Instead of trying to connect to every peer in the peer store when
we start a node, only connect to the peers that have been marked
with a `keep-alive` tag.
Uses peer tag values to select low-value connections to prune when we have too many connections open.
BREAKING CHANGE: `connectionManager.peerValue` has been removed, use `peerStore.tagPeer` instead
* feat: limit protocol streams per-connection
Uses the `maxInboundStreams` and `maxOutboundStreams` of the `registrar.handle`
opts to limit the number of concurrent streams open on each connection
on a per-protocol basis.
Both values default to 1 so some tuning will be necessary to set
appropriate values for some protocols.
* chore: make error codes consistent
* chore: fix up examples
There are a few places in the codebase where we send/receive data from the network without timeouts/abort controllers which means the user has to wait for the underlying socket to timeout which can take a long time depending on the platform, if at all.
This change ensures we can time out while running identify (both flavours), ping and fetch and adds tests to ensure there are no regressions.
**Motivation**
In lodestar, when we handle "peer:connect" event, we dial the peer which gives another "peer:connect" event and it causes other issues
**Motivation**
In `onConnect` function, "peer:connect" event should be emitted after we add connection to the `connections` map so that when app dial the peer in "peer:connect" event handler, it uses the same/existing connection
Instead of making the `.dht` and `.pubsub` properties optional, use dummy implementations that throw exceptions if they are not configured.
This way we don't have to null guard everywhere they are accessed.
Addresses PR comments from #1172 - fixes syntax of examples in docs, adds the transport manager to the exports map and renames fault tolerance enum for consistency.
Converts this module to typescript.
- Ecosystem modules renamed from (e.g.) `libp2p-tcp` to `@libp2p/tcp`
- Ecosystem module now have named exports
- Configuration has been updated, now pass instances of modules instead of classes:
- Some configuration keys have been renamed to make them more descriptive. `transport` -> `transports`, `connEncryption` -> `connectionEncryption`. In general where we pass multiple things, the key is now plural, e.g. `streamMuxer` -> `streamMuxers`, `contentRouting` -> `contentRouters`, etc. Where we are configuring a singleton the config key is singular, e.g. `connProtector` -> `connectionProtector` etc.
- Properties of the `modules` config key have been moved to the root
- Properties of the `config` config key have been moved to the root
```js
// before
import Libp2p from 'libp2p'
import TCP from 'libp2p-tcp'
await Libp2p.create({
modules: {
transport: [
TCP
],
}
config: {
transport: {
[TCP.tag]: {
foo: 'bar'
}
},
relay: {
enabled: true,
hop: {
enabled: true,
active: true
}
}
}
})
```
```js
// after
import { createLibp2p } from 'libp2p'
import { TCP } from '@libp2p/tcp'
await createLibp2p({
transports: [
new TCP({ foo: 'bar' })
],
relay: {
enabled: true,
hop: {
enabled: true,
active: true
}
}
})
```
- Use of `enabled` flag has been reduced - previously you could pass a module but disable it with config. Now if you don't want a feature, just don't pass an implementation. Eg:
```js
// before
await Libp2p.create({
modules: {
transport: [
TCP
],
pubsub: Gossipsub
},
config: {
pubsub: {
enabled: false
}
}
})
```
```js
// after
await createLibp2p({
transports: [
new TCP()
]
})
```
- `.multiaddrs` renamed to `.getMultiaddrs()` because it's not a property accessor, work is done by that method to calculate announce addresses, observed addresses, etc
- `/p2p/${peerId}` is now appended to all addresses returned by `.getMultiaddrs()` so they can be used opaquely (every consumer has to append the peer ID to the address to actually use it otherwise). If you need low-level unadulterated addresses, call methods on the address manager.
BREAKING CHANGE: types are no longer hand crafted, this module is now ESM only
We have a peerstore that keeps all data for all observed peers in memory with no eviction.
This is fine when you don't discover many peers but when using the DHT you encounter a significant number of peers so our peer storage grows and grows over time.
We have a persistent peer store, but it just periodically writes peers into the datastore to be read at startup, still keeping them in memory.
It also means a restart doesn't give you any temporary reprieve from the memory leak as the previously observed peer data is read into memory at startup.
This change refactors the peerstore to use a datastore by default, reading and writing peer info as it arrives. It can be configured with a MemoryDatastore if desired.
It was necessary to change the peerstore and *book interfaces to be asynchronous since the datastore api is asynchronous.
BREAKING CHANGE: `libp2p.handle`, `libp2p.registrar.register` and the peerstore methods have become async
The `AbortController` class is supported by browsers and node 14+ - we only support node 16+ (e.g. LTS+Current) so the `abort-controller` module isn't needed any more.
Looks like this project stopped running the `test:node` npm script when it was migrated to gh actions.
Re-enable it and fix all the related test failures.