nimbus-eth1/nimbus/db/ledger
Jacek Sieka f294d1e086
Clear account cache after each block (#2411)
When processing long ranges of blocks, the account cache grows unbounded
which cause huge memory spikes.

Here, we move the cache to a second-level cache after each block - the
second-level cache is cleared on the next block after that which creates
a simple LRU effect.

There's a small performance cost of course, though overall the freed-up
memory can now be reassigned to the rocksdb row cache which not only
makes up for the loss but overall leads to a performance increase.

The bump to 2gb of rocksdb row cache here needs more testing but is
slightly less and loosely basedy on the savings from this PR and the
circular ref fix in #2408 - another way to phrase this is that it's
better to give rocksdb more breathing room than let the memory sit
unused until circular ref collection happens ;)
2024-06-25 07:30:32 +02:00
..
base Coredb maintenance (#2398) 2024-06-19 14:13:12 +00:00
README.md Aristo resume off line syncing on pre loaded database (#2203) 2024-05-22 13:41:14 +00:00
accounts_ledger.nim Clear account cache after each block (#2411) 2024-06-25 07:30:32 +02:00
base.nim Clear account cache after each block (#2411) 2024-06-25 07:30:32 +02:00
base_iterators.nim Simplify AccountsLedgerRef complexity (#2239) 2024-05-29 13:06:49 +02:00
distinct_ledgers.nim Coredb maintenance (#2398) 2024-06-19 14:13:12 +00:00

README.md

The file accounts_cache.nim has been relocated

Background

The new LedgerRef module unifies different implementations of the legacy accounts_cache. It is intended to be used as new base method for all of the AccountsCache implementations. The legacy accounts_cache version has been removed.

This was needed to accomodate for different CoreDb API paradigms. While the overloaded legacy AccountsCache implementation is just a closure based wrapper around the accounts_cache module, the overloaded AccountsLedgerRef is a closure based wrapper around the accounts_ledger module with the new CoreDb API returning Result[] values and saparating the meaning of trie root hash and trie root reference.

This allows to use the legacy hexary database (with the new CoreDb API) as well as the Aristo database (only supported on new API.)

Instructions

Legacy notation LedgerRef replacement Comment
import accounts_cache import ledger preferred method,
AccountsCache.init(..) AccountsCache.init(..) wraps AccountsCache
methods
or
import ledger/accounts_cache stay with legacy
AccountsCache.init(..) version of
AccountsCache
--
fn(ac: AccountsCache) fn(ac: LedgerRef) function example for
preferred wrapper
or method
fn(ac: AccountsCache) with legacy version,
no change here

The constructor decides which CoreDb API is to be used

Legacy API constructor new API Constructor
import ledger import ledger
let w = AccountsCache.init(..) let w = AccountsLedgerRef.init(..)