nim-dagger/codex/node.nim
Eric 570a1f7b67
[marketplace] Availability improvements (#535)
## Problem
When Availabilities are created, the amount of bytes in the Availability are reserved in the repo, so those bytes on disk cannot be written to otherwise. When a request for storage is received by a node, if a previously created Availability is matched, an attempt will be made to fill a slot in the request (more accurately, the request's slots are added to the SlotQueue, and eventually those slots will be processed). During download, bytes that were reserved for the Availability were released (as they were written to disk). To prevent more bytes from being released than were reserved in the Availability, the Availability was marked as used during the download, so that no other requests would match the Availability, and therefore no new downloads (and byte releases) would begin. The unfortunate downside to this, is that the number of Availabilities a node has determines the download concurrency capacity. If, for example, a node creates a single Availability that covers all available disk space the operator is willing to use, that single Availability would mean that only one download could occur at a time, meaning the node could potentially miss out on storage opportunities.

## Solution
To alleviate the concurrency issue, each time a slot is processed, a Reservation is created, which takes size (aka reserved bytes) away from the Availability and stores them in the Reservation object. This can be done as many times as needed as long as there are enough bytes remaining in the Availability. Therefore, concurrent downloads are no longer limited by the number of Availabilities. Instead, they would more likely be limited to the SlotQueue's `maxWorkers`.

From a database design perspective, an Availability has zero or more Reservations.

Reservations are persisted in the RepoStore's metadata, along with Availabilities. The metadata store key path for Reservations is ` meta / sales / reservations / <availabilityId> / <reservationId>`, while Availabilities are stored one level up, eg `meta / sales / reservations / <availabilityId> `, allowing all Reservations for an Availability to be queried (this is not currently needed, but may be useful when work to restore Availability size is implemented, more on this later).

### Lifecycle
When a reservation is created, its size is deducted from the Availability, and when a reservation is deleted, any remaining size (bytes not written to disk) is returned to the Availability. If the request finishes, is cancelled (expired), or an error occurs, the Reservation is deleted (and any undownloaded bytes returned to the Availability). In addition, when the Sales module starts, any Reservations that are not actively being used in a filled slot, are deleted.

Having a Reservation persisted until after a storage request is completed, will allow for the originally set Availability size to be reclaimed once a request contract has been completed. This is a feature that is yet to be implemented, however the work in this PR is a step in the direction towards enabling this.

### Unknowns
Reservation size is determined by the `StorageAsk.slotSize`. If during download, more bytes than `slotSize` are attempted to be downloaded than this, then the Reservation update will fail, and the state machine will move to a `SaleErrored` state, deleting the Reservation. This will likely prevent the slot from being filled.

### Notes
Based on #514
2023-09-29 14:33:08 +10:00

429 lines
13 KiB
Nim
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

## Nim-Codex
## Copyright (c) 2021 Status Research & Development GmbH
## Licensed under either of
## * Apache License, version 2.0, ([LICENSE-APACHE](LICENSE-APACHE))
## * MIT license ([LICENSE-MIT](LICENSE-MIT))
## at your option.
## This file may not be copied, modified, or distributed except according to
## those terms.
import std/options
import std/tables
import std/sequtils
import std/strformat
import pkg/questionable
import pkg/questionable/results
import pkg/chronicles
import pkg/chronos
import pkg/libp2p/switch
import pkg/libp2p/stream/bufferstream
# TODO: remove once exported by libp2p
import pkg/libp2p/routing_record
import pkg/libp2p/signed_envelope
import ./chunker
import ./blocktype as bt
import ./manifest
import ./stores/blockstore
import ./blockexchange
import ./streams
import ./erasure
import ./discovery
import ./contracts
import ./node/batch
export batch
logScope:
topics = "codex node"
const
FetchBatch = 200
type
CodexError = object of CatchableError
Contracts* = tuple
client: ?ClientInteractions
host: ?HostInteractions
validator: ?ValidatorInteractions
CodexNodeRef* = ref object
switch*: Switch
networkId*: PeerId
blockStore*: BlockStore
engine*: BlockExcEngine
erasure*: Erasure
discovery*: Discovery
contracts*: Contracts
proc findPeer*(
node: CodexNodeRef,
peerId: PeerId): Future[?PeerRecord] {.async.} =
## Find peer using the discovery service from the given CodexNode
##
return await node.discovery.findPeer(peerId)
proc connect*(
node: CodexNodeRef,
peerId: PeerId,
addrs: seq[MultiAddress]
): Future[void] =
node.switch.connect(peerId, addrs)
proc fetchManifest*(
node: CodexNodeRef,
cid: Cid): Future[?!Manifest] {.async.} =
## Fetch and decode a manifest block
##
if err =? cid.isManifest.errorOption:
return failure "CID has invalid content type for manifest {$cid}"
trace "Retrieving manifest for cid", cid
without blk =? await node.blockStore.getBlock(cid), err:
trace "Error retrieve manifest block", cid, err = err.msg
return failure err
trace "Decoding manifest for cid", cid
without manifest =? Manifest.decode(blk), err:
trace "Unable to decode as manifest", err = err.msg
return failure("Unable to decode as manifest")
trace "Decoded manifest", cid
return manifest.success
proc fetchBatched*(
node: CodexNodeRef,
manifest: Manifest,
batchSize = FetchBatch,
onBatch: BatchProc = nil): Future[?!void] {.async, gcsafe.} =
## Fetch manifest in batches of `batchSize`
##
let
batches =
(manifest.blocks.len div batchSize) +
(manifest.blocks.len mod batchSize)
trace "Fetching blocks in batches of", size = batchSize
for blks in manifest.blocks.distribute(max(1, batches), true):
try:
let
blocks = blks.mapIt(node.blockStore.getBlock( it ))
await allFuturesThrowing(allFinished(blocks))
if not onBatch.isNil:
await onBatch(blocks.mapIt( it.read.get ))
except CancelledError as exc:
raise exc
except CatchableError as exc:
return failure(exc.msg)
return success()
proc retrieve*(
node: CodexNodeRef,
cid: Cid): Future[?!LPStream] {.async.} =
## Retrieve by Cid a single block or an entire dataset described by manifest
##
if manifest =? (await node.fetchManifest(cid)):
trace "Retrieving blocks from manifest", cid
if manifest.protected:
# Retrieve, decode and save to the local store all EС groups
proc erasureJob(): Future[void] {.async.} =
try:
# Spawn an erasure decoding job
without res =? (await node.erasure.decode(manifest)), error:
trace "Unable to erasure decode manifest", cid, exc = error.msg
except CatchableError as exc:
trace "Exception decoding manifest", cid, exc = exc.msg
asyncSpawn erasureJob()
# Retrieve all blocks of the dataset sequentially from the local store or network
trace "Creating store stream for manifest", cid
LPStream(StoreStream.new(node.blockStore, manifest, pad = false)).success
else:
let
stream = BufferStream.new()
without blk =? (await node.blockStore.getBlock(cid)), err:
return failure(err)
proc streamOneBlock(): Future[void] {.async.} =
try:
await stream.pushData(blk.data)
except CatchableError as exc:
trace "Unable to send block", cid, exc = exc.msg
discard
finally:
await stream.pushEof()
asyncSpawn streamOneBlock()
LPStream(stream).success()
proc store*(
self: CodexNodeRef,
stream: LPStream,
blockSize = DefaultBlockSize): Future[?!Cid] {.async.} =
## Save stream contents as dataset with given blockSize
## to nodes's BlockStore, and return Cid of its manifest
##
trace "Storing data"
without var blockManifest =? Manifest.new(blockSize = blockSize):
return failure("Unable to create Block Set")
# Manifest and chunker should use the same blockSize
let chunker = LPStreamChunker.new(stream, chunkSize = blockSize)
try:
while (
let chunk = await chunker.getBytes();
chunk.len > 0):
trace "Got data from stream", len = chunk.len
without blk =? bt.Block.new(chunk):
return failure("Unable to init block from chunk!")
blockManifest.add(blk.cid)
if err =? (await self.blockStore.putBlock(blk)).errorOption:
trace "Unable to store block", cid = blk.cid, err = err.msg
return failure(&"Unable to store block {blk.cid}")
except CancelledError as exc:
raise exc
except CatchableError as exc:
return failure(exc.msg)
finally:
await stream.close()
# Generate manifest
blockManifest.originalBytes = NBytes(chunker.offset) # store the exact file size
without data =? blockManifest.encode():
return failure(
newException(CodexError, "Could not generate dataset manifest!"))
# Store as a dag-pb block
without manifest =? bt.Block.new(data = data, codec = DagPBCodec):
trace "Unable to init block from manifest data!"
return failure("Unable to init block from manifest data!")
if isErr (await self.blockStore.putBlock(manifest)):
trace "Unable to store manifest", cid = manifest.cid
return failure("Unable to store manifest " & $manifest.cid)
without cid =? blockManifest.cid, error:
trace "Unable to generate manifest Cid!", exc = error.msg
return failure(error.msg)
trace "Stored data", manifestCid = manifest.cid,
contentCid = cid,
blocks = blockManifest.len,
size=blockManifest.originalBytes
# Announce manifest
await self.discovery.provide(manifest.cid)
return manifest.cid.success
proc requestStorage*(
self: CodexNodeRef,
cid: Cid,
duration: UInt256,
proofProbability: UInt256,
nodes: uint,
tolerance: uint,
reward: UInt256,
collateral: UInt256,
expiry = UInt256.none): Future[?!PurchaseId] {.async.} =
## Initiate a request for storage sequence, this might
## be a multistep procedure.
##
## Roughly the flow is as follows:
## - Get the original cid from the store (should have already been uploaded)
## - Erasure code it according to the nodes and tolerance parameters
## - Run the PoR setup on the erasure dataset
## - Call into the marketplace and purchasing contracts
##
trace "Received a request for storage!", cid, duration, nodes, tolerance, reward, proofProbability, collateral, expiry
without contracts =? self.contracts.client:
trace "Purchasing not available"
return failure "Purchasing not available"
without manifest =? await self.fetchManifest(cid), error:
trace "Unable to fetch manifest for cid", cid
raise error
# Erasure code the dataset according to provided parameters
without encoded =? (await self.erasure.encode(manifest, nodes.int, tolerance.int)), error:
trace "Unable to erasure code dataset", cid
return failure(error)
without encodedData =? encoded.encode(), error:
trace "Unable to encode protected manifest"
return failure(error)
without encodedBlk =? bt.Block.new(data = encodedData, codec = DagPBCodec), error:
trace "Unable to create block from encoded manifest"
return failure(error)
if isErr (await self.blockStore.putBlock(encodedBlk)):
trace "Unable to store encoded manifest block", cid = encodedBlk.cid
return failure("Unable to store encoded manifest block")
let request = StorageRequest(
ask: StorageAsk(
slots: nodes + tolerance,
# TODO: Specify slot-specific size (as below) once dispersal is
# implemented. The current implementation downloads the entire dataset, so
# it is currently set to be the size of the entire dataset. This is
# because the slotSize is used to determine the amount of bytes to reserve
# in a Reservations
# TODO: slotSize: (encoded.blockSize.int * encoded.steps).u256,
slotSize: (encoded.blockSize.int * encoded.blocks.len).u256,
duration: duration,
proofProbability: proofProbability,
reward: reward,
collateral: collateral,
maxSlotLoss: tolerance
),
content: StorageContent(
cid: $encodedBlk.cid,
erasure: StorageErasure(
totalChunks: encoded.len.uint64,
),
por: StoragePoR(
u: @[], # TODO: PoR setup
publicKey: @[], # TODO: PoR setup
name: @[] # TODO: PoR setup
)
),
expiry: expiry |? 0.u256
)
let purchase = await contracts.purchasing.purchase(request)
return success purchase.id
proc new*(
T: type CodexNodeRef,
switch: Switch,
store: BlockStore,
engine: BlockExcEngine,
erasure: Erasure,
discovery: Discovery,
contracts = Contracts.default): CodexNodeRef =
## Create new instance of a Codex node, call `start` to run it
##
CodexNodeRef(
switch: switch,
blockStore: store,
engine: engine,
erasure: erasure,
discovery: discovery,
contracts: contracts)
proc start*(node: CodexNodeRef) {.async.} =
if not node.engine.isNil:
await node.engine.start()
if not node.erasure.isNil:
await node.erasure.start()
if not node.discovery.isNil:
await node.discovery.start()
if hostContracts =? node.contracts.host:
# TODO: remove Sales callbacks, pass BlockStore and StorageProofs instead
hostContracts.sales.onStore = proc(request: StorageRequest,
slot: UInt256,
onBatch: BatchProc): Future[?!void] {.async.} =
## store data in local storage
##
without cid =? Cid.init(request.content.cid):
trace "Unable to parse Cid", cid
let error = newException(CodexError, "Unable to parse Cid")
return failure(error)
without manifest =? await node.fetchManifest(cid), error:
trace "Unable to fetch manifest for cid", cid
return failure(error)
trace "Fetching block for manifest", cid
# TODO: This will probably require a call to `getBlock` either way,
# since fetching of blocks will have to be selective according
# to a combination of parameters, such as node slot position
# and dataset geometry
if fetchErr =? (await node.fetchBatched(manifest, onBatch = onBatch)).errorOption:
let error = newException(CodexError, "Unable to retrieve blocks")
error.parent = fetchErr
return failure(error)
return success()
hostContracts.sales.onClear = proc(request: StorageRequest,
slotIndex: UInt256) =
# TODO: remove data from local storage
discard
hostContracts.sales.onProve = proc(slot: Slot): Future[seq[byte]] {.async.} =
# TODO: generate proof
return @[42'u8]
try:
await hostContracts.start()
except CatchableError as error:
error "Unable to start host contract interactions: ", error=error.msg
node.contracts.host = HostInteractions.none
if clientContracts =? node.contracts.client:
try:
await clientContracts.start()
except CatchableError as error:
error "Unable to start client contract interactions: ", error=error.msg
node.contracts.client = ClientInteractions.none
if validatorContracts =? node.contracts.validator:
try:
await validatorContracts.start()
except CatchableError as error:
error "Unable to start validator contract interactions: ", error=error.msg
node.contracts.validator = ValidatorInteractions.none
node.networkId = node.switch.peerInfo.peerId
notice "Started codex node", id = $node.networkId, addrs = node.switch.peerInfo.addrs
proc stop*(node: CodexNodeRef) {.async.} =
trace "Stopping node"
if not node.engine.isNil:
await node.engine.stop()
if not node.erasure.isNil:
await node.erasure.stop()
if not node.discovery.isNil:
await node.discovery.stop()
if clientContracts =? node.contracts.client:
await clientContracts.stop()
if hostContracts =? node.contracts.host:
await hostContracts.stop()
if validatorContracts =? node.contracts.validator:
await validatorContracts.stop()
if not node.blockStore.isNil:
await node.blockStore.close