind this, but it’s a good idea to crank up the default value for dbcache and change how it’s allocated as currently a large part goes to leveldb caches, but further benchmarks are needed to find the best values.
Gmaxwell will test against a patched version of 0.12.1 which skips signature checking before block 295000 to see if there’s any regression or this is normal behavior.
These tests where done before compact blocks was merged, so it can’t be affacted by this. Leveldb hasn’t had any changes recently so the only way for leveldb to be the issue is if it has crossed some performance cliff.
Sipa notes the testing has been done with txindex enabled (maintains a full transaction index), and performance might be influenced by this.
- Gmaxwell will benchmark different configurations: 0.12.1 vs master, txindex enabled/disabled, dbcache default/higher and with different cache splits.
Developers are working on a soft fork to introduce segregated witness onto Bitcoin mainnet. Segregated witness (segwit) allows transaction signature data to be stored outside of the data hashed to produce transaction identifiers, removing all known forms of third-party malleability, allowing full nodes to compile the current UTXO set without downloading all signatures, and laying the groundwork for fraud proofs that can allow lightweight (SPV) clients to help enforce more of the consensus rules. The segwit soft fork also allows miners to substitute 1 byte of block space with 4 bytes of segwit data, increasing transaction capacity for wallets that use segwit. segregated witness BIPs: BIP141, BIP142, BIP143, BIP144 and BIP145
Segwit has been rebased after the compact blocks merge.
Sipa has been running compact blocks + segwit and doesn’t see an impact on memory usage.
Everyone is in favor of merging segwit. (was merged after the meeting)
Gmaxwell suggests we’d put out ‘testnet binaries’, which has testnet enabled by default, right away to get more people testing. Up to now segwit testing was mostly done by technical people that aren’t likely to be confused by things like status changes in the UI or whatnot.
The activation of segwit on testnet was a very useful testing exercise as it was in an environment with most nodes not upgraded.
- Merge segwit
- Release testnet-default/only binaries
|wumpus||Wladimir van der Laan|
This summary was compiled without input from any of the participants in the discussion, so any errors are the fault of the summary author and not the discussion participants.