You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Note that bitcoin/bitcoin#25454 apparently also prevents a fingerprinting attack described in bitcoin/bitcoin#24571. (I'm not sure I understand the threat model under which that is considered an attack, though, tbh. There are many ways to fingerprint nodes.)
We initially tried to backport bitcoin/bitcoin#10345 as part of #6231, but that was reverted in #6276 due to a memory usage regression (#6268). If we want the headers-sync timer then I think there are two options:
a) implement the improvement to batch write memory tracking I suggested here and then (effectively) reapply #6231;
b) extract the addition of the timer (last three commits of #6231), without changing sync to headers-first. We would have to consider whether the timer semantically depends on the headers-first behaviour.
A recent zebrad bug (ZcashFoundation/zebra#8907), found when trying to repair the testnet after its NU6 activation, drew attention to the fact that zcashd is missing some improvements to the efficiency and robustness of header sync from Bitcoin Core. These were implemented in bitcoin/bitcoin#10345, bitcoin/bitcoin#25454, bitcoin/bitcoin#25720, and bitcoin/bitcoin#25717.
Backporting bitcoin/bitcoin#25720 would fix the issue that motivated the zebrad change that has been problematic on testnet (ZcashFoundation/zebra#1439).
The text was updated successfully, but these errors were encountered: