Commit graph

220 commits

Author SHA1 Message Date
Roman Khimov
7ee8f9c5d8 network: fix networking stalls caused by stale peers
We can leak sending goroutines and stall broadcasts because of already gone
peers that happened to be cached by some s.Peers() user (more than 800 of
these can be seen in nodoka log along with (*Server).run blocking on
CMDGetAddr send):

Feb 10 16:35:15 nodoka neo-go[1563]: goroutine 41 [chan send, 3320 minutes]:
Feb 10 16:35:15 nodoka neo-go[1563]: github.com/CityOfZion/neo-go/pkg/network.(*TCPPeer).putPacketIntoQueue(...)
Feb 10 16:35:15 nodoka neo-go[1563]:         /go/src/github.com/CityOfZion/neo-go/pkg/network/tcp_peer.go:81
Feb 10 16:35:15 nodoka neo-go[1563]: github.com/CityOfZion/neo-go/pkg/network.(*TCPPeer).EnqueueHPPacket(0xc0083d57a0, 0xc017206100, 0x18, 0x40, 0x136a240, 0xc018ef9720)
Feb 10 16:35:15 nodoka neo-go[1563]:         /go/src/github.com/CityOfZion/neo-go/pkg/network/tcp_peer.go:119 +0x98
Feb 10 16:35:15 nodoka neo-go[1563]: github.com/CityOfZion/neo-go/pkg/network.(*Server).iteratePeersWithSendMsg(0xc0000ca000, 0xc0001848a0, 0xcb4550, 0x0)
Feb 10 16:35:15 nodoka neo-go[1563]:         /go/src/github.com/CityOfZion/neo-go/pkg/network/server.go:720 +0x12a
Feb 10 16:35:15 nodoka neo-go[1563]: github.com/CityOfZion/neo-go/pkg/network.(*Server).broadcastHPMessage(...)
Feb 10 16:35:15 nodoka neo-go[1563]:         /go/src/github.com/CityOfZion/neo-go/pkg/network/server.go:731
Feb 10 16:35:15 nodoka neo-go[1563]: github.com/CityOfZion/neo-go/pkg/network.(*Server).run(0xc0000ca000)
Feb 10 16:35:15 nodoka neo-go[1563]:         /go/src/github.com/CityOfZion/neo-go/pkg/network/server.go:203 +0xee4
Feb 10 16:35:15 nodoka neo-go[1563]: github.com/CityOfZion/neo-go/pkg/network.(*Server).Start(0xc0000ca000, 0xc000072c60)
Feb 10 16:35:15 nodoka neo-go[1563]:         /go/src/github.com/CityOfZion/neo-go/pkg/network/server.go:173 +0x2ec
Feb 10 16:35:15 nodoka neo-go[1563]: created by github.com/CityOfZion/neo-go/cli/server.startServer
Feb 10 16:35:15 nodoka neo-go[1563]:         /go/src/github.com/CityOfZion/neo-go/cli/server/server.go:331 +0x476
2020-02-10 18:47:52 +03:00
Roman Khimov
c896e2b731 network: fix minor gofmt issues (spotted by goreportcard) 2020-02-08 15:53:08 +03:00
Roman Khimov
b9b77ac1be network: fix block relaying, don't spit out useless errors
We can only add one block of the given height and we have two competing
goroutines to do that --- consensus and block queue. Whomever adds the block
first shouldn't trigger an error in another one.

Fix block relaying for blocks added via the block queue also, previously one
consensus-generated blocks were broadcasted.
2020-02-06 15:41:56 +03:00
Roman Khimov
70b3839fd0 core/mempool: fix AddBlock and tx pooling concurrency issues
Eliminate races between tx checks and adding them to the mempool, ensure the
chain doesn't change while we're working with the new tx. Ensure only one
block addition attempt could be in progress.
2020-02-06 15:41:52 +03:00
Roman Khimov
f0bb886be3 mempool: make all methods pointer methods
Makes no sense copying the Pool around.
2020-02-04 17:36:11 +03:00
Roman Khimov
e5d0125a3f network: do async broadcast when processing P2P TX
It can lead to some goroutine explosion, but supposedly it's better than
stalling other processing and eventually all of these goroutines should finish
their sends. Note that this doesn't change the behavior for RPC-relayed
transactions that are still waiting for the broadcast to finish ensuring
proper transaction distribution before returning the result to the client.
2020-01-30 14:03:52 +03:00
Roman Khimov
0fcbc697ca network: only tell consensus service about new valid transactions
Transactions can be bad and can duplicate each other, consensus service
shouldn't be bothered with that.
2020-01-30 14:03:52 +03:00
Roman Khimov
fdbaac7a30 network: prevent broadcast queue starving, share time with p2p
Blocked broadcast queue of one peer may affect broadcasting capabilities of
the server, so prevent total blocking of it by p2p queue.
2020-01-30 14:03:52 +03:00
Roman Khimov
b2c4587dad network: fix PeerAddr() for not-yet-handshaked case
If we have already got Version message, we don't need the rest of handshake to
complete before being able to properly answer the PeerAddr() requests. Fixes
some duplicate connections between machines.
2020-01-30 14:03:52 +03:00
Roman Khimov
49bd7aada5 network: log peerCount on connection registration
Symmetrical to disconnects, make it easier to look through the logs.
2020-01-30 14:03:52 +03:00
Roman Khimov
32695b4f40 network: don't unregister connected peers
If we drop connection because we're already connected don't unregister this
connected address because it's connected anyway!
2020-01-30 14:03:52 +03:00
Roman Khimov
eb4ec61b8b network: register connected addr in handleVersionCmd()
Prevent useless attempts to connect to this peer if the peer has already made
a connection to us.
2020-01-30 14:03:52 +03:00
Roman Khimov
9eafec0d1d network: introduce peer-to-peer message queue
This one is designed to give more priority to direct nodes communication, that
is that their messaging would have more priority than generic broadcasts. It
should improve consensus process under TX pressure and allow to handle
pings in time (preventing disconnects).
2020-01-30 14:03:52 +03:00
Roman Khimov
72e4eb7172 network: fix wrong NewPing() parameters
They have the opposite order, height first and nonce second. It was done wrong
in 4e6ed902 and never fixed since. Fixes sending wrong peer state leading to
useless getheaders messages (and disconnects when the other side is lagging
behind).
2020-01-30 14:03:52 +03:00
Roman Khimov
da09fda575 network: add debug on message receival 2020-01-30 14:03:52 +03:00
Roman Khimov
1c28dd2567 network: add message type to disconnect error message
If it was caused by message processing, but only after the handshake to
preserve errIdenticalID and other handshaking errors.
2020-01-30 14:03:52 +03:00
Roman Khimov
b79abd1e27 network: detect duplicate connections before handshake completes
We can have more than one connection attempt in progress and not yet completed
the handshake, so if there is a Version already received we should look it.
2020-01-30 14:03:52 +03:00
Roman Khimov
06c3fbe455 network: rework ping sends, fix overpinging
Our node was too pingy because of wrong timer setups (that divided timeout
Duration by time.Second), it also was wrong in its time calculations (using
UTC time to calculate intervals). At the same time missing block is a
server-wide problem, so it's better solved with server-wide protocol loop.
2020-01-28 17:39:52 +03:00
Roman Khimov
39800aecb3 network: fix getdata handling
It was broken by 0ba6b2a7, the peer only responded with one message instead of
full requested set.
2020-01-28 17:39:52 +03:00
Roman Khimov
9aa5d0cf1e
Merge pull request #615 from nspcc-dev/network-peer-updates
Network peer updates
2020-01-23 13:15:43 +03:00
Roman Khimov
99dfdc19e7 network: drop now useless addrReq queue from the server
Just broadcast a high-priority message to everyone.
2020-01-22 11:28:59 +03:00
Roman Khimov
ea3b76ded1 network: make NewServer return an error, fix #612
It can return nil in two cases, so we're better return an error and handle
it.
2020-01-22 11:17:51 +03:00
Roman Khimov
f2ffffddb7 network: rework broadcasting functions, tune priorities
This gives more priority to anything related to consensus.
2020-01-22 11:01:13 +03:00
Roman Khimov
34b863d645 network: introduce Server's MkMsg()
That wraps NewMessage() for a configured network.
2020-01-21 17:31:51 +03:00
Roman Khimov
1f672e0da7 network: move SendVersion() to the Peer
Only leave server-specific `getVersionMsg()` in the Server, all the other
logic is peer-related.
2020-01-21 17:26:08 +03:00
Roman Khimov
f56383e9c8 network: use p.LastBlockIndex() in requestBlocks()
Always compare to the best known block index, comparing to the StartHeight is
just plain wrong now.
2020-01-20 19:37:17 +03:00
Roman Khimov
2c4ace022e network/config: redesign ping timeout handling a bit
1) Make timeout a timeout, don't do magic ping counts.
2) Drop additional timer from the main peer's protocol loop, create it
   dynamically and make it disconnect the peer.
3) Don't expose the ping counter to the outside, handle more logic inside the
   Peer.

Relates to #430.
2020-01-20 19:37:17 +03:00
Roman Khimov
62092c703d network: use local timestamp to decide when to ping
We don't and we won't have synchronized clocks in the network so the only
timestamp that we can compare our local time with is the one made
ourselves. What this ping mechanism is used for is to recover from missing the
block broadcast, thus it's appropriate for it to trigger after X seconds of
the local time since the last block received.

Relates to #430.
2020-01-20 19:37:17 +03:00
Roman Khimov
a8252ecc05 network: remove wrong ping condition
In reality it will never be true exactly in the case where we want this ping
mechanism to work --- when the node failed to get a block from the net. It
won't get the header either and thus its block height will be equal to header
height. The only moment when this condition is met is when the node does
initial synchronization and this synchronization works just fine without any
pings.

Relates to #430.
2020-01-20 19:37:17 +03:00
Roman Khimov
247cfa4165 network: either request blocks or ping a peer, but not both
It makes to sense to do both actions, pings are made for a different purpose.

Relates to #430.
2020-01-20 19:37:17 +03:00
Roman Khimov
0ba6b2a754 network: introduce peer sending queues
Two queues for high-priority and ordinary messages. Fixes #590. These queues
are deliberately made small to avoid buffer bloat problem, there is gonna be
another queueing layer above them to compensate for that. The queues are
designed to be synchronous in enqueueing, async capabilities are to be added
layer above later.
2020-01-20 17:23:26 +03:00
Roman Khimov
7f0882767c network: remove useless Done() method from the peer
It's internal state of the peer that no one should care about.
2020-01-20 17:23:26 +03:00
Roman Khimov
f39d5d5a10 network: fix unregistration on peer Disconnect
It should always signal to the server, not duplicating this send and not
missing it like it happened in the Server.run().
2020-01-20 17:23:26 +03:00
Roman Khimov
907a236285 network: move per-peer goroutines into the TCPPeer
As they're directly tied to it.
2020-01-20 17:23:26 +03:00
Roman Khimov
32213b1454
Merge pull request #601 from nspcc-dev/refactoring/core
core: refactor out Block, BlockBase and Header, closes #597.
2020-01-20 16:19:20 +03:00
Roman Khimov
bb80ba9b9e
Merge pull request #456 from nspcc-dev/pingpong_430
add ping pong processing
2020-01-20 16:10:29 +03:00
Vsevolod Brekelov
4e6ed9021c network: add ping pong processing
add pingInterval same as used in ref C# implementation with the same logic
add pingTimeout which is used to check whether pong received. If not -- drop the peer.
add pingLimit which is hardcoded to 4 in TCPPeer. It's limit for unsuccessful ping/pong calls (where pong wasn't received in pingTimeout interval)
2020-01-17 13:24:14 +03:00
Evgenii Stratonikov
28183b81d6 mempool: simplify names of exported types
With the move to a separate package, naming can be simplified:
MemPool -> Pool, PoolItem -> Item, PoolItems -> Items.
2020-01-16 10:16:24 +03:00
Evgenii Stratonikov
fed6fba9b6 core: refactor out MemPool 2020-01-16 10:16:24 +03:00
Evgenii Stratonikov
489b88afbb block: rename BlockBase to Base 2020-01-16 10:16:24 +03:00
Evgenii Stratonikov
63c56cca5c core: refactor out Block, BlockBase and Header structs
See #597.
2020-01-16 10:16:24 +03:00
Roman Khimov
7d4d57351e network: fix requestTx() behaviour for consensus service
It wasn't actually requesting transactions but rather sending an inventory
message telling everyone that we have them which is completely wrong and
easily leads to ChangeView that could be avoided.
2020-01-15 14:31:56 +03:00
Roman Khimov
0420d48e56 network: micro-optimize relayInventory
Do less allocations, we're sending the same message with the same payload.
2020-01-15 13:16:09 +03:00
Roman Khimov
7ba5267494
Merge pull request #596 from nspcc-dev/feature/single_node
network: allow single-node privnet setup. Closes #595.
2020-01-14 13:17:04 +03:00
Evgenii Stratonikov
70b23076f8 network: allow single-node privnet setup 2020-01-13 18:01:20 +03:00
Evgenii Stratonikov
5f5d0097e2 network: use TimePerBlock from config 2020-01-13 17:58:12 +03:00
Roman Khimov
2192b6602a network: fix logic and logging for sendVersion() errors
It doesn't make sense trying to read from a broken connection and the
disconnect is to be logger from the server on unregistration.
2020-01-13 17:33:29 +03:00
Evgenii Stratonikov
f3921ea3bd network: fix NPE in logging 2020-01-13 15:06:57 +03:00
Evgenii Stratonikov
9b8b77c9ea network: return error if header message is too big
Big messages can still be processed but only first
2000 headers will be used.
2020-01-10 11:14:29 +03:00
Evgenii Stratonikov
aecdf470e7 cli,pkg: use zap.Logger 2020-01-10 11:14:27 +03:00