Asynq 是一个 Go 库，用于对任务进行排队并与工作线程异步处理它们。它由Redis支持，旨在可扩展且易于入门。

Asynq 工作原理的高级概述：

• 客户端将任务放入队列
• 服务器从队列中取出任务并为每个任务启动一个工作协程
• 任务由多个worker同时处理

任务队列用作跨多台机器分配工作的机制。一个系统可以由多个工作服务器和代理组成，从而实现高可用性和水平扩展。

https://github.com/hibiken/asynq/tree/master

ZK-COUNTERS TABLE

Dynamic zk-counters

In the following doc, we calculate the cost of processing the opcode. It's important to also add the cost of calculating the cost

EXP

Inputs:
1- a: integer base.
2- exponent: integer exponent.

dynamic_gas = 50 * exponent_byte_size
We need to calculate the exponent byte size to get the gas cost. The counters cost is dynamic but we can't calculate the cost without consuming counters in a dynamic way. We need to find a way to get the size of the exponent in a constant manner or handle the counters limitations from the zkasm.
Maximum setted: Maxmimun byte syze = 256 bytes. Max counters = 256 * (2A + 4B) + 1B = 512A + 1025B

SHA3

L = input length
L/32 = A
L % 32 > 0 ? true -> B = 1, false -> B = 0

cnt_arith = 2 + B6
cnt_binary = 2 + A + B9
cnt_keccak = 1

CALLDATALOAD

L = byte offset in the calldata.
L/32 = A
L % 32 > 0 ? true -> B = 1, false -> B = 0

counters = divARITH + B*(SHLarith + SHRarith)

CALLDATACOPY

L = Length to copy.
L/32 = A
L % 32 > 0 ? true -> B = 1, false -> B = 0

counters = A(divARITH + SHLarith + SHRarith + MSTORE32) + SHLarith2 + SHRarith + MSTOREX

CODESIZE

counters = SLOAD

CODECOPY

L = bytes to copy.
if is createContract -> counters = CALLDATACOPY
else -> counters = LT + L*(LT + MEM_ALIGN_WR8)

EXTCODESIZE

counters = SLOAD

EXTCODECOPY

L = bytes to copy.
if is createContract -> counters = CALLDATACOPY
else -> counters = LT + L*(LT + MEM_ALIGN_WR8)

RETURNDATACOPY

L = Length to copy.
L/32 = A
L % 32 > 0 ? true -> B = 1, false -> B = 0
counters = 2EQ + LT + divARITH + mulARITH + A(MLOAD32 + MSTORE32) + B*(MLOADX + MSTOREX)

EXTCODEHASH

counters = SLOAD

SELFBALANCE

counters = SLOAD

MLOAD

counters = MLOAD32

MSTORE

counters = MSTORE32

SLOAD

counters = SLOAD

SSTORE

Cant calculate

JUMPI

isCreateContract ? true -> A = 1, false -> A = 0
isCreate ? true -> B = 1, false -> B = 0

counters = EQ + A(B(MLOADX + SHRarith) + (1-B)(EQ)) + (1-A)(EQ)

JUMP

isCreateContract ? true -> A = 1, false -> A = 0
isCreate ? true -> B = 1, false -> B = 0

counters = EQ + A(B(MLOADX + SHRarith) + (1-B)(EQ)) + (1-A)(EQ)

LOG

L = byte size to copy.
L/32 = A
L % 32 > 0 ? true -> B = 1, false -> B = 0
counters = AMLOAD32 + BMLOADX

CREATE

counters = computeGasSendCall + copySP + SLOAD + SSTORE + getLenBytes

CALL

argsLengthCall + retLength == 0 ? true -> A = 1, false -> A = 0
argsOffsetCall > memLength ? true -> B = 1, false -> B = 0
counters = addARITH + EQ + (1-A)(LT +BsaveMem ) + LT + isEmptyAccount + computeGasSendCall + copySP

CALLCODE

counters = 2EQ + LT2 + computeGasSendCall + copySP

DELEGATECALL

counters = 2EQ + LT2 + computeGasSendCall + copySP

CREATE2

counters = computeGasSendCall + copySP + SLOAD + SSTORE + getLenBytes

STATICCALL

counters = 2EQ + LT2 + computeGasSendCall + copySP

REGS TABLE

Dynamic regs

SSTORE

SLOAD

Should check how SLOAD is implemented

Functions TABLE

Dynamic functions

copySP

It depends on the stack size.
L = stack length
L/32 = A
L % 32 > 0 ? true -> B = 1, false -> B = 0
counters = MLOAD32 * (A + B)

MLOAD32

MSTORE32

MSTOREX

MLOADX

L = bytes length
L > 0 ? true -> A = 1, false -> A = 0
isMSTOREX ? true -> B = 1, false -> B = 0
counters = LT + B(2SHRarith + 2SHLarith) + (1-B)(C(2SHLarith + 4SHRarith) + (1-C)(2SHRarith + 2SHLarith) + MEM_ALIGN_WR)
MAX:
counters = 192A + 193B + 2MA

SHRarith

SHLarith

A -> bytes to shift
D -> times to shift (A << D)
E -> D > 256? true = 1, false = 0
counters= ARITH + EQ + (1-E) (D(LT + ARITH)) = 1A + 1B + (1-E)(D(1A + 1B)
MAX:
counters = 32A + 32B

isEmptyAccount

isNotPrecompiled ? true -> A = 1, false -> A = 0
zeroBalance ? true -> B = 1, false -> B = 0
zeroNonce ? true -> C = 1, false -> C = 0

counters = LT + A(SLOAD + LT + B(SLOAD + LT + C*(SLOAD + LT)))

https://github.com/0xPolygonHermez/zkevm-rom/blob/main/docs/opcode-cost-zk-counters.md

开源地址：https://github.com/Scratch-net/ethproxy

概述

Ethproxy 是一项为Cloudflare gateway支持的 API 方法 eth_getBlockByNumber 提供 LRU 缓存的服务。

• 它使用CCache作为缓存引擎。
• Singleflight可防止同时请求单个块
• 使用envconfig进行配置
• 最新 20 个块的默认 TTL 为 5 秒。从 20 到 1000 的区块的 TTL=“距最新的距离”* 5s
• 所有距离最新区块超过 1000 的区块都被认为是不可变的
• 请求过期的缓存项不会阻止服务。它在后台刷新请求的块
• 实现优雅关闭
• 短绒支持
• Docker 镜像构建。由于多阶段静态构建，生成的图像大小小于 6 mb

Blob Archiver 是一种存档并允许查询信标链中所有历史 Blob 的服务。它由两个部分组成：

Archiver - 跟踪信标链并将 blob 写入存储后端
API - 实现 blob sidecars API，允许客户端从存储后端检索 blob

贮存

目前支持两种存储选项：

磁盘存储 - Blob 写入磁盘的目录中
S3 存储 - Blob 写入 S3 存储桶（或兼容服务）
BLOB_API_DATA_STORE您可以通过将和设置BLOB_ARCHIVER_DATA_STORE为file或来控制使用哪个存储后端s3。

后端s3还可以使用（例如）Google Cloud Storage 存储桶（此处的说明）。

数据有效性

目前，归档器和 API 不验证信标节点的数据。因此，信任 Beacon 节点或验证客户端中的数据非常重要。 向归档器和 API 添加数据验证存在一个未解决的问题。

开源地址：https://github.com/base-org/blob-archiver

location =/aaa {
allow 127.0.0.0/24;
allow 192.168.0.0/16;
allow 172.16.0.0/12;
allow 10.0.0.0/8;
deny all;
}
/aaa为需要禁止访问的路径

https://blog.csdn.net/gjy11011/article/details/124982884