Crates.io | starknet-lb |
lib.rs | starknet-lb |
version | |
source | src |
created_at | 2024-12-08 00:26:17.8872 |
updated_at | 2024-12-08 02:11:07.142856 |
description | Pending block-aware Starknet-native RPC load balancer |
homepage | https://github.com/xJonathanLEI/starknet-lb |
repository | https://github.com/xJonathanLEI/starknet-lb |
max_upload_size | |
id | 1475967 |
Cargo.toml error: | TOML parse error at line 19, column 1 | 19 | autolib = false | ^^^^^^^ unknown field `autolib`, expected one of `name`, `version`, `edition`, `authors`, `description`, `readme`, `license`, `repository`, `homepage`, `documentation`, `build`, `resolver`, `links`, `default-run`, `default_dash_run`, `rust-version`, `rust_dash_version`, `rust_version`, `license-file`, `license_dash_file`, `license_file`, `licenseFile`, `license_capital_file`, `forced-target`, `forced_dash_target`, `autobins`, `autotests`, `autoexamples`, `autobenches`, `publish`, `metadata`, `keywords`, `categories`, `exclude`, `include` |
size | 0 |
Pending block-aware Starknet-native JSON-RPC load balancer
Running multiple Starknet full nodes but having trouble making sure requests don't hit staled instances? starknet-lb
is here to rescue! Built from scratch for Starknet, starknet-lb
is aware of Starknet's PENDING
block to make the most informed decision on traffic distribution.
Not only is starknet-lb
aware of the existence of the PENDING
block, it also knows that PENDING
blocks of the same height can still be different, where ones with more transactions are considered to be more up-to-date.
[!NOTE]
starknet-lb
is in early development. Currently, it only sends traffic to instances with the most up-to-date pending block, ensuring access to the latest network state at the cost of resource under-utilization. This should be improved in the future with configurable, context-based stalenss policy that's able to utilize staled instances to serve historical data (e.g. an instance that's 10 blocks behind chain head should have no problem serving astarknet_call
request on a block from 5 days ago).
Install the crate directly from crates.io:
cargo install --locked starknet-lb
Alternatively, use the pre-built Docker images from Docker Hub.
Currently, the load balancer can only be configured with command line options. To set an upstream, use --upstream <SPEC>
, which can be used multiple times to add multiple upstream specifications.
Two types of upstreams are supported:
Raw URL: Simply supply the full URL of the upstream endpoint as the value. For example: --upstream https://starknet-mainnet.public.blastapi.io/rpc/v0_7
.
DNS: Use the dns:
prefix to set a DNS name that will be resolved into actual hosts. This can be very useful in environments involving auto-scaling (e.g. Kubernetes), where the list of upstream hosts is not static.
An simple example of setting a DNS upstream is --upstream dns:pathfinder.default.svc.cluster.local
.
You can optionally set a port number if the upstream nodes are not serving from the default port of 9545
. For example: --upstream dns:pathfinder.default.svc.cluster.local:8080
.
It's also possible to specify a URL path to apply to all the resolved hosts. To use a path of /rpc/v0_7
: --upstream dns:pathfinder.default.svc.cluster.local/rpc/v0_7
.
The two optional components can also be combined. Here's an example of using both in the same specification: --upstream dns:pathfinder.default.svc.cluster.local:8080/rpc/v0_7
.
[!NOTE]
The load balancer is a work in progress. Specifically, DNS hostnames are only resolved at startup and are not continuously monitored. This should be implemented in the future.
Licensed under either of
at your option.