-
Notifications
You must be signed in to change notification settings - Fork 86
/
KnownPeers.hs
479 lines (437 loc) · 22.1 KB
/
KnownPeers.hs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
{-# LANGUAGE BangPatterns #-}
{-# LANGUAGE LambdaCase #-}
{-# LANGUAGE NamedFieldPuns #-}
{-# LANGUAGE RecordWildCards #-}
{-# LANGUAGE ScopedTypeVariables #-}
module Ouroboros.Network.PeerSelection.Governor.KnownPeers
( belowTarget
, aboveTarget
) where
import Data.Hashable
import Data.List (sortBy)
import Data.Maybe (fromMaybe)
import Data.Set qualified as Set
import GHC.Stack (HasCallStack)
import System.Random (random)
import Control.Concurrent.JobPool (Job (..))
import Control.Exception (Exception (..), SomeException, assert)
import Control.Monad.Class.MonadAsync
import Control.Monad.Class.MonadSTM
import Control.Monad.Class.MonadTime.SI
import Control.Monad.Class.MonadTimer.SI
import Ouroboros.Network.PeerSelection.Bootstrap (requiresBootstrapPeers)
import Ouroboros.Network.PeerSelection.Governor.Types
import Ouroboros.Network.PeerSelection.PeerAdvertise (PeerAdvertise (..))
import Ouroboros.Network.PeerSelection.PeerSharing (PeerSharing (..))
import Ouroboros.Network.PeerSelection.PublicRootPeers qualified as PublicRootPeers
import Ouroboros.Network.PeerSelection.State.EstablishedPeers qualified as EstablishedPeers
import Ouroboros.Network.PeerSelection.State.KnownPeers qualified as KnownPeers
import Ouroboros.Network.PeerSelection.State.LocalRootPeers qualified as LocalRootPeers
import Ouroboros.Network.Protocol.PeerSharing.Type (PeerSharingAmount)
---------------------------
-- Known peers below target
--
-- | If we are below the target of /known peers/ we peer share (if we are above
-- the peer share request threshold).
--
-- It should be noted if the node is in bootstrap mode (i.e. in a sensitive
-- state) then this monitoring action will be disabled.
--
belowTarget :: (MonadAsync m, MonadTimer m, Ord peeraddr, Hashable peeraddr,
HasCallStack)
=> PeerSelectionActions peeraddr peerconn m
-> MkGuardedDecision peeraddr peerconn m
belowTarget actions
policy@PeerSelectionPolicy {
policyMaxInProgressPeerShareReqs,
policyPickKnownPeersForPeerShare,
policyPeerShareRetryTime
}
st@PeerSelectionState {
knownPeers,
publicRootPeers,
establishedPeers,
inProgressPeerShareReqs,
inProgressDemoteToCold,
targets = PeerSelectionTargets {
targetNumberOfKnownPeers
},
ledgerStateJudgement,
bootstrapPeersFlag,
fuzzRng
}
-- Only start Peer Sharing request if PeerSharing was enabled
| PeerSharingEnabled <- peerSharing actions
-- Are we under target for number of known peers?
, numKnownPeers < targetNumberOfKnownPeers
-- Are we at our limit for number of peer share requests?
, numPeerShareReqsPossible > 0
-- Are there any known peers that we can send a peer share request to?
-- We can only ask ones where we have not asked them within a certain time.
, not (Set.null availableForPeerShare)
-- No peer share requests should be issued when the node is in a sensitive
-- state
, not (requiresBootstrapPeers bootstrapPeersFlag ledgerStateJudgement)
= Guarded Nothing $ do
-- Max selected should be <= numPeerShareReqsPossible
selectedForPeerShare <- pickPeers st
policyPickKnownPeersForPeerShare
availableForPeerShare
numPeerShareReqsPossible
let -- Should be <= numPeerShareReqsPossible
numPeerShareReqs = Set.size selectedForPeerShare
objective = targetNumberOfKnownPeers - numKnownPeers
-- Split current peer target objective across all peer sharing
-- candidates. If the objective is smaller than the number of
-- peer share requests available, ask for at 1 peer to each.
--
-- This is to increase diversity.
numPeersToReq :: PeerSharingAmount
!numPeersToReq = fromIntegral
$ min 255 (max 8 (objective `div` numPeerShareReqs))
(salt, fuzzRng') = random fuzzRng
return $ \now -> Decision {
decisionTrace = [TracePeerShareRequests
targetNumberOfKnownPeers
numKnownPeers
numPeersToReq
availableForPeerShare
selectedForPeerShare],
decisionState = st {
inProgressPeerShareReqs = inProgressPeerShareReqs
+ numPeerShareReqs,
establishedPeers = EstablishedPeers.setPeerShareTime
selectedForPeerShare
(addTime policyPeerShareRetryTime now)
establishedPeers,
fuzzRng = fuzzRng'
},
decisionJobs =
[jobPeerShare actions policy objective salt numPeersToReq
(Set.toList selectedForPeerShare)]
}
-- If we could peer share except that there are none currently available
-- then we return the next wakeup time (if any)
| numKnownPeers < targetNumberOfKnownPeers
, numPeerShareReqsPossible > 0
, Set.null availableForPeerShare
= GuardedSkip $ EstablishedPeers.minPeerShareTime establishedPeers
| otherwise
= GuardedSkip Nothing
where
numKnownPeers = Set.size $ KnownPeers.toSet knownPeers
Set.\\ PublicRootPeers.getBigLedgerPeers publicRootPeers
numPeerShareReqsPossible = policyMaxInProgressPeerShareReqs
- inProgressPeerShareReqs
-- Only peer which permit peersharing are available
availableForPeerShare = EstablishedPeers.availableForPeerShare establishedPeers
Set.\\ inProgressDemoteToCold
---------------------------
-- Peer sharing job
--
-- | The peer sharing job is run in two stages. The expected path is for all
-- peer sharing request to return within a short timeout. The second phase is
-- with a longer timeout for all still outstanding requests.
--
-- The result from each phase is filtered. Already known peers and big ledger
-- peers are removed before adding them to known peers. Big ledger peers are
-- popular so they don't need to be shared through peer sharing. However ledger
-- peers belonging to smaller pools shouldn't be discarded. Smaller pools could
-- use extra upstream peers and we spread out the load in the network.
--
-- If we ask for more peers than needed a random subset of the peers in the filtered result
-- is used.
jobPeerShare :: forall m peeraddr peerconn.
(MonadAsync m, MonadTimer m, Ord peeraddr, Hashable peeraddr)
=> PeerSelectionActions peeraddr peerconn m
-> PeerSelectionPolicy peeraddr m
-> Int
-> Int
-> PeerSharingAmount
-> [peeraddr]
-> Job () m (Completion m peeraddr peerconn)
jobPeerShare PeerSelectionActions{requestPeerShare}
PeerSelectionPolicy{..} salt maxAmount =
\amount peers -> Job (jobPhase1 amount peers) (handler peers) () "peerSharePhase1"
where
-- Return n random peers from a list of peers.
--
-- Every jobPeerShare will be called with a new random salt.
-- This means that even if presented with the same list peers their ordering
-- will be unpredictable.
takeNPeers :: Int -> [peeraddr] -> [peeraddr]
takeNPeers n addrs = take n $
sortBy (\a b -> compare (hashWithSalt salt a) (hashWithSalt salt b))
addrs
handler :: [peeraddr] -> SomeException -> m (Completion m peeraddr peerconn)
handler peers e = return $
Completion $ \st _ ->
Decision { decisionTrace = [TracePeerShareResults [ (p, Left e) | p <- peers ]],
decisionState =
st { inProgressPeerShareReqs = inProgressPeerShareReqs st
- length peers
},
decisionJobs = []
}
jobPhase1 :: PeerSharingAmount -> [peeraddr] -> m (Completion m peeraddr peerconn)
jobPhase1 amount peers = do
-- In the typical case, where most requests return within a short
-- timeout we want to collect all the responses into a batch and
-- add them to the known peers set in one go.
--
-- So fire them all off in one go:
peerShares <- sequence [ async (requestPeerShare amount peer)
| peer <- peers ]
-- First to finish synchronisation between /all/ the peer share requests
-- completing or the timeout (with whatever partial results we have at
-- the time)
results <- waitAllCatchOrTimeout peerShares policyPeerShareBatchWaitTime
case results of
Right totalResults ->
return $ Completion $ \st _ ->
let peerResults = zip peers totalResults
newPeers = takeNPeers maxAmount $
[ p | Right (PeerSharingResult ps) <- totalResults
, p <- ps
, not (KnownPeers.member p (knownPeers st))
, Set.notMember p (PublicRootPeers.getBigLedgerPeers (publicRootPeers st))]
in Decision { decisionTrace = [ TracePeerShareResults peerResults
, TracePeerShareResultsFiltered newPeers
]
, decisionState =
st { -- TODO: also update with the failures
knownPeers = KnownPeers.alter
(\x -> case x of
Nothing ->
KnownPeers.alterKnownPeerInfo
(Nothing, Just DoAdvertisePeer)
x
Just _ ->
KnownPeers.alterKnownPeerInfo
(Nothing, Nothing)
x
)
(Set.fromList newPeers)
(knownPeers st),
inProgressPeerShareReqs = inProgressPeerShareReqs st
- length peers
}
, decisionJobs = []
}
-- But if any don't make the first timeout then they'll be added later
-- when they do reply or never if we hit the hard timeout.
Left partialResults -> do
-- We have to keep track of the relationship between the peer
-- addresses and the peer share requests, completed and still in progress:
let peerResults = [ (p, r)
| (p, Just r) <- zip peers partialResults ]
peersRemaining = [ p
| (p, Nothing) <- zip peers partialResults ]
peerSharesRemaining = [ a
| (a, Nothing) <- zip peerShares partialResults ]
return $ Completion $ \st _ ->
let newPeers = takeNPeers maxAmount $
[ p | Just (Right (PeerSharingResult ps)) <- partialResults
, p <- ps
, not (KnownPeers.member p (knownPeers st))
, Set.notMember p (PublicRootPeers.getBigLedgerPeers (publicRootPeers st))]
in Decision { decisionTrace = [ TracePeerShareResults peerResults
, TracePeerShareResultsFiltered newPeers
]
, decisionState =
st { -- TODO: also update with the failures
knownPeers = KnownPeers.alter
(\x -> case x of
Nothing ->
KnownPeers.alterKnownPeerInfo
(Nothing, Just DoAdvertisePeer)
x
Just _ ->
KnownPeers.alterKnownPeerInfo
(Nothing, Nothing)
x
)
(Set.fromList newPeers)
(knownPeers st),
inProgressPeerShareReqs = inProgressPeerShareReqs st
- length peerResults
}
, decisionJobs = [Job (jobPhase2 (maxAmount - length newPeers) peersRemaining
peerSharesRemaining)
(handler peersRemaining)
()
"peerSharePhase2"]
}
jobPhase2 :: Int -> [peeraddr] -> [Async m (PeerSharingResult peeraddr)]
-> m (Completion m peeraddr peerconn)
jobPhase2 maxRemaining peers peerShares = do
-- Wait again, for all remaining to finish or a timeout.
results <- waitAllCatchOrTimeout
peerShares
(policyPeerShareOverallTimeout
- policyPeerShareBatchWaitTime)
let peerResults =
case results of
Right totalResults -> zip peers totalResults
Left partialResults -> [ (p, fromMaybe err r)
| (p, r) <- zip peers partialResults ]
where err = Left (toException AsyncCancelled)
peerSharesIncomplete =
case results of
Right _totalResults -> []
Left partialResults ->
[ a | (a, Nothing) <- zip peerShares partialResults ]
mapM_ cancel peerSharesIncomplete
return $ Completion $ \st _ ->
let newPeers = takeNPeers maxRemaining $
case results of
Right totalResults -> [ p | Right (PeerSharingResult ps) <- totalResults
, p <- ps
, not (KnownPeers.member p (knownPeers st))
, Set.notMember p (PublicRootPeers.getBigLedgerPeers (publicRootPeers st))]
Left partialResults -> [ p | Just (Right (PeerSharingResult ps)) <- partialResults
, p <- ps
, not (KnownPeers.member p (knownPeers st))
, Set.notMember p (PublicRootPeers.getBigLedgerPeers (publicRootPeers st))]
in Decision { decisionTrace = [ TracePeerShareResults peerResults
, TracePeerShareResultsFiltered newPeers
]
, decisionState =
st { -- TODO: also update with the failures
knownPeers = KnownPeers.alter
(\x -> case x of
Nothing ->
KnownPeers.alterKnownPeerInfo
(Nothing, Just DoAdvertisePeer)
x
Just _ ->
KnownPeers.alterKnownPeerInfo
(Nothing, Nothing)
x
)
(Set.fromList newPeers)
(knownPeers st),
inProgressPeerShareReqs = inProgressPeerShareReqs st
- length peers
}
, decisionJobs = []
}
---------------------------
-- Known peers above target
--
-- | If we are above the target of /known peers/ (i.e. /cold/, /warm/ and /hot/
-- combined), we drop some of the /cold peers/ but we protect the
-- 'targetNumberOfRootPeers' (from combined sets of /local/ and /public root/
-- peers). 'policyPickColdPeersToForget' policy is used to pick the peers.
--
aboveTarget :: (MonadSTM m, Ord peeraddr, HasCallStack)
=> MkGuardedDecision peeraddr peerconn m
aboveTarget PeerSelectionPolicy {
policyPickColdPeersToForget
}
st@PeerSelectionState {
localRootPeers,
publicRootPeers,
knownPeers,
establishedPeers,
inProgressPromoteCold,
targets = PeerSelectionTargets {
targetNumberOfKnownPeers,
targetNumberOfRootPeers
}
}
-- Are we above the target for number of known peers?
| numKnownPeers > targetNumberOfKnownPeers
-- Are there any cold peers we could pick to forget?
-- As a first cheap approximation, check if there are any cold peers.
, numKnownPeers > numEstablishedPeers
-- Beyond this it gets more complicated, and it is not clear that there
-- are any precise cheap checks. So we just do the full calculation.
-- In particular there can be overlap between cold peers and root peers
-- and we have constraints on forgetting root peers.
--
-- We must never pick local root peers to forget as this would violate
-- our invariant that the localRootPeers is a subset of the knownPeers.
--
-- We also need to avoid picking public root peers if that would put us
-- below the target for root peers.
--
, let numRootPeersCanForget = LocalRootPeers.size localRootPeers
+ PublicRootPeers.size publicRootPeers
- targetNumberOfRootPeers
availableToForget = KnownPeers.toSet knownPeers
Set.\\ EstablishedPeers.toSet establishedPeers
Set.\\ LocalRootPeers.keysSet localRootPeers
Set.\\ (if numRootPeersCanForget <= 0
then PublicRootPeers.toSet publicRootPeers
else Set.empty)
Set.\\ inProgressPromoteCold
Set.\\ bigLedgerPeersSet
, not (Set.null availableToForget)
= Guarded Nothing $ do
let numOtherPeersToForget = numKnownPeers
- targetNumberOfKnownPeers
numPeersToForget
| numRootPeersCanForget > 0 = min numRootPeersCanForget
numOtherPeersToForget
| otherwise = numOtherPeersToForget
-- If we /might/ pick a root peer, limit the number to forget so we do
-- not pick too many root peers. This may cause us to go round several
-- times but that is ok.
selectedToForget <- pickPeers st
policyPickColdPeersToForget
availableToForget
numPeersToForget
return $ \_now ->
let knownPeers' = KnownPeers.delete
selectedToForget
knownPeers
publicRootPeers' = publicRootPeers
`PublicRootPeers.difference` selectedToForget
in assert (Set.isSubsetOf
(PublicRootPeers.toSet publicRootPeers')
(KnownPeers.toSet knownPeers'))
Decision {
decisionTrace = [TraceForgetColdPeers
targetNumberOfKnownPeers
numKnownPeers
selectedToForget],
decisionState = st { knownPeers = knownPeers',
publicRootPeers = publicRootPeers' },
decisionJobs = []
}
| otherwise
= GuardedSkip Nothing
where
bigLedgerPeersSet = PublicRootPeers.getBigLedgerPeers publicRootPeers
numKnownPeers, numEstablishedPeers :: Int
numKnownPeers = Set.size $ KnownPeers.toSet knownPeers
Set.\\ bigLedgerPeersSet
numEstablishedPeers = Set.size $ EstablishedPeers.toSet establishedPeers
Set.\\ bigLedgerPeersSet
-------------------------------
-- Utils
--
-- | Perform a first-to-finish synchronisation between:
--
-- * /all/ the async actions completing; or
-- * the timeout with whatever partial results we have at the time
--
-- The result list is the same length and order as the asyncs, so the results
-- can be paired up.
--
waitAllCatchOrTimeout :: (MonadAsync m, MonadTimer m)
=> [Async m a]
-> DiffTime
-> m (Either [Maybe (Either SomeException a)]
[Either SomeException a])
waitAllCatchOrTimeout as time = do
(readTimeout, cancelTimeout) <- registerDelayCancellable time
results <- atomically $
(Right <$> mapM waitCatchSTM as)
`orElse` (Left <$> (readTimeout >>= \case TimeoutPending -> retry
_ -> mapM pollSTM as))
case results of
Right{} -> cancelTimeout
_ -> return ()
return results