[bisq-network/bisq] Optimize DaoState snapshot behaviour (#5609)

chimp1984 notifications at github.com
Fri Jul 9 21:48:46 CEST 2021

> I do notice much more disk activity now as it is persisting to disk rather than remaining in memory

The persistence of dao state has not changed. Only that we do not keep the daostate used for persisting in memory longer as needed.

I have started a full sync from genesis with a full node, still missing 10 000 blocks, but so far looks all as expected.

@sqrrm From my side you can merge it. Other potential optimizations can be done later.

Observations from dao parsing performance:
Initially parsing is very fast (10ms), getting the block via rpc is rather slow (about 1 sec.) - what I remember it was faster with the pre-segwit rpc library, but not sure. But a much bigger improvement might be done to request not each single block but a batch.
Over time when dao state gets bigger parsing gets slower as well. Not at block 680000 its about 200-300 ms, though that might be also because of the very long running process (runs since > 48 h). But there is likely more optimization options in that area - usually to add a hashmap as cache to avoid iterations. 

You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.bisq.network/pipermail/bisq-github/attachments/20210709/f69c1d2a/attachment.htm>

More information about the bisq-github mailing list