This fixes an issue where the mumbai testnet node struggle to find
peers. Before this fix in general test peer numbers are typically around
20 in total between eth66, eth67 and eth68. For new peers some can
struggle to find even a single peer after days of operation.
These are the numbers after 12 hours or running on a node which
previously could not find any peers: eth66=13, eth67=76, eth68=91.
The root cause of this issue is the following:
- A significant number of mumbai peers around the boot node return
network ids which are different from those currently available in the
DHT
- The available nodes are all consequently busy and return 'too many
peers' for long periods
These issues case a significant number of discovery timeouts, some of
the queries will never receive a response.
This causes the discovery read loop to enter a channel deadlock - which
means that no responses are processed, nor timeouts fired. This causes
the discovery process in the node to stop. From then on it just
re-requests handshakes from a relatively small number of peers.
This check in fixes this situation with the following changes:
- Remove the deadlock by running the timer in a separate go-routine so
it can run independently of the main request processing.
- Allow the discovery process matcher to match on port if no id match
can be established on initial ping. This allows subsequent node
validation to proceed and if the node proves to be valid via the
remainder of the look-up and handshake process it us used as a valid
peer.
- Completely unsolicited responses, i.e. those which come from a
completely unknown ip:port combination continue to be ignored.
-
Improve p2p error handling to propagate errors
from the origin up the call chain the Server peer removal code
using a new PeerError type containing a DiscReason and a more detailed
description.
The origin can be tracked down using PeerErrorCode (code) and DiscReason
(reason)
which looks like this in the log:
> [TRACE] [08-28|16:33:40.205] Removing p2p peer peercount=0
url=enode://d399f4b...@1.2.3.4:30303 duration=6.901ms
err="PeerError(code=remote disconnect reason, reason=too many peers,
err=<nil>, message=Peer.run got a remote DiscReason)"
* Switch peerId from 256 to 512 bit (as in stable)
* go mod tidy
* Fix some tests
* Fixed
* Fixes
* Fix tests
* Update to erigon-lib main
Co-authored-by: Alex Sharp <alexsharp@Alexs-MacBook-Pro.local>
Co-authored-by: Alexey Sharp <alexeysharp@Alexeys-iMac.local>
This change moves the RLPx protocol implementation into a separate package,
p2p/rlpx. The new package can be used to establish RLPx connections for
protocol testing purposes.
Co-authored-by: Felix Lange <fjl@twurst.com>
# Conflicts:
# p2p/rlpx/rlpx.go
# p2p/rlpx/rlpx_test.go
# p2p/server_test.go
* p2p: new dial scheduler
This change replaces the peer-to-peer dial scheduler with a new and
improved implementation. The new code is better than the previous
implementation in two key aspects:
- The time between discovery of a node and dialing that node is
significantly lower in the new version. The old dialState kept
a buffer of nodes and launched a task to refill it whenever the buffer
became empty. This worked well with the discovery interface we used to
have, but doesn't really work with the new iterator-based discovery
API.
- Selection of static dial candidates (created by Server.AddPeer or
through static-nodes.json) performs much better for large amounts of
static peers. Connections to static nodes are now limited like dynanic
dials and can no longer overstep MaxPeers or the dial ratio.
* p2p/simulations/adapters: adapt to new NodeDialer interface
* p2p: re-add check for self in checkDial
* p2p: remove peersetCh
* p2p: allow static dials when discovery is disabled
* p2p: add test for dialScheduler.removeStatic
* p2p: remove blank line
* p2p: fix documentation of maxDialPeers
* p2p: change "ok" to "added" in static node log
* p2p: improve dialTask docs
Also increase log level for "Can't resolve node"
* p2p: ensure dial resolver is truly nil without discovery
* p2p: add "looking for peers" log message
* p2p: clean up Server.run comments
* p2p: fix maxDialedConns for maxpeers < dialRatio
Always allocate at least one dial slot unless dialing is disabled using
NoDial or MaxPeers == 0. Most importantly, this fixes MaxPeers == 1 to
dedicate the sole slot to dialing instead of listening.
* p2p: fix RemovePeer to disconnect the peer again
Also make RemovePeer synchronous and add a test.
* p2p: remove "Connection set up" log message
* p2p: clean up connection logging
We previously logged outgoing connection failures up to three times.
- in SetupConn() as "Setting up connection failed addr=..."
- in setupConn() with an error-specific message and "id=... addr=..."
- in dial() as "Dial error task=..."
This commit ensures a single log message is emitted per failure and adds
"id=... addr=... conn=..." everywhere (id= omitted when the ID isn't
known yet).
Also avoid printing a log message when a static dial fails but can't be
resolved because discv4 is disabled. The light client hit this case all
the time, increasing the message count to four lines per failed
connection.
* p2p: document that RemovePeer blocks
The dialer limits itself to one attempt every 30s. Apply the same limit
in Server and reject peers which try to connect too eagerly. The check
against the limit happens right after accepting the connection.
Further changes in this commit ensure we pass the Server logger
down to Peer instances, discovery and dialState. Unit test logging now
works in all Server tests.
Package p2p/enode provides a generalized representation of p2p nodes
which can contain arbitrary information in key/value pairs. It is also
the new home for the node database. The "v4" identity scheme is also
moved here from p2p/enr to remove the dependency on Ethereum crypto from
that package.
Record signature handling is changed significantly. The identity scheme
registry is removed and acceptable schemes must be passed to any method
that needs identity. This means records must now be validated explicitly
after decoding.
The enode API is designed to make signature handling easy and safe: most
APIs around the codebase work with enode.Node, which is a wrapper around
a valid record. Going from enr.Record to enode.Node requires a valid
signature.
* p2p/discover: port to p2p/enode
This ports the discovery code to the new node representation in
p2p/enode. The wire protocol is unchanged, this can be considered a
refactoring change. The Kademlia table can now deal with nodes using an
arbitrary identity scheme. This requires a few incompatible API changes:
- Table.Lookup is not available anymore. It used to take a public key
as argument because v4 protocol requires one. Its replacement is
LookupRandom.
- Table.Resolve takes *enode.Node instead of NodeID. This is also for
v4 protocol compatibility because nodes cannot be looked up by ID
alone.
- Types Node and NodeID are gone. Further commits in the series will be
fixes all over the the codebase to deal with those removals.
* p2p: port to p2p/enode and discovery changes
This adapts package p2p to the changes in p2p/discover. All uses of
discover.Node and discover.NodeID are replaced by their equivalents from
p2p/enode.
New API is added to retrieve the enode.Node instance of a peer. The
behavior of Server.Self with discovery disabled is improved. It now
tries much harder to report a working IP address, falling back to
127.0.0.1 if no suitable address can be determined through other means.
These changes were needed for tests of other packages later in the
series.
* p2p/simulations, p2p/testing: port to p2p/enode
No surprises here, mostly replacements of discover.Node, discover.NodeID
with their new equivalents. The 'interesting' API changes are:
- testing.ProtocolSession tracks complete nodes, not just their IDs.
- adapters.NodeConfig has a new method to create a complete node.
These changes were needed to make swarm tests work.
Note that the NodeID change makes the code incompatible with old
simulation snapshots.
* whisper/whisperv5, whisper/whisperv6: port to p2p/enode
This port was easy because whisper uses []byte for node IDs and
URL strings in the API.
* eth: port to p2p/enode
Again, easy to port because eth uses strings for node IDs and doesn't
care about node information in any way.
* les: port to p2p/enode
Apart from replacing discover.NodeID with enode.ID, most changes are in
the server pool code. It now deals with complete nodes instead
of (Pubkey, IP, Port) triples. The database format is unchanged for now,
but we should probably change it to use the node database later.
* node: port to p2p/enode
This change simply replaces discover.Node and discover.NodeID with their
new equivalents.
* swarm/network: port to p2p/enode
Swarm has its own node address representation, BzzAddr, containing both
an overlay address (the hash of a secp256k1 public key) and an underlay
address (enode:// URL).
There are no changes to the BzzAddr format in this commit, but certain
operations such as creating a BzzAddr from a node ID are now impossible
because node IDs aren't public keys anymore.
Most swarm-related changes in the series remove uses of
NewAddrFromNodeID, replacing it with NewAddr which takes a complete node
as argument. ToOverlayAddr is removed because we can just use the node
ID directly.
The most visible change is event-based dialing, which should be an
improvement over the timer-based system that we have at the moment.
The dialer gets a chance to compute new tasks whenever peers change or
dials complete. This is better than checking peers on a timer because
dials happen faster. The dialer can now make more precise decisions
about whom to dial based on the peer set and we can test those
decisions without actually opening any sockets.
Peer management is easier to test because the tests can inject
connections at checkpoints (after enc handshake, after protocol
handshake).
Most of the handshake stuff is now part of the RLPx code. It could be
exported or move to its own package because it is no longer entangled
with Server logic.
The returned reason is currently not used except for the log
message. This change makes the log messages a bit more useful.
The handshake code also returns the remote reason.
There were multiple synchronization issues in the disconnect handling,
all caused by the odd special-casing of Peer.readLoop errors. Remove the
special handling of read errors and make readLoop part of the Peer
WaitGroup.
Thanks to @Gustav-Simonsson for pointing at arrows in a diagram
and playing rubber-duck.
Message encoding functions have been renamed to catch any uses.
The switch to the new encoder can cause subtle incompatibilities.
If there are any users outside of our tree, they will at least be
alerted that there was a change.
NewMsg no longer exists. The replacements for EncodeMsg are called
Send and SendItems.
With RLPx frames, the message code is contained in the
frame and is no longer part of the encoded data.
EncodeMsg, Msg.Decode have been updated to match.
Code that decodes RLP directly from Msg.Payload will need
to change.
The diff is a bit bigger than expected because the protocol handshake
logic has moved out of Peer. This is necessary because the protocol
handshake will have custom framing in the final protocol.
Overview of changes:
- ClientIdentity has been removed, use discover.NodeID
- Server now requires a private key to be set (instead of public key)
- Server performs the encryption handshake before launching Peer
- Dial logic takes peers from discover table
- Encryption handshake code has been cleaned up a bit
- baseProtocol is gone because we don't exchange peers anymore
- Some parts of baseProtocol have moved into Peer instead
...and make it a top-level function instead.
The original idea behind having EncodeMsg in the interface was that
implementations might be able to encode RLP data to their underlying
writer directly instead of buffering the encoded data. The encoder
will buffer anyway, so that doesn't matter anymore.
Given the recent problems with EncodeMsg (copy-pasted implementation
bug) I'd rather implement once, correctly.
