nim-libp2p/libp2p/protocols/pubsub/pubsubpeer.nim

240 lines
8.0 KiB
Nim
Raw Normal View History

2019-09-10 02:15:52 +00:00
## Nim-LibP2P
2019-09-24 17:48:23 +00:00
## Copyright (c) 2019 Status Research & Development GmbH
2019-09-10 02:15:52 +00:00
## Licensed under either of
## * Apache License, version 2.0, ([LICENSE-APACHE](LICENSE-APACHE))
## * MIT license ([LICENSE-MIT](LICENSE-MIT))
## at your option.
## This file may not be copied, modified, or distributed except according to
## those terms.
import std/[sequtils, strutils, tables, hashes]
2020-06-11 18:09:34 +00:00
import chronos, chronicles, nimcrypto/sha2, metrics
import rpc/[messages, message, protobuf],
../../peerid,
../../peerinfo,
../../stream/connection,
2019-09-12 05:46:08 +00:00
../../crypto/crypto,
../../protobuf/minprotobuf,
../../utility
2019-09-10 02:15:52 +00:00
export peerid, connection
2019-09-10 02:15:52 +00:00
logScope:
topics = "pubsubpeer"
2019-09-10 02:15:52 +00:00
when defined(libp2p_expensive_metrics):
declareCounter(libp2p_pubsub_sent_messages, "number of messages sent", labels = ["id", "topic"])
declareCounter(libp2p_pubsub_received_messages, "number of messages received", labels = ["id", "topic"])
declareCounter(libp2p_pubsub_skipped_received_messages, "number of received skipped messages", labels = ["id"])
declareCounter(libp2p_pubsub_skipped_sent_messages, "number of sent skipped messages", labels = ["id"])
2019-09-10 02:15:52 +00:00
type
PubSubObserver* = ref object
onRecv*: proc(peer: PubSubPeer; msgs: var RPCMsg) {.gcsafe, raises: [Defect].}
onSend*: proc(peer: PubSubPeer; msgs: var RPCMsg) {.gcsafe, raises: [Defect].}
2019-09-10 02:15:52 +00:00
PubSubPeerEventKind* {.pure.} = enum
Connected
Disconnected
PubsubPeerEvent* = object
kind*: PubSubPeerEventKind
GetConn* = proc(): Future[Connection] {.gcsafe.}
OnEvent* = proc(peer: PubSubPeer, event: PubsubPeerEvent) {.gcsafe.}
PubSubPeer* = ref object of RootObj
getConn*: GetConn # callback to establish a new send connection
onEvent*: OnEvent # Connectivity updates for peer
codec*: string # the protocol that this peer joined from
sendConn*: Connection # cached send connection
connections*: seq[Connection] # connections to this peer
peerId*: PeerID
handler*: RPCHandler
observers*: ref seq[PubSubObserver] # ref as in smart_ptr
Gossip one one (#240) * allow multiple codecs per protocol (without breaking things) * add 1.1 protocol to gossip * explicit peering part 1 * explicit peering part 2 * explicit peering part 3 * PeerInfo and ControlPrune protocols * fix encodePrune * validated always, even explicit peers * prune by score (score is stub still) * add a way to pass parameters to gossip * standard setup fixes * take into account explicit direct peers in publish * add floodPublish logic * small fixes, publish still half broken * make sure to waitsub in sparse test * use var semantics to optimize table access * wip... lvalues don't work properly sadly... * big publish refactor, replenish and balance * fix internal tests * use g.peers for fanout (todo: don't include flood peers) * exclude non gossip from fanout * internal test fixes * fix flood tests * fix test's trypublish * test interop fixes * make sure to not remove peers from gossip table * restore old replenishFanout * cleanups * restore utility module import * restore trace vs debug in gossip * improve fanout replenish behavior further * triage publish nil peers (issue is on master too but just hidden behind a if/in) * getGossipPeers fixes * remove topics from pubsubpeer (was unused) * simplify rebalanceMesh (following spec) and make it finally reach D_high * better diagnostics * merge new pubsubpeer, copy 1.1 to new module * fix up merge * conditional enable gossip11 module * add back topics in peers, re-enable flood publish * add more heartbeat locking to prevent races * actually lock the heartbeat * minor fixes * with sugar * merge 1.0 * remove assertion in publish * fix multistream 1.1 multi proto * Fix merge oops * wip * fix gossip 11 upstream * gossipsub11 -> gossipsub * support interop testing * tests fixing * fix directchat build * control prune updates (pb) * wip parameters * gossip internal tests fixes * parameters wip * finishup with params * cleanups/wip * small sugar * grafted and pruned procs * wip updateScores * wip * fix logging issue * pubsubpeer, chronicles explicit override * fix internal gossip tests * wip * tables troubleshooting * score wip * score wip * fixes * fix test utils generateNodes * don't delete while iterating in score update * fix grafted defect * add a handleConnect in subscribeTopic * pruning improvements * wip * score fixes * post merge - builds gossip tests * further merge fixes * rebalance improvements and opportunistic grafting * fix test for now * restore explicit peering * implement peer exchange graft message * add an hard cap to PX * backoff time management * IWANT cap/budget * Adaptive gossip dissemination * outbound mesh quota, internal tests fixing * oversub prune score based, finish outbound quota * finishup with score and ihave budget * use go daemon 0.3.0 * import fixes * byScore cleanup score sorting * remove pointless scaling in `/` Duration operator * revert using libp2p org for daemon * interop fixes * fixes and cleanup * remove heartbeat assertion, minor debug fixes * logging improvements and cleaning up * (to revert) add some traces * add explicit topic to gossip rpcs * pubsub merge fixes and type fix in switch * Revert "(to revert) add some traces" This reverts commit 4663eaab6cc336c81cee50bc54025cf0b7bcbd99. * cleanup some now irrelevant todo * shuffle peers anyway as score might be disabled * add missing shuffle * old merge fix * more merge fixes * debug improvements * re-enable gossip internal tests * add gossip10 fallback (dormant but tested) * split gossipsub internal tests into 1.0 and 1.1 Co-authored-by: Dmitriy Ryajov <dryajov@gmail.com>
2020-09-21 09:16:29 +00:00
score*: float64
iWantBudget*: int
iHaveBudget*: int
outbound*: bool # if this is an outbound connection
appScore*: float64 # application specific score
behaviourPenalty*: float64 # the eventual penalty score
RPCHandler* = proc(peer: PubSubPeer, msg: RPCMsg): Future[void] {.gcsafe.}
2019-09-10 02:15:52 +00:00
func hash*(p: PubSubPeer): Hash =
# int is either 32/64, so intptr basically, pubsubpeer is a ref
cast[pointer](p).hash
func shortLog*(p: PubSubPeer): string =
if p.isNil: "PubSubPeer(nil)"
else: shortLog(p.peerId)
chronicles.formatIt(PubSubPeer): shortLog(it)
PubSub (Gossip & Flood) Implementation (#36) This adds gossipsub and floodsub, as well as basic interop testing with the go libp2p daemon. * add close event * wip: gossipsub * splitting rpc message * making message handling more consistent * initial gossipsub implementation * feat: nim 1.0 cleanup * wip: gossipsub protobuf * adding encoding/decoding of gossipsub messages * add disconnect handler * add proper gossipsub msg handling * misc: cleanup for nim 1.0 * splitting floodsub and gossipsub tests * feat: add mesh rebalansing * test pubsub * add mesh rebalansing tests * testing mesh maintenance * finishing mcache implementatin * wip: commenting out broken tests * wip: don't run heartbeat for now * switchout debug for trace logging * testing gossip peer selection algorithm * test stream piping * more work around message amplification * get the peerid from message * use timed cache as backing store * allow setting timeout in constructor * several changes to improve performance * more through testing of msg amplification * prevent gc issues * allow piping to self and prevent deadlocks * improove floodsub * allow running hook on cache eviction * prevent race conditions * prevent race conditions and improove tests * use hashes as cache keys * removing useless file * don't create a new seq * re-enable pubsub tests * fix imports * reduce number of runs to speed up tests * break out control message processing * normalize sleeps between steps * implement proper transport filtering * initial interop testing * clean up floodsub publish logic * allow dialing without a protocol * adding multiple reads/writes * use protobuf varint in mplex * don't loose conn's peerInfo * initial interop pubsub tests * don't duplicate connections/peers * bring back interop tests * wip: interop * re-enable interop and daemon tests * add multiple read write tests from handlers * don't cleanup channel prematurely * use correct channel to send/receive msgs * adjust tests with latest changes * include interop tests * remove temp logging output * fix ci * use correct public key serialization * additional tests for pubsub interop
2019-12-06 02:16:18 +00:00
proc connected*(p: PubSubPeer): bool =
not p.sendConn.isNil and not
(p.sendConn.closed or p.sendConn.atEof)
proc hasObservers(p: PubSubPeer): bool =
p.observers != nil and anyIt(p.observers[], it != nil)
proc recvObservers(p: PubSubPeer, msg: var RPCMsg) =
# trigger hooks
if not(isNil(p.observers)) and p.observers[].len > 0:
for obs in p.observers[]:
if not(isNil(obs)): # TODO: should never be nil, but...
obs.onRecv(p, msg)
proc sendObservers(p: PubSubPeer, msg: var RPCMsg) =
# trigger hooks
if not(isNil(p.observers)) and p.observers[].len > 0:
for obs in p.observers[]:
if not(isNil(obs)): # TODO: should never be nil, but...
obs.onSend(p, msg)
proc handle*(p: PubSubPeer, conn: Connection) {.async.} =
debug "starting pubsub read loop",
conn, peer = p, closed = conn.closed
2019-09-10 02:15:52 +00:00
try:
2020-05-25 14:33:24 +00:00
try:
while not conn.atEof:
trace "waiting for data", conn, peer = p, closed = conn.closed
let data = await conn.readLp(64 * 1024)
trace "read data from peer",
conn, peer = p, closed = conn.closed,
data = data.shortLog
2020-05-25 14:33:24 +00:00
var rmsg = decodeRpcMsg(data)
if rmsg.isErr():
notice "failed to decode msg from peer",
conn, peer = p, closed = conn.closed,
err = rmsg.error()
break
trace "decoded msg from peer",
conn, peer = p, closed = conn.closed,
msg = rmsg.get().shortLog
2020-05-25 14:33:24 +00:00
# trigger hooks
p.recvObservers(rmsg.get())
when defined(libp2p_expensive_metrics):
for m in rmsg.get().messages:
for t in m.topicIDs:
# metrics
libp2p_pubsub_received_messages.inc(labelValues = [$p.peerId, t])
2020-06-11 18:09:34 +00:00
await p.handler(p, rmsg.get())
2020-05-25 14:33:24 +00:00
finally:
await conn.close()
except CancelledError:
# This is top-level procedure which will work as separate task, so it
# do not need to propogate CancelledError.
trace "Unexpected cancellation in PubSubPeer.handle"
PubSub (Gossip & Flood) Implementation (#36) This adds gossipsub and floodsub, as well as basic interop testing with the go libp2p daemon. * add close event * wip: gossipsub * splitting rpc message * making message handling more consistent * initial gossipsub implementation * feat: nim 1.0 cleanup * wip: gossipsub protobuf * adding encoding/decoding of gossipsub messages * add disconnect handler * add proper gossipsub msg handling * misc: cleanup for nim 1.0 * splitting floodsub and gossipsub tests * feat: add mesh rebalansing * test pubsub * add mesh rebalansing tests * testing mesh maintenance * finishing mcache implementatin * wip: commenting out broken tests * wip: don't run heartbeat for now * switchout debug for trace logging * testing gossip peer selection algorithm * test stream piping * more work around message amplification * get the peerid from message * use timed cache as backing store * allow setting timeout in constructor * several changes to improve performance * more through testing of msg amplification * prevent gc issues * allow piping to self and prevent deadlocks * improove floodsub * allow running hook on cache eviction * prevent race conditions * prevent race conditions and improove tests * use hashes as cache keys * removing useless file * don't create a new seq * re-enable pubsub tests * fix imports * reduce number of runs to speed up tests * break out control message processing * normalize sleeps between steps * implement proper transport filtering * initial interop testing * clean up floodsub publish logic * allow dialing without a protocol * adding multiple reads/writes * use protobuf varint in mplex * don't loose conn's peerInfo * initial interop pubsub tests * don't duplicate connections/peers * bring back interop tests * wip: interop * re-enable interop and daemon tests * add multiple read write tests from handlers * don't cleanup channel prematurely * use correct channel to send/receive msgs * adjust tests with latest changes * include interop tests * remove temp logging output * fix ci * use correct public key serialization * additional tests for pubsub interop
2019-12-06 02:16:18 +00:00
except CatchableError as exc:
trace "Exception occurred in PubSubPeer.handle",
conn, peer = p, closed = conn.closed, exc = exc.msg
finally:
debug "exiting pubsub read loop",
conn, peer = p, closed = conn.closed
2019-09-10 02:15:52 +00:00
proc connectOnce(p: PubSubPeer): Future[void] {.async.} =
try:
let newConn = await p.getConn()
if newConn.isNil:
raise (ref CatchableError)(msg: "Cannot establish send connection")
# When the send channel goes up, subscriptions need to be sent to the
# remote peer - if we had multiple channels up and one goes down, all
# stop working so we make an effort to only keep a single channel alive
trace "Get new send connection", p, newConn
p.sendConn = newConn
if p.onEvent != nil:
p.onEvent(p, PubsubPeerEvent(kind: PubSubPeerEventKind.Connected))
await handle(p, newConn)
finally:
if p.sendConn != nil:
trace "Removing send connection", p, conn = p.sendConn
await p.sendConn.close()
p.sendConn = nil
if p.onEvent != nil:
p.onEvent(p, PubsubPeerEvent(kind: PubSubPeerEventKind.Disconnected))
proc connectImpl(p: PubSubPeer) {.async.} =
try:
# Keep trying to establish a connection while it's possible to do so - the
# send connection might get disconnected due to a timeout or an unrelated
# issue so we try to get a new on
while true:
await connectOnce(p)
except CatchableError as exc:
debug "Could not establish send connection", msg = exc.msg
proc connect*(p: PubSubPeer) =
asyncSpawn connectImpl(p)
proc sendImpl(conn: Connection, encoded: seq[byte]) {.async.} =
try:
trace "sending encoded msgs to peer", conn, encoded = shortLog(encoded)
await conn.writeLp(encoded)
trace "sent pubsub message to remote", conn
except CatchableError as exc:
# Because we detach the send call from the currently executing task using
# asyncSpawn, no exceptions may leak out of it
trace "Unable to send to remote", conn, msg = exc.msg
# Next time sendConn is used, it will be have its close flag set and thus
# will be recycled
await conn.close() # This will clean up the send connection
proc send*(p: PubSubPeer, msg: RPCMsg) =
doAssert(not isNil(p), "pubsubpeer nil!")
let conn = p.sendConn
if conn == nil or conn.closed():
trace "No send connection, skipping message", p, msg
return
trace "sending msg to peer", peer = p, rpcMsg = shortLog(msg)
let encoded = if p.hasObservers():
# trigger send hooks
var mm = msg # hooks can modify the message
p.sendObservers(mm)
encodeRpcMsg(mm)
else:
# If there are no send hooks, we redundantly re-encode the message to
# protobuf for every peer - this could easily be improved!
encodeRpcMsg(msg)
if encoded.len <= 0:
debug "empty message, skipping", p, msg
return
# To limit the size of the closure, we only pass the encoded message and
# connection to the spawned send task
asyncSpawn sendImpl(conn, encoded)
PubSub (Gossip & Flood) Implementation (#36) This adds gossipsub and floodsub, as well as basic interop testing with the go libp2p daemon. * add close event * wip: gossipsub * splitting rpc message * making message handling more consistent * initial gossipsub implementation * feat: nim 1.0 cleanup * wip: gossipsub protobuf * adding encoding/decoding of gossipsub messages * add disconnect handler * add proper gossipsub msg handling * misc: cleanup for nim 1.0 * splitting floodsub and gossipsub tests * feat: add mesh rebalansing * test pubsub * add mesh rebalansing tests * testing mesh maintenance * finishing mcache implementatin * wip: commenting out broken tests * wip: don't run heartbeat for now * switchout debug for trace logging * testing gossip peer selection algorithm * test stream piping * more work around message amplification * get the peerid from message * use timed cache as backing store * allow setting timeout in constructor * several changes to improve performance * more through testing of msg amplification * prevent gc issues * allow piping to self and prevent deadlocks * improove floodsub * allow running hook on cache eviction * prevent race conditions * prevent race conditions and improove tests * use hashes as cache keys * removing useless file * don't create a new seq * re-enable pubsub tests * fix imports * reduce number of runs to speed up tests * break out control message processing * normalize sleeps between steps * implement proper transport filtering * initial interop testing * clean up floodsub publish logic * allow dialing without a protocol * adding multiple reads/writes * use protobuf varint in mplex * don't loose conn's peerInfo * initial interop pubsub tests * don't duplicate connections/peers * bring back interop tests * wip: interop * re-enable interop and daemon tests * add multiple read write tests from handlers * don't cleanup channel prematurely * use correct channel to send/receive msgs * adjust tests with latest changes * include interop tests * remove temp logging output * fix ci * use correct public key serialization * additional tests for pubsub interop
2019-12-06 02:16:18 +00:00
when defined(libp2p_expensive_metrics):
for x in mm.messages:
for t in x.topicIDs:
# metrics
libp2p_pubsub_sent_messages.inc(labelValues = [$p.peerId, t])
proc newPubSubPeer*(peerId: PeerID,
getConn: GetConn,
onEvent: OnEvent,
codec: string): PubSubPeer =
PubSubPeer(
getConn: getConn,
onEvent: onEvent,
codec: codec,
peerId: peerId,
)