nimbus-eth1/nimbus/sync/snap/worker/com/get_trie_nodes.nim

166 lines
5.2 KiB
Nim
Raw Normal View History

Prep for full sync after snap (#1253) * Split fetch accounts into sub-modules details: There will be separated modules for accounts snapshot, storage snapshot, and healing for either. * Allow to rebase pivot before negotiated header why: Peers seem to have not too many snapshots available. By setting back the pivot block header slightly, the chances might be higher to find more peers to serve this pivot. Experiment on mainnet showed that setting back too much (tested with 1024), the chances to find matching snapshot peers seem to decrease. * Add accounts healing * Update variable/field naming in `worker_desc` for readability * Handle leaf nodes in accounts healing why: There is no need to fetch accounts when they had been added by the healing process. On the flip side, these accounts must be checked for storage data and the batch queue updated, accordingly. * Reorganising accounts hash ranges batch queue why: The aim is to formally cover as many accounts as possible for different pivot state root environments. Formerly, this was tried by starting the accounts batch queue at a random value for each pivot (and wrapping around.) Now, each pivot environment starts with an interval set mutually disjunct from any interval set retrieved with other pivot state roots. also: Stop fishing for more pivots in `worker` if 100% download is reached * Reorganise/update accounts healing why: Error handling was wrong and the (math. complexity of) whole process could be better managed. details: Much of the algorithm is now documented at the top of the file `heal_accounts.nim`
2022-10-08 17:20:50 +00:00
# Nimbus
# Copyright (c) 2018-2021 Status Research & Development GmbH
# Licensed and distributed under either of
# * Apache License, version 2.0, ([LICENSE-APACHE](LICENSE-APACHE) or
# http://www.apache.org/licenses/LICENSE-2.0)
# * MIT license ([LICENSE-MIT](LICENSE-MIT) or
# http://opensource.org/licenses/MIT)
# at your option. This file may not be copied, modified, or distributed
# except according to those terms.
import
std/[options, sequtils],
chronos,
eth/[common, p2p],
"../../.."/[protocol, protocol/trace_config],
"../.."/[constants, range_desc, worker_desc],
Prep for full sync after snap (#1253) * Split fetch accounts into sub-modules details: There will be separated modules for accounts snapshot, storage snapshot, and healing for either. * Allow to rebase pivot before negotiated header why: Peers seem to have not too many snapshots available. By setting back the pivot block header slightly, the chances might be higher to find more peers to serve this pivot. Experiment on mainnet showed that setting back too much (tested with 1024), the chances to find matching snapshot peers seem to decrease. * Add accounts healing * Update variable/field naming in `worker_desc` for readability * Handle leaf nodes in accounts healing why: There is no need to fetch accounts when they had been added by the healing process. On the flip side, these accounts must be checked for storage data and the batch queue updated, accordingly. * Reorganising accounts hash ranges batch queue why: The aim is to formally cover as many accounts as possible for different pivot state root environments. Formerly, this was tried by starting the accounts batch queue at a random value for each pivot (and wrapping around.) Now, each pivot environment starts with an interval set mutually disjunct from any interval set retrieved with other pivot state roots. also: Stop fishing for more pivots in `worker` if 100% download is reached * Reorganise/update accounts healing why: Error handling was wrong and the (math. complexity of) whole process could be better managed. details: Much of the algorithm is now documented at the top of the file `heal_accounts.nim`
2022-10-08 17:20:50 +00:00
./com_error
{.push raises: [Defect].}
logScope:
topics = "snap-fetch"
type
# SnapTrieNodes = object
# nodes*: seq[Blob]
GetTrieNodes* = object
leftOver*: seq[seq[Blob]]
Prep for full sync after snap make 4 (#1282) * Re-arrange fetching storage slots in batch module why; Previously, fetching partial slot ranges first has a chance of terminating the worker peer 9due to network error) while there were many inheritable storage slots on the queue. Now, inheritance is checked first, then full slot ranges and finally partial ranges. * Update logging * Bundled node information for healing into single object `NodeSpecs` why: Previously, partial paths and node keys were kept in separate variables. This approach was error prone due to copying/reassembling function argument objects. As all partial paths, keys, and node data types are more or less handled as `Blob`s over the network (using Eth/6x, or Snap/1) it makes sense to hold these `Blob`s as named field in a single object (even if not all fields are active for the current purpose.) * For good housekeeping, using `NodeKey` type only for account keys why: previously, a mixture of `NodeKey` and `Hash256` was used. Now, only state or storage root keys use the `Hash256` type. * Always accept latest pivot (and not a slightly older one) why; For testing it was tried to use a slightly older pivot state root than available. Some anecdotal tests seemed to suggest an advantage so that more peers are willing to serve on that older pivot. But this could not be confirmed in subsequent tests (still anecdotal, though.) As a side note, the distance of the latest pivot to its predecessor is at least 128 (or whatever the constant `minPivotBlockDistance` is assigned to.) * Reshuffle name components for some file and function names why: Clarifies purpose: "storages" becomes: "storage slots" "store" becomes: "range fetch" * Stash away currently unused modules in sub-folder named "notused"
2022-10-27 13:49:28 +00:00
nodes*: seq[NodeSpecs] ## `nodeKey` field unused with `NodeSpecs`
# ------------------------------------------------------------------------------
# Private functions
# ------------------------------------------------------------------------------
proc getTrieNodesReq(
buddy: SnapBuddyRef;
stateRoot: Hash256;
paths: seq[seq[Blob]];
pivot: string;
): Future[Result[Option[SnapTrieNodes],void]]
{.async.} =
let
peer = buddy.peer
try:
Snap sync refactor healing (#1397) * Simplify accounts healing threshold management why: Was over-engineered. details: Previously, healing was based on recursive hexary trie perusal. Due to "cheap" envelope decomposition of a range complement for the hexary trie, the cost of running extra laps have become time-affordable again and a simple trigger mechanism for healing will do. * Control number of dangling result nodes in `hexaryInspectTrie()` also: + Returns number of visited nodes available for logging so the maximum number of nodes can be tuned accordingly. + Some code and docu update * Update names of constants why: Declutter, more systematic naming * Re-implemented `worker_desc.merge()` for storage slots why: Provided as proper queue management in `storage_queue_helper`. details: + Several append modes (replaces `merge()`) + Added third queue to record entries currently fetched by a worker. So another parallel running worker can safe the complete set of storage slots in as checkpoint. This was previously lost. * Refactor healing why: Simplify and remove deep hexary trie perusal for finding completeness. Due to "cheap" envelope decomposition of a range complement for the hexary trie, the cost of running extra laps have become time-affordable again and a simple trigger mechanism for healing will do. * Docu update * Run a storage job only once in download loop why: Download failure or rejection (i.e. missing data) lead to repeated fetch requests until peer disconnects, otherwise.
2022-12-24 09:54:18 +00:00
let reply = await peer.getTrieNodes(
stateRoot, paths, fetchRequestBytesLimit)
return ok(reply)
except CatchableError as e:
trace trSnapRecvError & "waiting for GetByteCodes reply", peer, pivot,
error=e.msg
return err()
# ------------------------------------------------------------------------------
# Public functions
# ------------------------------------------------------------------------------
proc getTrieNodes*(
buddy: SnapBuddyRef;
stateRoot: Hash256; ## Current DB base (see `pivot` for logging)
paths: seq[seq[Blob]]; ## Nodes to fetch
pivot: string; ## For logging, instead of `stateRoot`
): Future[Result[GetTrieNodes,ComError]]
{.async.} =
## Fetch data using the `snap#` protocol, returns the trie nodes requested
## (if any.)
let
peer = buddy.peer
nPaths = paths.len
if nPaths == 0:
return err(ComEmptyRequestArguments)
let nTotal = paths.mapIt(it.len).foldl(a+b, 0)
if trSnapTracePacketsOk:
Snap sync refactor healing (#1397) * Simplify accounts healing threshold management why: Was over-engineered. details: Previously, healing was based on recursive hexary trie perusal. Due to "cheap" envelope decomposition of a range complement for the hexary trie, the cost of running extra laps have become time-affordable again and a simple trigger mechanism for healing will do. * Control number of dangling result nodes in `hexaryInspectTrie()` also: + Returns number of visited nodes available for logging so the maximum number of nodes can be tuned accordingly. + Some code and docu update * Update names of constants why: Declutter, more systematic naming * Re-implemented `worker_desc.merge()` for storage slots why: Provided as proper queue management in `storage_queue_helper`. details: + Several append modes (replaces `merge()`) + Added third queue to record entries currently fetched by a worker. So another parallel running worker can safe the complete set of storage slots in as checkpoint. This was previously lost. * Refactor healing why: Simplify and remove deep hexary trie perusal for finding completeness. Due to "cheap" envelope decomposition of a range complement for the hexary trie, the cost of running extra laps have become time-affordable again and a simple trigger mechanism for healing will do. * Docu update * Run a storage job only once in download loop why: Download failure or rejection (i.e. missing data) lead to repeated fetch requests until peer disconnects, otherwise.
2022-12-24 09:54:18 +00:00
trace trSnapSendSending & "GetTrieNodes", peer, pivot, nPaths, nTotal
let trieNodes = block:
let rc = await buddy.getTrieNodesReq(stateRoot, paths, pivot)
if rc.isErr:
return err(ComNetworkProblem)
if rc.value.isNone:
trace trSnapRecvTimeoutWaiting & "for TrieNodes", peer, pivot, nPaths
return err(ComResponseTimeout)
let blobs = rc.value.get.nodes
if nTotal < blobs.len:
# Ooops, makes no sense
return err(ComTooManyTrieNodes)
blobs
let
nNodes = trieNodes.len
if nNodes == 0:
# github.com/ethereum/devp2p/blob/master/caps/snap.md#gettrienodes-0x06
#
# Notes:
# * Nodes must always respond to the query.
# * The returned nodes must be in the request order.
# * If the node does not have the state for the requested state root or for
# any requested account paths, it must return an empty reply. It is the
# responsibility of the caller to query an state not older than 128
# blocks; and the caller is expected to only ever query existing trie
# nodes.
# * The responding node is allowed to return less data than requested
# (serving QoS limits), but the node must return at least one trie node.
trace trSnapRecvReceived & "empty TrieNodes", peer, pivot, nPaths, nNodes
return err(ComNoByteCodesAvailable)
# Assemble return value
Prep for full sync after snap make 4 (#1282) * Re-arrange fetching storage slots in batch module why; Previously, fetching partial slot ranges first has a chance of terminating the worker peer 9due to network error) while there were many inheritable storage slots on the queue. Now, inheritance is checked first, then full slot ranges and finally partial ranges. * Update logging * Bundled node information for healing into single object `NodeSpecs` why: Previously, partial paths and node keys were kept in separate variables. This approach was error prone due to copying/reassembling function argument objects. As all partial paths, keys, and node data types are more or less handled as `Blob`s over the network (using Eth/6x, or Snap/1) it makes sense to hold these `Blob`s as named field in a single object (even if not all fields are active for the current purpose.) * For good housekeeping, using `NodeKey` type only for account keys why: previously, a mixture of `NodeKey` and `Hash256` was used. Now, only state or storage root keys use the `Hash256` type. * Always accept latest pivot (and not a slightly older one) why; For testing it was tried to use a slightly older pivot state root than available. Some anecdotal tests seemed to suggest an advantage so that more peers are willing to serve on that older pivot. But this could not be confirmed in subsequent tests (still anecdotal, though.) As a side note, the distance of the latest pivot to its predecessor is at least 128 (or whatever the constant `minPivotBlockDistance` is assigned to.) * Reshuffle name components for some file and function names why: Clarifies purpose: "storages" becomes: "storage slots" "store" becomes: "range fetch" * Stash away currently unused modules in sub-folder named "notused"
2022-10-27 13:49:28 +00:00
var dd = GetTrieNodes()
# For each request group/sub-sequence, analyse the results
var nInx = 0
block loop:
for n in 0 ..< nPaths:
let pathLen = paths[n].len
# Account node request
if pathLen < 2:
if trieNodes[nInx].len == 0:
dd.leftOver.add paths[n]
Prep for full sync after snap make 4 (#1282) * Re-arrange fetching storage slots in batch module why; Previously, fetching partial slot ranges first has a chance of terminating the worker peer 9due to network error) while there were many inheritable storage slots on the queue. Now, inheritance is checked first, then full slot ranges and finally partial ranges. * Update logging * Bundled node information for healing into single object `NodeSpecs` why: Previously, partial paths and node keys were kept in separate variables. This approach was error prone due to copying/reassembling function argument objects. As all partial paths, keys, and node data types are more or less handled as `Blob`s over the network (using Eth/6x, or Snap/1) it makes sense to hold these `Blob`s as named field in a single object (even if not all fields are active for the current purpose.) * For good housekeeping, using `NodeKey` type only for account keys why: previously, a mixture of `NodeKey` and `Hash256` was used. Now, only state or storage root keys use the `Hash256` type. * Always accept latest pivot (and not a slightly older one) why; For testing it was tried to use a slightly older pivot state root than available. Some anecdotal tests seemed to suggest an advantage so that more peers are willing to serve on that older pivot. But this could not be confirmed in subsequent tests (still anecdotal, though.) As a side note, the distance of the latest pivot to its predecessor is at least 128 (or whatever the constant `minPivotBlockDistance` is assigned to.) * Reshuffle name components for some file and function names why: Clarifies purpose: "storages" becomes: "storage slots" "store" becomes: "range fetch" * Stash away currently unused modules in sub-folder named "notused"
2022-10-27 13:49:28 +00:00
else:
dd.nodes.add NodeSpecs(
partialPath: paths[n][0],
data: trieNodes[nInx])
nInx.inc
if nInx < nNodes:
continue
# all the rest needs to be re-processed
dd.leftOver = dd.leftOver & paths[n+1 ..< nPaths]
break loop
Prep for full sync after snap make 4 (#1282) * Re-arrange fetching storage slots in batch module why; Previously, fetching partial slot ranges first has a chance of terminating the worker peer 9due to network error) while there were many inheritable storage slots on the queue. Now, inheritance is checked first, then full slot ranges and finally partial ranges. * Update logging * Bundled node information for healing into single object `NodeSpecs` why: Previously, partial paths and node keys were kept in separate variables. This approach was error prone due to copying/reassembling function argument objects. As all partial paths, keys, and node data types are more or less handled as `Blob`s over the network (using Eth/6x, or Snap/1) it makes sense to hold these `Blob`s as named field in a single object (even if not all fields are active for the current purpose.) * For good housekeeping, using `NodeKey` type only for account keys why: previously, a mixture of `NodeKey` and `Hash256` was used. Now, only state or storage root keys use the `Hash256` type. * Always accept latest pivot (and not a slightly older one) why; For testing it was tried to use a slightly older pivot state root than available. Some anecdotal tests seemed to suggest an advantage so that more peers are willing to serve on that older pivot. But this could not be confirmed in subsequent tests (still anecdotal, though.) As a side note, the distance of the latest pivot to its predecessor is at least 128 (or whatever the constant `minPivotBlockDistance` is assigned to.) * Reshuffle name components for some file and function names why: Clarifies purpose: "storages" becomes: "storage slots" "store" becomes: "range fetch" * Stash away currently unused modules in sub-folder named "notused"
2022-10-27 13:49:28 +00:00
# Storage request for account followed by storage slot paths
if 1 < pathLen:
var pushBack: seq[Blob]
for i in 1 ..< pathLen:
if trieNodes[nInx].len == 0:
pushBack.add paths[n][i]
Prep for full sync after snap make 4 (#1282) * Re-arrange fetching storage slots in batch module why; Previously, fetching partial slot ranges first has a chance of terminating the worker peer 9due to network error) while there were many inheritable storage slots on the queue. Now, inheritance is checked first, then full slot ranges and finally partial ranges. * Update logging * Bundled node information for healing into single object `NodeSpecs` why: Previously, partial paths and node keys were kept in separate variables. This approach was error prone due to copying/reassembling function argument objects. As all partial paths, keys, and node data types are more or less handled as `Blob`s over the network (using Eth/6x, or Snap/1) it makes sense to hold these `Blob`s as named field in a single object (even if not all fields are active for the current purpose.) * For good housekeeping, using `NodeKey` type only for account keys why: previously, a mixture of `NodeKey` and `Hash256` was used. Now, only state or storage root keys use the `Hash256` type. * Always accept latest pivot (and not a slightly older one) why; For testing it was tried to use a slightly older pivot state root than available. Some anecdotal tests seemed to suggest an advantage so that more peers are willing to serve on that older pivot. But this could not be confirmed in subsequent tests (still anecdotal, though.) As a side note, the distance of the latest pivot to its predecessor is at least 128 (or whatever the constant `minPivotBlockDistance` is assigned to.) * Reshuffle name components for some file and function names why: Clarifies purpose: "storages" becomes: "storage slots" "store" becomes: "range fetch" * Stash away currently unused modules in sub-folder named "notused"
2022-10-27 13:49:28 +00:00
else:
dd.nodes.add NodeSpecs(
partialPath: paths[n][i],
data: trieNodes[nInx])
nInx.inc
if nInx < nNodes:
continue
# all the rest needs to be re-processed
#
# add: account & pushBack & rest ...
dd.leftOver.add paths[n][0] & pushBack & paths[n][i+1 ..< pathLen]
dd.leftOver = dd.leftOver & paths[n+1 ..< nPaths]
break loop
if 0 < pushBack.len:
dd.leftOver.add paths[n][0] & pushBack
trace trSnapRecvReceived & "TrieNodes", peer, pivot,
nPaths, nNodes, nLeftOver=dd.leftOver.len
return ok(dd)
# ------------------------------------------------------------------------------
# End
# ------------------------------------------------------------------------------