Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
# nim-eth
|
|
|
|
# Copyright (c) 2018-2021 Status Research & Development GmbH
|
|
|
|
# Licensed and distributed under either of
|
2022-05-13 16:30:10 +00:00
|
|
|
# * MIT license (license terms in the root directory or at
|
|
|
|
# https://opensource.org/licenses/MIT).
|
|
|
|
# * Apache v2 license (license terms in the root directory or at
|
|
|
|
# https://www.apache.org/licenses/LICENSE-2.0).
|
|
|
|
# at your option. This file may not be copied, modified, or distributed
|
|
|
|
# except according to those terms.
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
|
|
|
import
|
Sync fix: `GetBlockBodies` logic preventing sync, dropping peers
Fixes #864 "Sync progress stops at Goerli block 4494913", and equivalent on
other networks.
The block body fetcher in `blockchain_sync.nim` had an incorrect assumption
about how peers respond to `GetBlockBodies`. It was issuing requests for N
block bodies and incorrectly handling replies which contained fewer than N
bodies.
Having received up to 192 headers in a batch, it split the range into smaller
`GetBlockBodies` requests, fetched each reply, then combined replies. The
effect was Nimbus requested batches of 128+64 block bodies, received gaps in
the reply sequence, then aborted.
That meant it repeatedly fetched data, then discarded it, and fetched it again,
dropping good peers in the process.
Aborted and restarted batches occurred with earlier blocks too, but this became
more pronounced until there were no suitable peers at batch 4494913..4495104.
Here's a trace:
```
TRC 2021-09-29 02:40:24.977+01:00 Requesting block headers file=blockchain_sync.nim:224 start=4494913 count=192 peer=<ENODE>
TRC 2021-09-29 02:40:24.977+01:00 >> Sending eth.GetBlockHeaders (0x03) file=protocol_eth65.nim:51 peer=<PEER> startBlock=4494913 max=192
TRC 2021-09-29 02:40:25.005+01:00 << Got reply eth.BlockHeaders (0x04) file=protocol_eth65.nim:51 peer=<PEER> count=192
TRC 2021-09-29 02:40:25.007+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=128
TRC 2021-09-29 02:40:25.209+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=13
TRC 2021-09-29 02:40:25.210+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=64
TRC 2021-09-29 02:40:25.290+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=64
WRN 2021-09-29 02:40:25.306+01:00 Bodies len != headers.len file=blockchain_sync.nim:276 bodies=77 headers=192
TRC 2021-09-29 02:40:25.306+01:00 peer disconnected file=blockchain_sync.nim:403 peer=<PEER>
TRC 2021-09-29 02:40:25.306+01:00 Finished obtaining blocks file=blockchain_sync.nim:303 peer=<PEER>
```
In practice, for modern peers, Nimbus received shorter replies than it assumed
depending on the block sizes on the chain. Geth/Erigon has 2MiB `BlockBodies`
response size soft limit. OpenEthereum has 4MiB.
Up to Berlin (EIP-2929), Nimbus's fetcher failed often, but there were still
some peers serving what Nimbus needed.
Just after the start of Berlin, at batch 4494913..4495104 on Goerli, zero peers
responded with full size replies for the whole batch, so Nimbus couldn't
progress past that point. But there was already a problem happening before
that for large blocks, dropping good peers and repeatedly fetching the same
block data.
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-10-04 17:13:33 +00:00
|
|
|
std/[sets, options, random, hashes, sequtils],
|
2022-05-13 16:30:10 +00:00
|
|
|
chronicles,
|
|
|
|
chronos,
|
|
|
|
eth/[common/eth_types, p2p],
|
|
|
|
eth/p2p/[private/p2p_types, peer_pool],
|
|
|
|
stew/byteutils,
|
2022-05-23 16:53:19 +00:00
|
|
|
"."/[protocol, types]
|
2022-03-21 17:12:07 +00:00
|
|
|
|
|
|
|
{.push raises:[Defect].}
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
2022-05-23 16:53:19 +00:00
|
|
|
logScope:
|
2022-06-16 08:58:50 +00:00
|
|
|
topics = "fast-sync"
|
2022-05-23 16:53:19 +00:00
|
|
|
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
const
|
|
|
|
minPeersToStartSync* = 2 # Wait for consensus of at least this
|
|
|
|
# number of peers before syncing
|
|
|
|
|
|
|
|
type
|
2022-05-13 16:30:10 +00:00
|
|
|
#SyncStatus = enum
|
|
|
|
# syncSuccess
|
|
|
|
# syncNotEnoughPeers
|
|
|
|
# syncTimeOut
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
2022-03-21 17:12:07 +00:00
|
|
|
BlockchainSyncDefect* = object of Defect
|
|
|
|
## Catch and relay exception
|
|
|
|
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
WantedBlocksState = enum
|
|
|
|
Initial,
|
|
|
|
Requested,
|
|
|
|
Received,
|
|
|
|
Persisted
|
|
|
|
|
|
|
|
WantedBlocks = object
|
|
|
|
startIndex: BlockNumber
|
|
|
|
numBlocks: uint
|
|
|
|
state: WantedBlocksState
|
|
|
|
headers: seq[BlockHeader]
|
|
|
|
bodies: seq[BlockBody]
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
FastSyncCtx* = ref object
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
workQueue: seq[WantedBlocks]
|
|
|
|
endBlockNumber: BlockNumber
|
|
|
|
finalizedBlock: BlockNumber # Block which was downloaded and verified
|
|
|
|
chain: AbstractChainDB
|
|
|
|
peerPool: PeerPool
|
|
|
|
trustedPeers: HashSet[Peer]
|
|
|
|
hasOutOfOrderBlocks: bool
|
|
|
|
|
|
|
|
proc hash*(p: Peer): Hash = hash(cast[pointer](p))
|
|
|
|
|
|
|
|
proc endIndex(b: WantedBlocks): BlockNumber =
|
|
|
|
result = b.startIndex
|
|
|
|
result += (b.numBlocks - 1).toBlockNumber
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc availableWorkItem(ctx: FastSyncCtx): int =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
var maxPendingBlock = ctx.finalizedBlock # the last downloaded & processed
|
|
|
|
trace "queue len", length = ctx.workQueue.len
|
|
|
|
result = -1
|
|
|
|
for i in 0 .. ctx.workQueue.high:
|
|
|
|
case ctx.workQueue[i].state
|
|
|
|
of Initial:
|
|
|
|
# When there is a work item at Initial state, immediatly use this one.
|
|
|
|
# This usually means a previous work item that failed somewhere in the
|
|
|
|
# process, and thus can be reused to work on.
|
|
|
|
return i
|
|
|
|
of Persisted:
|
|
|
|
# In case of Persisted, we can reset this work item to a new one.
|
|
|
|
result = i
|
|
|
|
# No break here to give work items in Initial state priority and to
|
|
|
|
# calculate endBlock.
|
|
|
|
else:
|
|
|
|
discard
|
|
|
|
|
|
|
|
# Check all endBlocks of all workqueue items to decide on next range of
|
|
|
|
# blocks to collect & process.
|
|
|
|
let endBlock = ctx.workQueue[i].endIndex
|
|
|
|
if endBlock > maxPendingBlock:
|
|
|
|
maxPendingBlock = endBlock
|
|
|
|
|
|
|
|
let nextRequestedBlock = maxPendingBlock + 1
|
|
|
|
# If this next block doesn't exist yet according to any of our peers, don't
|
|
|
|
# return a work item (and sync will be stopped).
|
|
|
|
if nextRequestedBlock >= ctx.endBlockNumber:
|
|
|
|
return -1
|
|
|
|
|
|
|
|
# Increase queue when there are no free (Initial / Persisted) work items in
|
|
|
|
# the queue. At start, queue will be empty.
|
|
|
|
if result == -1:
|
|
|
|
result = ctx.workQueue.len
|
|
|
|
ctx.workQueue.setLen(result + 1)
|
|
|
|
|
|
|
|
# Create new work item when queue was increased, reset when selected work item
|
|
|
|
# is at Persisted state.
|
2022-03-21 17:12:07 +00:00
|
|
|
var numBlocks = (ctx.endBlockNumber - nextRequestedBlock).truncate(int)
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
if numBlocks > maxHeadersFetch:
|
|
|
|
numBlocks = maxHeadersFetch
|
|
|
|
ctx.workQueue[result] = WantedBlocks(startIndex: nextRequestedBlock, numBlocks: numBlocks.uint, state: Initial)
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc persistWorkItem(ctx: FastSyncCtx, wi: var WantedBlocks): ValidationResult
|
2022-03-21 17:12:07 +00:00
|
|
|
{.gcsafe, raises:[Defect,CatchableError].} =
|
2022-07-06 15:00:51 +00:00
|
|
|
try:
|
2022-03-21 17:12:07 +00:00
|
|
|
result = ctx.chain.persistBlocks(wi.headers, wi.bodies)
|
2022-07-06 15:00:51 +00:00
|
|
|
except CatchableError as e:
|
|
|
|
error "storing persistent blocks failed",
|
|
|
|
error = $e.name, msg = e.msg
|
|
|
|
result = ValidationResult.Error
|
|
|
|
except Defect as e:
|
|
|
|
# Pass through
|
2022-07-21 12:14:41 +00:00
|
|
|
raise e
|
2022-07-06 15:00:51 +00:00
|
|
|
except Exception as e:
|
2022-07-21 12:14:41 +00:00
|
|
|
# Notorious case where the `Chain` reference applied to `persistBlocks()`
|
|
|
|
# has the compiler traced a possible `Exception` (i.e. `ctx.chain` could
|
|
|
|
# be uninitialised.)
|
2022-07-06 15:00:51 +00:00
|
|
|
error "exception while storing persistent blocks",
|
|
|
|
error = $e.name, msg = e.msg
|
2022-07-21 12:14:41 +00:00
|
|
|
raise (ref Defect)(msg: $e.name & ": " & e.msg)
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
case result
|
|
|
|
of ValidationResult.OK:
|
|
|
|
ctx.finalizedBlock = wi.endIndex
|
|
|
|
wi.state = Persisted
|
|
|
|
of ValidationResult.Error:
|
|
|
|
wi.state = Initial
|
|
|
|
# successful or not, we're done with these blocks
|
|
|
|
wi.headers = @[]
|
|
|
|
wi.bodies = @[]
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc persistPendingWorkItems(ctx: FastSyncCtx): (int, ValidationResult)
|
2022-03-21 17:12:07 +00:00
|
|
|
{.gcsafe, raises:[Defect,CatchableError].} =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
var nextStartIndex = ctx.finalizedBlock + 1
|
|
|
|
var keepRunning = true
|
|
|
|
var hasOutOfOrderBlocks = false
|
|
|
|
trace "Looking for out of order blocks"
|
|
|
|
while keepRunning:
|
|
|
|
keepRunning = false
|
|
|
|
hasOutOfOrderBlocks = false
|
|
|
|
# Go over the full work queue and check for every work item if it is in
|
|
|
|
# Received state and has the next blocks in line to be processed.
|
|
|
|
for i in 0 ..< ctx.workQueue.len:
|
|
|
|
let start = ctx.workQueue[i].startIndex
|
|
|
|
# There should be at least 1 like this, namely the just received work item
|
|
|
|
# that initiated this call.
|
|
|
|
if ctx.workQueue[i].state == Received:
|
|
|
|
if start == nextStartIndex:
|
|
|
|
trace "Processing pending work item", number = start
|
|
|
|
result = (i, ctx.persistWorkItem(ctx.workQueue[i]))
|
|
|
|
# TODO: We can stop here on failure, but have to set
|
|
|
|
# hasOutofORderBlocks. Is this always valid?
|
|
|
|
nextStartIndex = ctx.finalizedBlock + 1
|
|
|
|
keepRunning = true
|
|
|
|
break
|
|
|
|
else:
|
|
|
|
hasOutOfOrderBlocks = true
|
|
|
|
|
|
|
|
ctx.hasOutOfOrderBlocks = hasOutOfOrderBlocks
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc returnWorkItem(ctx: FastSyncCtx, workItem: int): ValidationResult
|
2022-03-21 17:12:07 +00:00
|
|
|
{.gcsafe, raises:[Defect,CatchableError].} =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
let wi = addr ctx.workQueue[workItem]
|
|
|
|
let askedBlocks = wi.numBlocks.int
|
|
|
|
let receivedBlocks = wi.headers.len
|
|
|
|
let start = wi.startIndex
|
|
|
|
|
|
|
|
if askedBlocks == receivedBlocks:
|
|
|
|
trace "Work item complete",
|
|
|
|
start,
|
|
|
|
askedBlocks,
|
|
|
|
receivedBlocks
|
|
|
|
|
|
|
|
if wi.startIndex != ctx.finalizedBlock + 1:
|
|
|
|
trace "Blocks out of order", start, final = ctx.finalizedBlock
|
|
|
|
ctx.hasOutOfOrderBlocks = true
|
|
|
|
|
|
|
|
if ctx.hasOutOfOrderBlocks:
|
|
|
|
let (index, validation) = ctx.persistPendingWorkItems()
|
|
|
|
# Only report an error if it was this peer's work item that failed
|
2022-04-08 04:54:11 +00:00
|
|
|
if validation == ValidationResult.Error and index == workItem:
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
result = ValidationResult.Error
|
|
|
|
# TODO: What about failures on other peers' work items?
|
|
|
|
# In that case the peer will probably get disconnected on future erroneous
|
|
|
|
# work items, but before this occurs, several more blocks (that will fail)
|
|
|
|
# might get downloaded from this peer. This will delay the sync and this
|
|
|
|
# should be improved.
|
|
|
|
else:
|
|
|
|
trace "Processing work item", number = wi.startIndex
|
|
|
|
# Validation result needs to be returned so that higher up can be decided
|
|
|
|
# to disconnect from this peer in case of error.
|
|
|
|
result = ctx.persistWorkItem(wi[])
|
|
|
|
else:
|
|
|
|
trace "Work item complete but we got fewer blocks than requested, so we're ditching the whole thing.",
|
|
|
|
start,
|
|
|
|
askedBlocks,
|
|
|
|
receivedBlocks
|
|
|
|
return ValidationResult.Error
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc handleLostPeer(ctx: FastSyncCtx) =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
# TODO: ask the PeerPool for new connections and then call
|
|
|
|
# `obtainBlocksFromPeer`
|
|
|
|
discard
|
|
|
|
|
|
|
|
proc getBestBlockNumber(p: Peer): Future[BlockNumber] {.async.} =
|
|
|
|
let request = BlocksRequest(
|
|
|
|
startBlock: HashOrNum(isHash: true,
|
|
|
|
hash: p.state(eth).bestBlockHash),
|
|
|
|
maxResults: 1,
|
|
|
|
skip: 0,
|
|
|
|
reverse: true)
|
|
|
|
|
2022-06-06 13:42:08 +00:00
|
|
|
trace trEthSendSendingGetBlockHeaders, peer=p,
|
2022-03-21 17:12:07 +00:00
|
|
|
startBlock=request.startBlock.hash.toHex, max=request.maxResults
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
let latestBlock = await p.getBlockHeaders(request)
|
|
|
|
|
2021-07-22 16:36:17 +00:00
|
|
|
if latestBlock.isSome:
|
|
|
|
if latestBlock.get.headers.len > 0:
|
|
|
|
result = latestBlock.get.headers[0].blockNumber
|
2022-06-06 13:42:08 +00:00
|
|
|
trace trEthRecvReceivedBlockHeaders, peer=p,
|
2022-05-23 16:53:19 +00:00
|
|
|
count=latestBlock.get.headers.len,
|
|
|
|
blockNumber=(if latestBlock.get.headers.len > 0: $result else: "missing")
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc obtainBlocksFromPeer(syncCtx: FastSyncCtx, peer: Peer) {.async.} =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
# Update our best block number
|
|
|
|
try:
|
|
|
|
let bestBlockNumber = await peer.getBestBlockNumber()
|
|
|
|
if bestBlockNumber > syncCtx.endBlockNumber:
|
|
|
|
trace "New sync end block number", number = bestBlockNumber
|
|
|
|
syncCtx.endBlockNumber = bestBlockNumber
|
|
|
|
except TransportError:
|
|
|
|
debug "Transport got closed during obtainBlocksFromPeer"
|
|
|
|
except CatchableError as e:
|
|
|
|
debug "Exception in getBestBlockNumber()", exc = e.name, err = e.msg
|
|
|
|
# no need to exit here, because the context might still have blocks to fetch
|
|
|
|
# from this peer
|
|
|
|
|
|
|
|
while (let workItemIdx = syncCtx.availableWorkItem(); workItemIdx != -1 and
|
|
|
|
peer.connectionState notin {Disconnecting, Disconnected}):
|
|
|
|
template workItem: auto = syncCtx.workQueue[workItemIdx]
|
|
|
|
workItem.state = Requested
|
|
|
|
trace "Requesting block headers", start = workItem.startIndex,
|
|
|
|
count = workItem.numBlocks, peer = peer.remote.node
|
|
|
|
let request = BlocksRequest(
|
|
|
|
startBlock: HashOrNum(isHash: false, number: workItem.startIndex),
|
|
|
|
maxResults: workItem.numBlocks,
|
|
|
|
skip: 0,
|
|
|
|
reverse: false)
|
|
|
|
|
|
|
|
var dataReceived = false
|
|
|
|
try:
|
2022-06-06 13:42:08 +00:00
|
|
|
trace trEthSendSendingGetBlockHeaders, peer,
|
2022-05-09 14:04:48 +00:00
|
|
|
startBlock=request.startBlock.number, max=request.maxResults,
|
|
|
|
step=traceStep(request)
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
let results = await peer.getBlockHeaders(request)
|
|
|
|
if results.isSome:
|
2022-06-06 13:42:08 +00:00
|
|
|
trace trEthRecvReceivedBlockHeaders, peer,
|
2022-05-09 14:04:48 +00:00
|
|
|
count=results.get.headers.len, requested=request.maxResults
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
shallowCopy(workItem.headers, results.get.headers)
|
|
|
|
|
Sync fix: `GetBlockBodies` logic preventing sync, dropping peers
Fixes #864 "Sync progress stops at Goerli block 4494913", and equivalent on
other networks.
The block body fetcher in `blockchain_sync.nim` had an incorrect assumption
about how peers respond to `GetBlockBodies`. It was issuing requests for N
block bodies and incorrectly handling replies which contained fewer than N
bodies.
Having received up to 192 headers in a batch, it split the range into smaller
`GetBlockBodies` requests, fetched each reply, then combined replies. The
effect was Nimbus requested batches of 128+64 block bodies, received gaps in
the reply sequence, then aborted.
That meant it repeatedly fetched data, then discarded it, and fetched it again,
dropping good peers in the process.
Aborted and restarted batches occurred with earlier blocks too, but this became
more pronounced until there were no suitable peers at batch 4494913..4495104.
Here's a trace:
```
TRC 2021-09-29 02:40:24.977+01:00 Requesting block headers file=blockchain_sync.nim:224 start=4494913 count=192 peer=<ENODE>
TRC 2021-09-29 02:40:24.977+01:00 >> Sending eth.GetBlockHeaders (0x03) file=protocol_eth65.nim:51 peer=<PEER> startBlock=4494913 max=192
TRC 2021-09-29 02:40:25.005+01:00 << Got reply eth.BlockHeaders (0x04) file=protocol_eth65.nim:51 peer=<PEER> count=192
TRC 2021-09-29 02:40:25.007+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=128
TRC 2021-09-29 02:40:25.209+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=13
TRC 2021-09-29 02:40:25.210+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=64
TRC 2021-09-29 02:40:25.290+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=64
WRN 2021-09-29 02:40:25.306+01:00 Bodies len != headers.len file=blockchain_sync.nim:276 bodies=77 headers=192
TRC 2021-09-29 02:40:25.306+01:00 peer disconnected file=blockchain_sync.nim:403 peer=<PEER>
TRC 2021-09-29 02:40:25.306+01:00 Finished obtaining blocks file=blockchain_sync.nim:303 peer=<PEER>
```
In practice, for modern peers, Nimbus received shorter replies than it assumed
depending on the block sizes on the chain. Geth/Erigon has 2MiB `BlockBodies`
response size soft limit. OpenEthereum has 4MiB.
Up to Berlin (EIP-2929), Nimbus's fetcher failed often, but there were still
some peers serving what Nimbus needed.
Just after the start of Berlin, at batch 4494913..4495104 on Goerli, zero peers
responded with full size replies for the whole batch, so Nimbus couldn't
progress past that point. But there was already a problem happening before
that for large blocks, dropping good peers and repeatedly fetching the same
block data.
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-10-04 17:13:33 +00:00
|
|
|
var bodies = newSeqOfCap[BlockBody](workItem.headers.len)
|
|
|
|
var hashes = newSeqOfCap[KeccakHash](maxBodiesFetch)
|
|
|
|
template fetchBodies() =
|
2022-06-06 13:42:08 +00:00
|
|
|
trace trEthSendSendingGetBlockBodies, peer,
|
2022-05-09 14:04:48 +00:00
|
|
|
hashes=hashes.len
|
Sync fix: `GetBlockBodies` logic preventing sync, dropping peers
Fixes #864 "Sync progress stops at Goerli block 4494913", and equivalent on
other networks.
The block body fetcher in `blockchain_sync.nim` had an incorrect assumption
about how peers respond to `GetBlockBodies`. It was issuing requests for N
block bodies and incorrectly handling replies which contained fewer than N
bodies.
Having received up to 192 headers in a batch, it split the range into smaller
`GetBlockBodies` requests, fetched each reply, then combined replies. The
effect was Nimbus requested batches of 128+64 block bodies, received gaps in
the reply sequence, then aborted.
That meant it repeatedly fetched data, then discarded it, and fetched it again,
dropping good peers in the process.
Aborted and restarted batches occurred with earlier blocks too, but this became
more pronounced until there were no suitable peers at batch 4494913..4495104.
Here's a trace:
```
TRC 2021-09-29 02:40:24.977+01:00 Requesting block headers file=blockchain_sync.nim:224 start=4494913 count=192 peer=<ENODE>
TRC 2021-09-29 02:40:24.977+01:00 >> Sending eth.GetBlockHeaders (0x03) file=protocol_eth65.nim:51 peer=<PEER> startBlock=4494913 max=192
TRC 2021-09-29 02:40:25.005+01:00 << Got reply eth.BlockHeaders (0x04) file=protocol_eth65.nim:51 peer=<PEER> count=192
TRC 2021-09-29 02:40:25.007+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=128
TRC 2021-09-29 02:40:25.209+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=13
TRC 2021-09-29 02:40:25.210+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=64
TRC 2021-09-29 02:40:25.290+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=64
WRN 2021-09-29 02:40:25.306+01:00 Bodies len != headers.len file=blockchain_sync.nim:276 bodies=77 headers=192
TRC 2021-09-29 02:40:25.306+01:00 peer disconnected file=blockchain_sync.nim:403 peer=<PEER>
TRC 2021-09-29 02:40:25.306+01:00 Finished obtaining blocks file=blockchain_sync.nim:303 peer=<PEER>
```
In practice, for modern peers, Nimbus received shorter replies than it assumed
depending on the block sizes on the chain. Geth/Erigon has 2MiB `BlockBodies`
response size soft limit. OpenEthereum has 4MiB.
Up to Berlin (EIP-2929), Nimbus's fetcher failed often, but there were still
some peers serving what Nimbus needed.
Just after the start of Berlin, at batch 4494913..4495104 on Goerli, zero peers
responded with full size replies for the whole batch, so Nimbus couldn't
progress past that point. But there was already a problem happening before
that for large blocks, dropping good peers and repeatedly fetching the same
block data.
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-10-04 17:13:33 +00:00
|
|
|
let b = await peer.getBlockBodies(hashes)
|
|
|
|
if b.isNone:
|
2022-05-23 16:53:19 +00:00
|
|
|
raise newException(
|
|
|
|
CatchableError, "Was not able to get the block bodies")
|
Sync fix: `GetBlockBodies` logic preventing sync, dropping peers
Fixes #864 "Sync progress stops at Goerli block 4494913", and equivalent on
other networks.
The block body fetcher in `blockchain_sync.nim` had an incorrect assumption
about how peers respond to `GetBlockBodies`. It was issuing requests for N
block bodies and incorrectly handling replies which contained fewer than N
bodies.
Having received up to 192 headers in a batch, it split the range into smaller
`GetBlockBodies` requests, fetched each reply, then combined replies. The
effect was Nimbus requested batches of 128+64 block bodies, received gaps in
the reply sequence, then aborted.
That meant it repeatedly fetched data, then discarded it, and fetched it again,
dropping good peers in the process.
Aborted and restarted batches occurred with earlier blocks too, but this became
more pronounced until there were no suitable peers at batch 4494913..4495104.
Here's a trace:
```
TRC 2021-09-29 02:40:24.977+01:00 Requesting block headers file=blockchain_sync.nim:224 start=4494913 count=192 peer=<ENODE>
TRC 2021-09-29 02:40:24.977+01:00 >> Sending eth.GetBlockHeaders (0x03) file=protocol_eth65.nim:51 peer=<PEER> startBlock=4494913 max=192
TRC 2021-09-29 02:40:25.005+01:00 << Got reply eth.BlockHeaders (0x04) file=protocol_eth65.nim:51 peer=<PEER> count=192
TRC 2021-09-29 02:40:25.007+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=128
TRC 2021-09-29 02:40:25.209+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=13
TRC 2021-09-29 02:40:25.210+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=64
TRC 2021-09-29 02:40:25.290+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=64
WRN 2021-09-29 02:40:25.306+01:00 Bodies len != headers.len file=blockchain_sync.nim:276 bodies=77 headers=192
TRC 2021-09-29 02:40:25.306+01:00 peer disconnected file=blockchain_sync.nim:403 peer=<PEER>
TRC 2021-09-29 02:40:25.306+01:00 Finished obtaining blocks file=blockchain_sync.nim:303 peer=<PEER>
```
In practice, for modern peers, Nimbus received shorter replies than it assumed
depending on the block sizes on the chain. Geth/Erigon has 2MiB `BlockBodies`
response size soft limit. OpenEthereum has 4MiB.
Up to Berlin (EIP-2929), Nimbus's fetcher failed often, but there were still
some peers serving what Nimbus needed.
Just after the start of Berlin, at batch 4494913..4495104 on Goerli, zero peers
responded with full size replies for the whole batch, so Nimbus couldn't
progress past that point. But there was already a problem happening before
that for large blocks, dropping good peers and repeatedly fetching the same
block data.
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-10-04 17:13:33 +00:00
|
|
|
let bodiesLen = b.get.blocks.len
|
2022-06-06 13:42:08 +00:00
|
|
|
trace trEthRecvReceivedBlockBodies, peer,
|
2022-05-09 14:04:48 +00:00
|
|
|
count=bodiesLen, requested=hashes.len
|
Sync fix: `GetBlockBodies` logic preventing sync, dropping peers
Fixes #864 "Sync progress stops at Goerli block 4494913", and equivalent on
other networks.
The block body fetcher in `blockchain_sync.nim` had an incorrect assumption
about how peers respond to `GetBlockBodies`. It was issuing requests for N
block bodies and incorrectly handling replies which contained fewer than N
bodies.
Having received up to 192 headers in a batch, it split the range into smaller
`GetBlockBodies` requests, fetched each reply, then combined replies. The
effect was Nimbus requested batches of 128+64 block bodies, received gaps in
the reply sequence, then aborted.
That meant it repeatedly fetched data, then discarded it, and fetched it again,
dropping good peers in the process.
Aborted and restarted batches occurred with earlier blocks too, but this became
more pronounced until there were no suitable peers at batch 4494913..4495104.
Here's a trace:
```
TRC 2021-09-29 02:40:24.977+01:00 Requesting block headers file=blockchain_sync.nim:224 start=4494913 count=192 peer=<ENODE>
TRC 2021-09-29 02:40:24.977+01:00 >> Sending eth.GetBlockHeaders (0x03) file=protocol_eth65.nim:51 peer=<PEER> startBlock=4494913 max=192
TRC 2021-09-29 02:40:25.005+01:00 << Got reply eth.BlockHeaders (0x04) file=protocol_eth65.nim:51 peer=<PEER> count=192
TRC 2021-09-29 02:40:25.007+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=128
TRC 2021-09-29 02:40:25.209+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=13
TRC 2021-09-29 02:40:25.210+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=64
TRC 2021-09-29 02:40:25.290+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=64
WRN 2021-09-29 02:40:25.306+01:00 Bodies len != headers.len file=blockchain_sync.nim:276 bodies=77 headers=192
TRC 2021-09-29 02:40:25.306+01:00 peer disconnected file=blockchain_sync.nim:403 peer=<PEER>
TRC 2021-09-29 02:40:25.306+01:00 Finished obtaining blocks file=blockchain_sync.nim:303 peer=<PEER>
```
In practice, for modern peers, Nimbus received shorter replies than it assumed
depending on the block sizes on the chain. Geth/Erigon has 2MiB `BlockBodies`
response size soft limit. OpenEthereum has 4MiB.
Up to Berlin (EIP-2929), Nimbus's fetcher failed often, but there were still
some peers serving what Nimbus needed.
Just after the start of Berlin, at batch 4494913..4495104 on Goerli, zero peers
responded with full size replies for the whole batch, so Nimbus couldn't
progress past that point. But there was already a problem happening before
that for large blocks, dropping good peers and repeatedly fetching the same
block data.
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-10-04 17:13:33 +00:00
|
|
|
if bodiesLen == 0:
|
|
|
|
raise newException(CatchableError, "Zero block bodies received for request")
|
|
|
|
elif bodiesLen < hashes.len:
|
|
|
|
hashes.delete(0, bodiesLen - 1)
|
|
|
|
elif bodiesLen == hashes.len:
|
|
|
|
hashes.setLen(0)
|
|
|
|
else:
|
|
|
|
raise newException(CatchableError, "Too many block bodies received for request")
|
|
|
|
bodies.add(b.get.blocks)
|
|
|
|
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
var nextIndex = workItem.startIndex
|
|
|
|
for i in workItem.headers:
|
|
|
|
if i.blockNumber != nextIndex:
|
|
|
|
raise newException(CatchableError, "The block numbers are not in sequence. Not processing this workItem.")
|
|
|
|
else:
|
|
|
|
nextIndex = nextIndex + 1
|
|
|
|
hashes.add(blockHash(i))
|
|
|
|
if hashes.len == maxBodiesFetch:
|
Sync fix: `GetBlockBodies` logic preventing sync, dropping peers
Fixes #864 "Sync progress stops at Goerli block 4494913", and equivalent on
other networks.
The block body fetcher in `blockchain_sync.nim` had an incorrect assumption
about how peers respond to `GetBlockBodies`. It was issuing requests for N
block bodies and incorrectly handling replies which contained fewer than N
bodies.
Having received up to 192 headers in a batch, it split the range into smaller
`GetBlockBodies` requests, fetched each reply, then combined replies. The
effect was Nimbus requested batches of 128+64 block bodies, received gaps in
the reply sequence, then aborted.
That meant it repeatedly fetched data, then discarded it, and fetched it again,
dropping good peers in the process.
Aborted and restarted batches occurred with earlier blocks too, but this became
more pronounced until there were no suitable peers at batch 4494913..4495104.
Here's a trace:
```
TRC 2021-09-29 02:40:24.977+01:00 Requesting block headers file=blockchain_sync.nim:224 start=4494913 count=192 peer=<ENODE>
TRC 2021-09-29 02:40:24.977+01:00 >> Sending eth.GetBlockHeaders (0x03) file=protocol_eth65.nim:51 peer=<PEER> startBlock=4494913 max=192
TRC 2021-09-29 02:40:25.005+01:00 << Got reply eth.BlockHeaders (0x04) file=protocol_eth65.nim:51 peer=<PEER> count=192
TRC 2021-09-29 02:40:25.007+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=128
TRC 2021-09-29 02:40:25.209+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=13
TRC 2021-09-29 02:40:25.210+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=64
TRC 2021-09-29 02:40:25.290+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=64
WRN 2021-09-29 02:40:25.306+01:00 Bodies len != headers.len file=blockchain_sync.nim:276 bodies=77 headers=192
TRC 2021-09-29 02:40:25.306+01:00 peer disconnected file=blockchain_sync.nim:403 peer=<PEER>
TRC 2021-09-29 02:40:25.306+01:00 Finished obtaining blocks file=blockchain_sync.nim:303 peer=<PEER>
```
In practice, for modern peers, Nimbus received shorter replies than it assumed
depending on the block sizes on the chain. Geth/Erigon has 2MiB `BlockBodies`
response size soft limit. OpenEthereum has 4MiB.
Up to Berlin (EIP-2929), Nimbus's fetcher failed often, but there were still
some peers serving what Nimbus needed.
Just after the start of Berlin, at batch 4494913..4495104 on Goerli, zero peers
responded with full size replies for the whole batch, so Nimbus couldn't
progress past that point. But there was already a problem happening before
that for large blocks, dropping good peers and repeatedly fetching the same
block data.
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-10-04 17:13:33 +00:00
|
|
|
fetchBodies()
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
Sync fix: `GetBlockBodies` logic preventing sync, dropping peers
Fixes #864 "Sync progress stops at Goerli block 4494913", and equivalent on
other networks.
The block body fetcher in `blockchain_sync.nim` had an incorrect assumption
about how peers respond to `GetBlockBodies`. It was issuing requests for N
block bodies and incorrectly handling replies which contained fewer than N
bodies.
Having received up to 192 headers in a batch, it split the range into smaller
`GetBlockBodies` requests, fetched each reply, then combined replies. The
effect was Nimbus requested batches of 128+64 block bodies, received gaps in
the reply sequence, then aborted.
That meant it repeatedly fetched data, then discarded it, and fetched it again,
dropping good peers in the process.
Aborted and restarted batches occurred with earlier blocks too, but this became
more pronounced until there were no suitable peers at batch 4494913..4495104.
Here's a trace:
```
TRC 2021-09-29 02:40:24.977+01:00 Requesting block headers file=blockchain_sync.nim:224 start=4494913 count=192 peer=<ENODE>
TRC 2021-09-29 02:40:24.977+01:00 >> Sending eth.GetBlockHeaders (0x03) file=protocol_eth65.nim:51 peer=<PEER> startBlock=4494913 max=192
TRC 2021-09-29 02:40:25.005+01:00 << Got reply eth.BlockHeaders (0x04) file=protocol_eth65.nim:51 peer=<PEER> count=192
TRC 2021-09-29 02:40:25.007+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=128
TRC 2021-09-29 02:40:25.209+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=13
TRC 2021-09-29 02:40:25.210+01:00 >> Sending eth.GetBlockBodies (0x05) file=protocol_eth65.nim:51 peer=<PEER> count=64
TRC 2021-09-29 02:40:25.290+01:00 << Got reply eth.BlockBodies (0x06) file=protocol_eth65.nim:51 peer=<PEER> count=64
WRN 2021-09-29 02:40:25.306+01:00 Bodies len != headers.len file=blockchain_sync.nim:276 bodies=77 headers=192
TRC 2021-09-29 02:40:25.306+01:00 peer disconnected file=blockchain_sync.nim:403 peer=<PEER>
TRC 2021-09-29 02:40:25.306+01:00 Finished obtaining blocks file=blockchain_sync.nim:303 peer=<PEER>
```
In practice, for modern peers, Nimbus received shorter replies than it assumed
depending on the block sizes on the chain. Geth/Erigon has 2MiB `BlockBodies`
response size soft limit. OpenEthereum has 4MiB.
Up to Berlin (EIP-2929), Nimbus's fetcher failed often, but there were still
some peers serving what Nimbus needed.
Just after the start of Berlin, at batch 4494913..4495104 on Goerli, zero peers
responded with full size replies for the whole batch, so Nimbus couldn't
progress past that point. But there was already a problem happening before
that for large blocks, dropping good peers and repeatedly fetching the same
block data.
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-10-04 17:13:33 +00:00
|
|
|
while hashes.len != 0:
|
|
|
|
fetchBodies()
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
|
|
|
if bodies.len == workItem.headers.len:
|
|
|
|
shallowCopy(workItem.bodies, bodies)
|
|
|
|
dataReceived = true
|
|
|
|
else:
|
|
|
|
warn "Bodies len != headers.len", bodies = bodies.len, headers = workItem.headers.len
|
|
|
|
except TransportError:
|
|
|
|
debug "Transport got closed during obtainBlocksFromPeer"
|
|
|
|
except CatchableError as e:
|
|
|
|
# the success case sets `dataReceived`, so we can just fall back to the
|
|
|
|
# failure path below. If we signal time-outs with exceptions such
|
|
|
|
# failures will be easier to handle.
|
|
|
|
debug "Exception in obtainBlocksFromPeer()", exc = e.name, err = e.msg
|
|
|
|
|
|
|
|
var giveUpOnPeer = false
|
|
|
|
|
|
|
|
if dataReceived:
|
|
|
|
workItem.state = Received
|
|
|
|
if syncCtx.returnWorkItem(workItemIdx) != ValidationResult.OK:
|
|
|
|
giveUpOnPeer = true
|
|
|
|
else:
|
|
|
|
giveUpOnPeer = true
|
|
|
|
|
|
|
|
if giveUpOnPeer:
|
|
|
|
workItem.state = Initial
|
|
|
|
try:
|
|
|
|
await peer.disconnect(SubprotocolReason)
|
|
|
|
except CatchableError:
|
|
|
|
discard
|
|
|
|
syncCtx.handleLostPeer()
|
|
|
|
break
|
|
|
|
|
|
|
|
trace "Finished obtaining blocks", peer
|
|
|
|
|
|
|
|
proc peersAgreeOnChain(a, b: Peer): Future[bool] {.async.} =
|
|
|
|
# Returns true if one of the peers acknowledges existence of the best block
|
|
|
|
# of another peer.
|
|
|
|
var
|
|
|
|
a = a
|
|
|
|
b = b
|
|
|
|
|
|
|
|
if a.state(eth).bestDifficulty < b.state(eth).bestDifficulty:
|
|
|
|
swap(a, b)
|
|
|
|
|
|
|
|
let request = BlocksRequest(
|
|
|
|
startBlock: HashOrNum(isHash: true,
|
|
|
|
hash: b.state(eth).bestBlockHash),
|
|
|
|
maxResults: 1,
|
|
|
|
skip: 0,
|
|
|
|
reverse: true)
|
|
|
|
|
2022-06-06 13:42:08 +00:00
|
|
|
trace trEthSendSendingGetBlockHeaders, peer=a,
|
2022-03-21 17:12:07 +00:00
|
|
|
startBlock=request.startBlock.hash.toHex, max=request.maxResults
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
let latestBlock = await a.getBlockHeaders(request)
|
2021-07-22 16:36:17 +00:00
|
|
|
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
result = latestBlock.isSome and latestBlock.get.headers.len > 0
|
2022-05-23 16:53:19 +00:00
|
|
|
if latestBlock.isSome:
|
2021-07-22 16:36:17 +00:00
|
|
|
let blockNumber = if result: $latestBlock.get.headers[0].blockNumber
|
|
|
|
else: "missing"
|
2022-06-06 13:42:08 +00:00
|
|
|
trace trEthRecvReceivedBlockHeaders, peer=a,
|
2021-07-22 16:36:17 +00:00
|
|
|
count=latestBlock.get.headers.len, blockNumber
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc randomTrustedPeer(ctx: FastSyncCtx): Peer =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
var k = rand(ctx.trustedPeers.len - 1)
|
|
|
|
var i = 0
|
|
|
|
for p in ctx.trustedPeers:
|
|
|
|
result = p
|
|
|
|
if i == k: return
|
|
|
|
inc i
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc startSyncWithPeer(ctx: FastSyncCtx, peer: Peer) {.async.} =
|
2022-05-23 16:53:19 +00:00
|
|
|
trace "Start sync", peer, trustedPeers = ctx.trustedPeers.len
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
if ctx.trustedPeers.len >= minPeersToStartSync:
|
|
|
|
# We have enough trusted peers. Validate new peer against trusted
|
|
|
|
if await peersAgreeOnChain(peer, ctx.randomTrustedPeer()):
|
|
|
|
ctx.trustedPeers.incl(peer)
|
2022-05-09 14:04:48 +00:00
|
|
|
asyncSpawn ctx.obtainBlocksFromPeer(peer)
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
elif ctx.trustedPeers.len == 0:
|
|
|
|
# Assume the peer is trusted, but don't start sync until we reevaluate
|
|
|
|
# it with more peers
|
|
|
|
trace "Assume trusted peer", peer
|
|
|
|
ctx.trustedPeers.incl(peer)
|
|
|
|
else:
|
|
|
|
# At this point we have some "trusted" candidates, but they are not
|
|
|
|
# "trusted" enough. We evaluate `peer` against all other candidates.
|
|
|
|
# If one of the candidates disagrees, we swap it for `peer`. If all
|
|
|
|
# candidates agree, we add `peer` to trusted set. The peers in the set
|
|
|
|
# will become "fully trusted" (and sync will start) when the set is big
|
|
|
|
# enough
|
|
|
|
var
|
|
|
|
agreeScore = 0
|
|
|
|
disagreedPeer: Peer
|
|
|
|
|
|
|
|
for tp in ctx.trustedPeers:
|
|
|
|
if await peersAgreeOnChain(peer, tp):
|
|
|
|
inc agreeScore
|
|
|
|
else:
|
|
|
|
disagreedPeer = tp
|
|
|
|
|
|
|
|
let disagreeScore = ctx.trustedPeers.len - agreeScore
|
|
|
|
|
|
|
|
if agreeScore == ctx.trustedPeers.len:
|
|
|
|
ctx.trustedPeers.incl(peer) # The best possible outcome
|
|
|
|
elif disagreeScore == 1:
|
|
|
|
trace "Peer is no longer trusted for sync", peer
|
|
|
|
ctx.trustedPeers.excl(disagreedPeer)
|
|
|
|
ctx.trustedPeers.incl(peer)
|
|
|
|
else:
|
|
|
|
trace "Peer not trusted for sync", peer
|
|
|
|
|
|
|
|
if ctx.trustedPeers.len == minPeersToStartSync:
|
|
|
|
for p in ctx.trustedPeers:
|
2022-05-09 14:04:48 +00:00
|
|
|
asyncSpawn ctx.obtainBlocksFromPeer(p)
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc onPeerConnected(ctx: FastSyncCtx, peer: Peer) =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
trace "New candidate for sync", peer
|
|
|
|
try:
|
|
|
|
let f = ctx.startSyncWithPeer(peer)
|
|
|
|
f.callback = proc(data: pointer) {.gcsafe.} =
|
|
|
|
if f.failed:
|
|
|
|
if f.error of TransportError:
|
|
|
|
debug "Transport got closed during startSyncWithPeer"
|
|
|
|
else:
|
|
|
|
error "startSyncWithPeer failed", msg = f.readError.msg, peer
|
|
|
|
except TransportError:
|
|
|
|
debug "Transport got closed during startSyncWithPeer"
|
|
|
|
except CatchableError as e:
|
|
|
|
debug "Exception in startSyncWithPeer()", exc = e.name, err = e.msg
|
|
|
|
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc onPeerDisconnected(ctx: FastSyncCtx, p: Peer) =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
trace "peer disconnected ", peer = p
|
|
|
|
ctx.trustedPeers.excl(p)
|
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
proc new*(T: type FastSyncCtx; ethNode: EthereumNode): T
|
|
|
|
{.gcsafe, raises:[Defect,CatchableError].} =
|
|
|
|
FastSyncCtx(
|
|
|
|
# workQueue: n/a
|
|
|
|
# endBlockNumber: n/a
|
|
|
|
# hasOutOfOrderBlocks: n/a
|
|
|
|
chain: ethNode.chain,
|
|
|
|
peerPool: ethNode.peerPool,
|
|
|
|
trustedPeers: initHashSet[Peer](),
|
|
|
|
finalizedBlock: ethNode.chain.getBestBlockHeader.blockNumber)
|
|
|
|
|
|
|
|
proc start*(ctx: FastSyncCtx) =
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
## Code for the fast blockchain sync procedure:
|
2022-05-13 16:30:10 +00:00
|
|
|
## <https://github.com/ethereum/wiki/wiki/Parallel-Block-Downloads>_
|
|
|
|
## <https://github.com/ethereum/go-ethereum/pull/1889__
|
|
|
|
var po = PeerObserver(
|
|
|
|
onPeerConnected:
|
|
|
|
proc(p: Peer) {.gcsafe.} =
|
|
|
|
ctx.onPeerConnected(p),
|
|
|
|
onPeerDisconnected:
|
|
|
|
proc(p: Peer) {.gcsafe.} =
|
|
|
|
ctx.onPeerDisconnected(p))
|
|
|
|
po.setProtocol eth
|
|
|
|
ctx.peerPool.addObserver(ctx, po)
|
Sync: Move `blockchain_sync` code and use it with `eth/65`
Move `blockchain_sync.nim` from `nim-eth` to `nimbus-eth1`.
This lets `blockchain_sync` use the `eth/65` protocol to synchronise with more
modern peers than before.
Practically, the effect is the sync process runs more quickly and reliably than
before. It finds usable peers, and they are up to date.
Note, this is mostly old code, and it mostly performs "classic sync", the
original Ethereum method. Here's a summary of this code:
- It decides on a blockchain canonical head by sampling a few peers.
- Starting from block 0 (genesis), it downloads each block header and
block, mostly in order.
- After it downloads each block, it executes the EVM transactions in that block
and updates state trie from that, before going to the next block.
- This way the database state is updated by EVM executions in block order,
and new state is persisted to the trie database after each block.
Even though it mentions Geth "fast sync" (comments near end of file), and has
some elements, it isn't really. The most obvious missing part is this code
_doesn't download a state trie_, it calculates all state from block 0.
Geth "fast sync" has several parts:
1. Find an agreed common chain among several peers to treat as probably secure,
and a sufficiently long suffix to provide "statistical economic consensus"
when it is validated.
2. Perform a subset of PoW calculations, skipping forward over a segment to
verify some of the PoWs according to a pattern in the relevant paper.
3. Download the state trie from the block at the start of that last segment.
4. Execute only the blocks/transactions in that last segment, using the
downloaded state trie, to fill out the later states and properly validate the
blocks in the last segment.
Some other issues with `blockchain_sync` code:
- If it ever reaches the head of the chain, it doesn't follow new blocks with
increasing block numbers, at least not rapidly.
- If the chain undergoes a reorg, this code won't fetch a block number it has
already fetched, so it can't accept the reorg. It will end up conflicted
with peers. This hasn't mattered because the development focus has been on
the bulk of the catching up process, not the real-time head and reorgs.
- So it probably doesn't work correctly when it gets close to the head due to
many small reorgs, though it might for subtle reasons.
- Some of the network message handling isn't sufficiently robust, and it
discards some replies that have valid data according to specification.
- On rare occasions the initial query mapping block hash to number can
fail (because the peer's state changes).
- It makes some assumptions about the state of peers based on their responses
which may not be valid (I'm not convinced they are). The method for working
out "trusted" peers that agree a common chain prefix is clever. It compares
peers by asking each peer if it has the header matching another peer's
canonical head block by hash. But it's not clear that merely knowing about a
block constitutes agreement about the canonical chain. (If it did, query by
block number would give the same answer more authoritatively.)
Nonetheless, being able to run this sync process on `eth/65` is useful.
<# interactive rebase in progress; onto 66532e8a
Signed-off-by: Jamie Lokier <jamie@shareable.org>
2021-07-22 13:36:10 +00:00
|
|
|
|
2022-05-13 16:30:10 +00:00
|
|
|
# End
|