storj/satellite/overlay/service.go

713 lines
28 KiB
Go
Raw Normal View History

2019-01-24 20:15:10 +00:00
// Copyright (C) 2019 Storj Labs, Inc.
2018-04-18 17:55:28 +01:00
// See LICENSE for copying information.
package overlay
2018-04-18 16:34:15 +01:00
import (
"context"
"errors"
"net"
"time"
2018-04-18 16:34:15 +01:00
"github.com/zeebo/errs"
"go.uber.org/zap"
2018-11-16 16:31:14 +00:00
"storj.io/common/pb"
"storj.io/common/storj"
"storj.io/common/storj/location"
"storj.io/storj/satellite/geoip"
"storj.io/storj/satellite/metabase"
2018-04-18 16:34:15 +01:00
)
// ErrEmptyNode is returned when the nodeID is empty.
2018-12-17 18:47:26 +00:00
var ErrEmptyNode = errs.New("empty node ID")
// ErrNodeNotFound is returned if a node does not exist in database.
var ErrNodeNotFound = errs.Class("node not found")
// ErrNodeOffline is returned if a nodes is offline.
var ErrNodeOffline = errs.Class("node is offline")
// ErrNodeDisqualified is returned if a nodes is disqualified.
var ErrNodeDisqualified = errs.Class("node is disqualified")
// ErrNodeFinishedGE is returned if a node has finished graceful exit.
var ErrNodeFinishedGE = errs.Class("node finished graceful exit")
// ErrNotEnoughNodes is when selecting nodes failed with the given parameters.
var ErrNotEnoughNodes = errs.Class("not enough nodes")
// DB implements the database for overlay.Service.
2019-09-10 14:24:16 +01:00
//
// architecture: Database
type DB interface {
// GetOnlineNodesForGetDelete returns a map of nodes for the supplied nodeIDs
GetOnlineNodesForGetDelete(ctx context.Context, nodeIDs []storj.NodeID, onlineWindow time.Duration) (map[storj.NodeID]*SelectedNode, error)
// GetOnlineNodesForAuditRepair returns a map of nodes for the supplied nodeIDs.
// The return value contains necessary information to create orders as well as nodes'
// current reputation status.
GetOnlineNodesForAuditRepair(ctx context.Context, nodeIDs []storj.NodeID, onlineWindow time.Duration) (map[storj.NodeID]*NodeReputation, error)
// SelectStorageNodes looks up nodes based on criteria
SelectStorageNodes(ctx context.Context, totalNeededNodes, newNodeCount int, criteria *NodeCriteria) ([]*SelectedNode, error)
// SelectAllStorageNodesUpload returns all nodes that qualify to store data, organized as reputable nodes and new nodes
SelectAllStorageNodesUpload(ctx context.Context, selectionCfg NodeSelectionConfig) (reputable, new []*SelectedNode, err error)
// SelectAllStorageNodesDownload returns a nodes that are ready for downloading
SelectAllStorageNodesDownload(ctx context.Context, onlineWindow time.Duration, asOf AsOfSystemTimeConfig) ([]*SelectedNode, error)
// Get looks up the node by nodeID
Get(ctx context.Context, nodeID storj.NodeID) (*NodeDossier, error)
// KnownOffline filters a set of nodes to offline nodes
KnownOffline(context.Context, *NodeCriteria, storj.NodeIDList) (storj.NodeIDList, error)
// KnownUnreliableOrOffline filters a set of nodes to unhealth or offlines node, independent of new
KnownUnreliableOrOffline(context.Context, *NodeCriteria, storj.NodeIDList) (storj.NodeIDList, error)
// KnownReliableInExcludedCountries filters healthy nodes that are in excluded countries.
KnownReliableInExcludedCountries(context.Context, *NodeCriteria, storj.NodeIDList) (storj.NodeIDList, error)
// KnownReliable filters a set of nodes to reliable (online and qualified) nodes.
KnownReliable(ctx context.Context, onlineWindow time.Duration, nodeIDs storj.NodeIDList) ([]*pb.Node, error)
// Reliable returns all nodes that are reliable
Reliable(context.Context, *NodeCriteria) (storj.NodeIDList, error)
// UpdateReputation updates the DB columns for all reputation fields in ReputationStatus.
UpdateReputation(ctx context.Context, id storj.NodeID, request ReputationUpdate) error
Add Version Information into KAD Network and SatelliteDB & Change Selection Process (#1648) * Initial Webserver Draft for Version Controlling * Rename type to avoid confusion * Move Function Calls into Version Package * Fix Linting and Language Typos * Fix Linting and Spelling Mistakes * Include Copyright * Include Copyright * Adjust Version-Control Server to return list of Versions * Linting * Improve Request Handling and Readability * Add Configuration File Option Add Systemd Service file * Add Logging to File * Smaller Changes * Add Semantic Versioning and refuses outdated Software from Startup (#1612) * implements internal Semantic Version library * adds version logging + reporting to process * Advance SemVer struct for easier handling * Add Accepted Version Store * Fix Function * Restructure * Type Conversion * Handle Version String properly * Add Note about array index * Set temporary Default Version * Add Copyright * Adding Version to Dashboard * Adding Version Info Log * Renaming and adding CheckerProcess * Iteration Sync * Iteration V2 * linting * made LogAndReportVersion a go routine * Refactor to Go Routine * Add Context to Go Routine and allow Operation if Lookup to Control Server fails * Handle Unmarshal properly * Linting * Relocate Version Checks * Relocating Version Check and specified default Version for now * Linting Error Prevention * Refuse Startup on outdated Version * Add Startup Check Function * Straighten Logging * Dont force Shutdown if --dev flag is set * Create full Service/Peer Structure for ControlServer * Linting * Straighting Naming * Finish VersionControl Service Layout * Improve Error Handling * Change Listening Address * Move Checker Function * Remove VersionControl Peer * Linting * Linting * Create VersionClient Service * Renaming * Add Version Client to Peer Definitions * Linting and Renaming * Linting * Remove Transport Checks for now * Move to Client Side Flag * Remove check * Linting * Transport Client Version Intro * Adding Version Client to Transport Client * Add missing parameter * Adding Version Check, to set Allowed = true * Set Default to true, testing * Restructuring Code * Uplink Changes * Add more proper Defaults * Renaming of Version struct * Dont pass Service use Pointer * Set Defaults for Versioning Checks * Put HTTP Server in go routine * Add Versioncontrol to Storj-Sim * Testplanet Fixes * Linting * Add Error Handling and new Server Struct * Move Lock slightly * Reduce Race Potentials * Remove unnecessary files * Linting * Add Proper Transport Handling * small fixes * add fence for allowed check * Add Startup Version Check and Service Naming * make errormessage private * Add Comments about VersionedClient * Linting * Remove Checks that refuse outgoing connections * Remove release cmd * Add Release Script * Linting * Update to use correct Values * Change Timestamp handling * Adding Protobuf changes back in * Adding SatelliteDB Changes and adding Storj Node Version to PB * Add Migration Table * Add Default Stats for Creation * Move to BigInt * Proper SQL Migration * Ensure minimum Version is passed to the node selection * Linting... * Remove VersionedClient and adjust smaller changes from prior merge * Linting * Fix PB Message Handling and Query for Node Selection * some future-proofing type changes Change-Id: I3cb5018dcccdbc9739fe004d859065992720caaf * fix a compiler error Change-Id: If66bb92d8b98e31cd618ecec9c6448ab9b037fa5 * Comment on Constant for Overlay * Remove NOT NULL and add epoch call as function * add versions to bootstrap and satellites Change-Id: I436944589ea5f21600cdd997742a84fe0b16e47b * Change Update Migration * Fix DB Migration * Increase Timeout temporarily, to see whats going on * Remove unnecessary const and vars Cleanup Function calls from deprecated NodeVersion struct * Updated Protopuf, removed depcreated Code from Inspector * Implement NodeVersion into InfoResponse * Regenerated locked.go * Linting * Fix Tests * Remove unnecessary constant * Update Function and Flag Description * Remove Empty Stat Creation * return properly with error * Remove unnecessary struct * simplify migration step * Update Inspector to return Version Info * Update local Endpoint Version Handling * Reset Travis Timeout * Add Default for CommitHash * single quotes
2019-04-10 07:04:24 +01:00
// UpdateNodeInfo updates node dossier with info requested from the node itself like node type, email, wallet, capacity, and version.
UpdateNodeInfo(ctx context.Context, node storj.NodeID, nodeInfo *InfoResponse) (stats *NodeDossier, err error)
// UpdateCheckIn updates a single storagenode's check-in stats.
UpdateCheckIn(ctx context.Context, node NodeCheckInInfo, timestamp time.Time, config NodeSelectionConfig) (err error)
// AllPieceCounts returns a map of node IDs to piece counts from the db.
AllPieceCounts(ctx context.Context) (pieceCounts map[storj.NodeID]int, err error)
// UpdatePieceCounts sets the piece count field for the given node IDs.
UpdatePieceCounts(ctx context.Context, pieceCounts map[storj.NodeID]int) (err error)
// UpdateExitStatus is used to update a node's graceful exit status.
UpdateExitStatus(ctx context.Context, request *ExitStatusRequest) (_ *NodeDossier, err error)
// GetExitingNodes returns nodes who have initiated a graceful exit, but have not completed it.
GetExitingNodes(ctx context.Context) (exitingNodes []*ExitStatus, err error)
// GetGracefulExitCompletedByTimeFrame returns nodes who have completed graceful exit within a time window (time window is around graceful exit completion).
GetGracefulExitCompletedByTimeFrame(ctx context.Context, begin, end time.Time) (exitedNodes storj.NodeIDList, err error)
// GetGracefulExitIncompleteByTimeFrame returns nodes who have initiated, but not completed graceful exit within a time window (time window is around graceful exit initiation).
GetGracefulExitIncompleteByTimeFrame(ctx context.Context, begin, end time.Time) (exitingNodes storj.NodeIDList, err error)
// GetExitStatus returns a node's graceful exit status.
GetExitStatus(ctx context.Context, nodeID storj.NodeID) (exitStatus *ExitStatus, err error)
// GetNodesNetwork returns the /24 subnet for each storage node, order is not guaranteed.
GetNodesNetwork(ctx context.Context, nodeIDs []storj.NodeID) (nodeNets []string, err error)
// DisqualifyNode disqualifies a storage node.
DisqualifyNode(ctx context.Context, nodeID storj.NodeID, disqualifiedAt time.Time, reason DisqualificationReason) (err error)
// DQNodesLastSeenBefore disqualifies a limited number of nodes where last_contact_success < cutoff except those already disqualified
// or gracefully exited or where last_contact_success = '0001-01-01 00:00:00+00'.
DQNodesLastSeenBefore(ctx context.Context, cutoff time.Time, limit int) (count int, err error)
// TestSuspendNodeUnknownAudit suspends a storage node for unknown audits.
TestSuspendNodeUnknownAudit(ctx context.Context, nodeID storj.NodeID, suspendedAt time.Time) (err error)
// TestUnsuspendNodeUnknownAudit unsuspends a storage node for unknown audits.
TestUnsuspendNodeUnknownAudit(ctx context.Context, nodeID storj.NodeID) (err error)
// TestVetNode directly sets a node's vetted_at timestamp to make testing easier
TestVetNode(ctx context.Context, nodeID storj.NodeID) (vettedTime *time.Time, err error)
// TestUnvetNode directly sets a node's vetted_at timestamp to null to make testing easier
TestUnvetNode(ctx context.Context, nodeID storj.NodeID) (err error)
// TestVetNode directly sets a node's offline_suspended timestamp to make testing easier
TestSuspendNodeOffline(ctx context.Context, nodeID storj.NodeID, suspendedAt time.Time) (err error)
// TestNodeCountryCode sets node country code.
TestNodeCountryCode(ctx context.Context, nodeID storj.NodeID, countryCode string) (err error)
// IterateAllNodes will call cb on all known nodes (used in restore trash contexts).
IterateAllNodes(context.Context, func(context.Context, *SelectedNode) error) error
// IterateAllNodes will call cb on all known nodes (used for invoice generation).
IterateAllNodeDossiers(context.Context, func(context.Context, *NodeDossier) error) error
}
// DisqualificationReason is disqualification reason enum type.
type DisqualificationReason int
const (
// DisqualificationReasonUnknown denotes undetermined disqualification reason.
DisqualificationReasonUnknown DisqualificationReason = 0
// DisqualificationReasonAuditFailure denotes disqualification due to audit score falling below threshold.
DisqualificationReasonAuditFailure DisqualificationReason = 1
// DisqualificationReasonSuspension denotes disqualification due to unknown audit failure after grace period for unknown audits
// has elapsed.
DisqualificationReasonSuspension DisqualificationReason = 2
// DisqualificationReasonNodeOffline denotes disqualification due to node's online score falling below threshold after tracking
// period has elapsed.
DisqualificationReasonNodeOffline DisqualificationReason = 3
)
// NodeCheckInInfo contains all the info that will be updated when a node checkins.
type NodeCheckInInfo struct {
NodeID storj.NodeID
Address *pb.NodeAddress
LastNet string
LastIPPort string
IsUp bool
Operator *pb.NodeOperator
Capacity *pb.NodeCapacity
Version *pb.NodeVersion
CountryCode location.CountryCode
}
// InfoResponse contains node dossier info requested from the storage node.
type InfoResponse struct {
Type pb.NodeType
Operator *pb.NodeOperator
Capacity *pb.NodeCapacity
Version *pb.NodeVersion
}
2019-03-23 08:06:11 +00:00
// FindStorageNodesRequest defines easy request parameters.
type FindStorageNodesRequest struct {
RequestedCount int
ExcludedIDs []storj.NodeID
MinimumVersion string // semver or empty
AsOfSystemInterval time.Duration // only used for CRDB queries
Placement storj.PlacementConstraint
2019-03-23 08:06:11 +00:00
}
// NodeCriteria are the requirements for selecting nodes.
2019-03-23 08:06:11 +00:00
type NodeCriteria struct {
FreeDisk int64
ExcludedIDs []storj.NodeID
ExcludedNetworks []string // the /24 subnet IPv4 or /64 subnet IPv6 for nodes
MinimumVersion string // semver or empty
OnlineWindow time.Duration
DistinctIP bool
AsOfSystemInterval time.Duration // only used for CRDB queries
ExcludedCountries []string
2019-03-23 08:06:11 +00:00
}
// ReputationStatus indicates current reputation status for a node.
type ReputationStatus struct {
Disqualified *time.Time
DisqualificationReason *DisqualificationReason
UnknownAuditSuspended *time.Time
OfflineSuspended *time.Time
VettedAt *time.Time
}
// ReputationUpdate contains reputation update data for a node.
type ReputationUpdate struct {
Disqualified *time.Time
DisqualificationReason DisqualificationReason
UnknownAuditSuspended *time.Time
OfflineSuspended *time.Time
VettedAt *time.Time
}
// ExitStatus is used for reading graceful exit status.
type ExitStatus struct {
NodeID storj.NodeID
ExitInitiatedAt *time.Time
ExitLoopCompletedAt *time.Time
ExitFinishedAt *time.Time
ExitSuccess bool
}
// ExitStatusRequest is used to update a node's graceful exit status.
type ExitStatusRequest struct {
NodeID storj.NodeID
ExitInitiatedAt time.Time
ExitLoopCompletedAt time.Time
ExitFinishedAt time.Time
ExitSuccess bool
}
// NodeDossier is the complete info that the satellite tracks for a storage node.
type NodeDossier struct {
pb.Node
Type pb.NodeType
Operator pb.NodeOperator
Capacity pb.NodeCapacity
Reputation NodeStats
Version pb.NodeVersion
Contained bool
Disqualified *time.Time
DisqualificationReason *DisqualificationReason
UnknownAuditSuspended *time.Time
OfflineSuspended *time.Time
OfflineUnderReview *time.Time
PieceCount int64
ExitStatus ExitStatus
CreatedAt time.Time
LastNet string
LastIPPort string
CountryCode location.CountryCode
}
// NodeStats contains statistics about a node.
type NodeStats struct {
Latency90 int64
LastContactSuccess time.Time
LastContactFailure time.Time
OfflineUnderReview *time.Time
Status ReputationStatus
}
// NodeLastContact contains the ID, address, and timestamp.
type NodeLastContact struct {
URL storj.NodeURL
LastIPPort string
LastContactSuccess time.Time
LastContactFailure time.Time
}
// SelectedNode is used as a result for creating orders limits.
type SelectedNode struct {
ID storj.NodeID
Address *pb.NodeAddress
LastNet string
LastIPPort string
CountryCode location.CountryCode
}
// NodeReputation is used as a result for creating orders limits for audits.
type NodeReputation struct {
ID storj.NodeID
Address *pb.NodeAddress
LastNet string
LastIPPort string
Reputation ReputationStatus
}
// Clone returns a deep clone of the selected node.
func (node *SelectedNode) Clone() *SelectedNode {
return &SelectedNode{
ID: node.ID,
Address: &pb.NodeAddress{
Transport: node.Address.Transport,
Address: node.Address.Address,
},
LastNet: node.LastNet,
LastIPPort: node.LastIPPort,
}
}
// Service is used to store and handle node information.
2019-09-10 14:24:16 +01:00
//
// architecture: Service
type Service struct {
log *zap.Logger
db DB
config Config
GeoIP geoip.IPToCountry
UploadSelectionCache *UploadSelectionCache
DownloadSelectionCache *DownloadSelectionCache
2018-04-18 16:34:15 +01:00
}
// NewService returns a new Service.
func NewService(log *zap.Logger, db DB, config Config) (*Service, error) {
err := config.Node.AsOfSystemTime.isValid()
if err != nil {
return nil, err
}
var geoIP geoip.IPToCountry = geoip.NewMockIPToCountry(config.GeoIP.MockCountries)
if config.GeoIP.DB != "" {
geoIP, err = geoip.OpenMaxmindDB(config.GeoIP.DB)
if err != nil {
return nil, Error.Wrap(err)
}
}
return &Service{
log: log,
db: db,
config: config,
GeoIP: geoIP,
UploadSelectionCache: NewUploadSelectionCache(log, db,
config.NodeSelectionCache.Staleness, config.Node,
),
DownloadSelectionCache: NewDownloadSelectionCache(log, db, DownloadSelectionCacheConfig{
Staleness: config.NodeSelectionCache.Staleness,
OnlineWindow: config.Node.OnlineWindow,
AsOfSystemTime: config.Node.AsOfSystemTime,
}),
}, nil
}
// Close closes resources.
func (service *Service) Close() error {
return service.GeoIP.Close()
}
Satellite Peer (#1034) * add satellite peer * Add overlay * reorganize kademlia * add RunRefresh * add refresh to storagenode.Peer * add discovery * add agreements and metainfo * rename * add datarepair checker * add repair * add todo notes for audit * add testing interface * add into testplanet * fixes * fix compilation errors * fix compilation errors * make testplanet run * remove audit refrences * ensure that audit tests run * dev * checker tests compilable * fix discovery * fix compilation * fix * fix * dev * fix * disable auth * fixes * revert go.mod/sum * fix linter errors * fix * fix copyright * Add address param for SN dashboard (#1076) * Rename storj-sdk to storj-sim (#1078) * Storagenode logs and config improvements (#1075) * Add more info to SN logs * remove config-dir from user config * add output where config was stored * add message for successful connection * fix linter * remove storage.path from user config * resolve config path * move success message to info * log improvements * Remove captplanet (#1070) * pkg/server: include production cert (#1082) Change-Id: Ie8e6fe78550be83c3bd797db7a1e58d37c684792 * Generate Payments Report (#1079) * memory.Size: autoformat sizes based on value entropy (#1081) * Jj/bytes (#1085) * run tally and rollup * sets dev default tally and rollup intervals * nonessential storj-sim edits (#1086) * Closing context doesn't stop storage node (#1084) * Print when cancelled * Close properly * Don't log nil * Don't print error when closing dashboard * Fix panic in inspector if ping fails (#1088) * Consolidate identity management to identity cli commands (#1083) * Consolidate identity management: Move identity cretaion/signing out of storagenode setup command. * fixes * linters * Consolidate identity management: Move identity cretaion/signing out of storagenode setup command. * fixes * sava backups before saving signed certs * add "-prebuilt-test-cmds" test flag * linters * prepare cli tests for travis * linter fixes * more fixes * linter gods * sp/sdk/sim * remove ca.difficulty * remove unused difficulty * return setup to its rightful place * wip travis * Revert "wip travis" This reverts commit 56834849dcf066d3cc0a4f139033fc3f6d7188ca. * typo in travis.yaml * remove tests * remove more * make it only create one identity at a time for consistency * add config-dir for consitency * add identity creation to storj-sim * add flags * simplify * fix nolint and compile * prevent overwrite and pass difficulty, concurrency, and parent creds * goimports
2019-01-18 13:54:08 +00:00
// Get looks up the provided nodeID from the overlay.
func (service *Service) Get(ctx context.Context, nodeID storj.NodeID) (_ *NodeDossier, err error) {
2019-03-23 08:06:11 +00:00
defer mon.Task()(&ctx)(&err)
2018-12-17 18:47:26 +00:00
if nodeID.IsZero() {
return nil, ErrEmptyNode
}
return service.db.Get(ctx, nodeID)
2018-04-18 16:34:15 +01:00
}
// GetOnlineNodesForGetDelete returns a map of nodes for the supplied nodeIDs.
func (service *Service) GetOnlineNodesForGetDelete(ctx context.Context, nodeIDs []storj.NodeID) (_ map[storj.NodeID]*SelectedNode, err error) {
defer mon.Task()(&ctx)(&err)
return service.db.GetOnlineNodesForGetDelete(ctx, nodeIDs, service.config.Node.OnlineWindow)
}
// GetOnlineNodesForAuditRepair returns a map of nodes for the supplied nodeIDs.
func (service *Service) GetOnlineNodesForAuditRepair(ctx context.Context, nodeIDs []storj.NodeID) (_ map[storj.NodeID]*NodeReputation, err error) {
defer mon.Task()(&ctx)(&err)
return service.db.GetOnlineNodesForAuditRepair(ctx, nodeIDs, service.config.Node.OnlineWindow)
}
// GetNodeIPs returns a map of node ip:port for the supplied nodeIDs.
func (service *Service) GetNodeIPs(ctx context.Context, nodeIDs []storj.NodeID) (_ map[storj.NodeID]string, err error) {
defer mon.Task()(&ctx)(&err)
return service.DownloadSelectionCache.GetNodeIPs(ctx, nodeIDs)
}
// IsOnline checks if a node is 'online' based on the collected statistics.
func (service *Service) IsOnline(node *NodeDossier) bool {
return time.Since(node.Reputation.LastContactSuccess) < service.config.Node.OnlineWindow
}
// FindStorageNodesForGracefulExit searches the overlay network for nodes that meet the provided requirements for graceful-exit requests.
func (service *Service) FindStorageNodesForGracefulExit(ctx context.Context, req FindStorageNodesRequest) (_ []*SelectedNode, err error) {
defer mon.Task()(&ctx)(&err)
return service.UploadSelectionCache.GetNodes(ctx, req)
}
// FindStorageNodesForUpload searches the overlay network for nodes that meet the provided requirements for upload.
//
// When enabled it uses the cache to select nodes.
// When the node selection from the cache fails, it falls back to the old implementation.
func (service *Service) FindStorageNodesForUpload(ctx context.Context, req FindStorageNodesRequest) (_ []*SelectedNode, err error) {
defer mon.Task()(&ctx)(&err)
if service.config.Node.AsOfSystemTime.Enabled && service.config.Node.AsOfSystemTime.DefaultInterval < 0 {
req.AsOfSystemInterval = service.config.Node.AsOfSystemTime.DefaultInterval
}
// TODO excluding country codes on upload if cache is disabled is not implemented
if service.config.NodeSelectionCache.Disabled {
return service.FindStorageNodesWithPreferences(ctx, req, &service.config.Node)
}
selectedNodes, err := service.UploadSelectionCache.GetNodes(ctx, req)
if err != nil {
return selectedNodes, err
}
if len(selectedNodes) < req.RequestedCount {
excludedIDs := make([]string, 0)
for _, e := range req.ExcludedIDs {
excludedIDs = append(excludedIDs, e.String())
}
service.log.Warn("Not enough nodes are available from Node Cache",
zap.String("minVersion", req.MinimumVersion),
zap.Strings("excludedIDs", excludedIDs),
zap.Duration("asOfSystemInterval", req.AsOfSystemInterval),
zap.Int("requested", req.RequestedCount),
zap.Int("available", len(selectedNodes)),
zap.Uint16("placement", uint16(req.Placement)))
}
return selectedNodes, err
}
// FindStorageNodesWithPreferences searches the overlay network for nodes that meet the provided criteria.
//
// This does not use a cache.
func (service *Service) FindStorageNodesWithPreferences(ctx context.Context, req FindStorageNodesRequest, preferences *NodeSelectionConfig) (nodes []*SelectedNode, err error) {
2019-03-23 08:06:11 +00:00
defer mon.Task()(&ctx)(&err)
// TODO: add sanity limits to requested node count
// TODO: add sanity limits to excluded nodes
totalNeededNodes := req.RequestedCount
excludedIDs := req.ExcludedIDs
// if distinctIP is enabled, keep track of the network
// to make sure we only select nodes from different networks
var excludedNetworks []string
if preferences.DistinctIP && len(excludedIDs) > 0 {
excludedNetworks, err = service.db.GetNodesNetwork(ctx, excludedIDs)
if err != nil {
return nil, Error.Wrap(err)
}
}
newNodeCount := 0
if preferences.NewNodeFraction > 0 {
newNodeCount = int(float64(totalNeededNodes) * preferences.NewNodeFraction)
}
criteria := NodeCriteria{
FreeDisk: preferences.MinimumDiskSpace.Int64(),
ExcludedIDs: excludedIDs,
ExcludedNetworks: excludedNetworks,
MinimumVersion: preferences.MinimumVersion,
OnlineWindow: preferences.OnlineWindow,
DistinctIP: preferences.DistinctIP,
AsOfSystemInterval: req.AsOfSystemInterval,
}
nodes, err = service.db.SelectStorageNodes(ctx, totalNeededNodes, newNodeCount, &criteria)
if err != nil {
return nil, Error.Wrap(err)
}
if len(nodes) < totalNeededNodes {
return nodes, ErrNotEnoughNodes.New("requested %d found %d; %+v ", totalNeededNodes, len(nodes), criteria)
}
return nodes, nil
}
// KnownOffline filters a set of nodes to offline nodes.
func (service *Service) KnownOffline(ctx context.Context, nodeIds storj.NodeIDList) (offlineNodes storj.NodeIDList, err error) {
defer mon.Task()(&ctx)(&err)
criteria := &NodeCriteria{
OnlineWindow: service.config.Node.OnlineWindow,
}
return service.db.KnownOffline(ctx, criteria, nodeIds)
}
// KnownUnreliableOrOffline filters a set of nodes to unhealth or offlines node, independent of new.
func (service *Service) KnownUnreliableOrOffline(ctx context.Context, nodeIds storj.NodeIDList) (badNodes storj.NodeIDList, err error) {
2019-03-23 08:06:11 +00:00
defer mon.Task()(&ctx)(&err)
criteria := &NodeCriteria{
OnlineWindow: service.config.Node.OnlineWindow,
}
return service.db.KnownUnreliableOrOffline(ctx, criteria, nodeIds)
}
// KnownReliableInExcludedCountries filters healthy nodes that are in excluded countries.
func (service *Service) KnownReliableInExcludedCountries(ctx context.Context, nodeIds storj.NodeIDList) (reliableInExcluded storj.NodeIDList, err error) {
defer mon.Task()(&ctx)(&err)
criteria := &NodeCriteria{
OnlineWindow: service.config.Node.OnlineWindow,
ExcludedCountries: service.config.RepairExcludedCountryCodes,
}
return service.db.KnownReliableInExcludedCountries(ctx, criteria, nodeIds)
}
// KnownReliable filters a set of nodes to reliable (online and qualified) nodes.
func (service *Service) KnownReliable(ctx context.Context, nodeIDs storj.NodeIDList) (nodes []*pb.Node, err error) {
defer mon.Task()(&ctx)(&err)
return service.db.KnownReliable(ctx, service.config.Node.OnlineWindow, nodeIDs)
}
// Reliable filters a set of nodes that are reliable, independent of new.
func (service *Service) Reliable(ctx context.Context) (nodes storj.NodeIDList, err error) {
defer mon.Task()(&ctx)(&err)
criteria := &NodeCriteria{
OnlineWindow: service.config.Node.OnlineWindow,
}
criteria.ExcludedCountries = service.config.RepairExcludedCountryCodes
return service.db.Reliable(ctx, criteria)
}
// UpdateReputation updates the DB columns for any of the reputation fields.
func (service *Service) UpdateReputation(ctx context.Context, id storj.NodeID, request ReputationUpdate) (err error) {
defer mon.Task()(&ctx)(&err)
return service.db.UpdateReputation(ctx, id, request)
}
Add Version Information into KAD Network and SatelliteDB & Change Selection Process (#1648) * Initial Webserver Draft for Version Controlling * Rename type to avoid confusion * Move Function Calls into Version Package * Fix Linting and Language Typos * Fix Linting and Spelling Mistakes * Include Copyright * Include Copyright * Adjust Version-Control Server to return list of Versions * Linting * Improve Request Handling and Readability * Add Configuration File Option Add Systemd Service file * Add Logging to File * Smaller Changes * Add Semantic Versioning and refuses outdated Software from Startup (#1612) * implements internal Semantic Version library * adds version logging + reporting to process * Advance SemVer struct for easier handling * Add Accepted Version Store * Fix Function * Restructure * Type Conversion * Handle Version String properly * Add Note about array index * Set temporary Default Version * Add Copyright * Adding Version to Dashboard * Adding Version Info Log * Renaming and adding CheckerProcess * Iteration Sync * Iteration V2 * linting * made LogAndReportVersion a go routine * Refactor to Go Routine * Add Context to Go Routine and allow Operation if Lookup to Control Server fails * Handle Unmarshal properly * Linting * Relocate Version Checks * Relocating Version Check and specified default Version for now * Linting Error Prevention * Refuse Startup on outdated Version * Add Startup Check Function * Straighten Logging * Dont force Shutdown if --dev flag is set * Create full Service/Peer Structure for ControlServer * Linting * Straighting Naming * Finish VersionControl Service Layout * Improve Error Handling * Change Listening Address * Move Checker Function * Remove VersionControl Peer * Linting * Linting * Create VersionClient Service * Renaming * Add Version Client to Peer Definitions * Linting and Renaming * Linting * Remove Transport Checks for now * Move to Client Side Flag * Remove check * Linting * Transport Client Version Intro * Adding Version Client to Transport Client * Add missing parameter * Adding Version Check, to set Allowed = true * Set Default to true, testing * Restructuring Code * Uplink Changes * Add more proper Defaults * Renaming of Version struct * Dont pass Service use Pointer * Set Defaults for Versioning Checks * Put HTTP Server in go routine * Add Versioncontrol to Storj-Sim * Testplanet Fixes * Linting * Add Error Handling and new Server Struct * Move Lock slightly * Reduce Race Potentials * Remove unnecessary files * Linting * Add Proper Transport Handling * small fixes * add fence for allowed check * Add Startup Version Check and Service Naming * make errormessage private * Add Comments about VersionedClient * Linting * Remove Checks that refuse outgoing connections * Remove release cmd * Add Release Script * Linting * Update to use correct Values * Change Timestamp handling * Adding Protobuf changes back in * Adding SatelliteDB Changes and adding Storj Node Version to PB * Add Migration Table * Add Default Stats for Creation * Move to BigInt * Proper SQL Migration * Ensure minimum Version is passed to the node selection * Linting... * Remove VersionedClient and adjust smaller changes from prior merge * Linting * Fix PB Message Handling and Query for Node Selection * some future-proofing type changes Change-Id: I3cb5018dcccdbc9739fe004d859065992720caaf * fix a compiler error Change-Id: If66bb92d8b98e31cd618ecec9c6448ab9b037fa5 * Comment on Constant for Overlay * Remove NOT NULL and add epoch call as function * add versions to bootstrap and satellites Change-Id: I436944589ea5f21600cdd997742a84fe0b16e47b * Change Update Migration * Fix DB Migration * Increase Timeout temporarily, to see whats going on * Remove unnecessary const and vars Cleanup Function calls from deprecated NodeVersion struct * Updated Protopuf, removed depcreated Code from Inspector * Implement NodeVersion into InfoResponse * Regenerated locked.go * Linting * Fix Tests * Remove unnecessary constant * Update Function and Flag Description * Remove Empty Stat Creation * return properly with error * Remove unnecessary struct * simplify migration step * Update Inspector to return Version Info * Update local Endpoint Version Handling * Reset Travis Timeout * Add Default for CommitHash * single quotes
2019-04-10 07:04:24 +01:00
// UpdateNodeInfo updates node dossier with info requested from the node itself like node type, email, wallet, capacity, and version.
func (service *Service) UpdateNodeInfo(ctx context.Context, node storj.NodeID, nodeInfo *InfoResponse) (stats *NodeDossier, err error) {
defer mon.Task()(&ctx)(&err)
return service.db.UpdateNodeInfo(ctx, node, nodeInfo)
}
// UpdateCheckIn updates a single storagenode's check-in info if needed.
/*
The check-in info is updated in the database if:
(1) there is no previous entry;
(2) it has been too long since the last known entry; or
(3) the node hostname, IP address, port, wallet, sw version, or disk capacity
has changed.
Note that there can be a race between acquiring the previous entry and
performing the update, so if two updates happen at about the same time it is
not defined which one will end up in the database.
*/
func (service *Service) UpdateCheckIn(ctx context.Context, node NodeCheckInInfo, timestamp time.Time) (err error) {
defer mon.Task()(&ctx)(&err)
failureMeter := mon.Meter("geofencing_lookup_failed")
oldInfo, err := service.Get(ctx, node.NodeID)
if err != nil && !ErrNodeNotFound.Has(err) {
return Error.New("failed to get node info from DB")
}
if oldInfo == nil {
node.CountryCode, err = service.GeoIP.LookupISOCountryCode(node.LastIPPort)
if err != nil {
failureMeter.Mark(1)
service.log.Debug("failed to resolve country code for node",
zap.String("node address", node.Address.Address),
zap.Stringer("Node ID", node.NodeID),
zap.Error(err))
}
return service.db.UpdateCheckIn(ctx, node, timestamp, service.config.Node)
}
lastUp, lastDown := oldInfo.Reputation.LastContactSuccess, oldInfo.Reputation.LastContactFailure
lastContact := lastUp
if lastContact.Before(lastDown) {
lastContact = lastDown
}
dbStale := lastContact.Add(service.config.NodeCheckInWaitPeriod).Before(timestamp) ||
(node.IsUp && lastUp.Before(lastDown)) || (!node.IsUp && lastDown.Before(lastUp))
addrChanged := ((node.Address == nil) != (oldInfo.Address == nil)) ||
(oldInfo.Address != nil && node.Address != nil && oldInfo.Address.Address != node.Address.Address)
walletChanged := (node.Operator == nil && oldInfo.Operator.Wallet != "") ||
(node.Operator != nil && oldInfo.Operator.Wallet != node.Operator.Wallet)
verChanged := (node.Version == nil && oldInfo.Version.Version != "") ||
(node.Version != nil && oldInfo.Version.Version != node.Version.Version)
spaceChanged := (node.Capacity == nil && oldInfo.Capacity.FreeDisk != 0) ||
(node.Capacity != nil && node.Capacity.FreeDisk != oldInfo.Capacity.FreeDisk)
if oldInfo.CountryCode == location.CountryCode(0) || oldInfo.LastIPPort != node.LastIPPort {
node.CountryCode, err = service.GeoIP.LookupISOCountryCode(node.LastIPPort)
if err != nil {
failureMeter.Mark(1)
service.log.Debug("failed to resolve country code for node",
zap.String("node address", node.Address.Address),
zap.Stringer("Node ID", node.NodeID),
zap.Error(err))
}
} else {
node.CountryCode = oldInfo.CountryCode
}
if dbStale || addrChanged || walletChanged || verChanged || spaceChanged ||
oldInfo.LastNet != node.LastNet || oldInfo.LastIPPort != node.LastIPPort ||
oldInfo.CountryCode != node.CountryCode {
return service.db.UpdateCheckIn(ctx, node, timestamp, service.config.Node)
}
service.log.Debug("ignoring unnecessary check-in",
zap.String("node address", node.Address.Address),
zap.Stringer("Node ID", node.NodeID))
mon.Event("unnecessary_node_check_in")
return nil
}
// GetMissingPieces returns the list of offline nodes and the corresponding pieces.
func (service *Service) GetMissingPieces(ctx context.Context, pieces metabase.Pieces) (missingPieces []uint16, err error) {
defer mon.Task()(&ctx)(&err)
var nodeIDs storj.NodeIDList
for _, p := range pieces {
nodeIDs = append(nodeIDs, p.StorageNode)
}
badNodeIDs, err := service.KnownUnreliableOrOffline(ctx, nodeIDs)
if err != nil {
return nil, Error.New("error getting nodes %s", err)
}
for _, p := range pieces {
for _, nodeID := range badNodeIDs {
if nodeID == p.StorageNode {
missingPieces = append(missingPieces, p.Number)
}
}
}
return missingPieces, nil
}
// GetReliablePiecesInExcludedCountries returns the list of pieces held by nodes located in excluded countries.
func (service *Service) GetReliablePiecesInExcludedCountries(ctx context.Context, pieces metabase.Pieces) (piecesInExcluded []uint16, err error) {
defer mon.Task()(&ctx)(&err)
var nodeIDs storj.NodeIDList
for _, p := range pieces {
nodeIDs = append(nodeIDs, p.StorageNode)
}
inExcluded, err := service.KnownReliableInExcludedCountries(ctx, nodeIDs)
if err != nil {
return nil, Error.New("error getting nodes %s", err)
}
for _, p := range pieces {
for _, nodeID := range inExcluded {
if nodeID == p.StorageNode {
piecesInExcluded = append(piecesInExcluded, p.Number)
}
}
}
return piecesInExcluded, nil
}
// DisqualifyNode disqualifies a storage node.
func (service *Service) DisqualifyNode(ctx context.Context, nodeID storj.NodeID, reason DisqualificationReason) (err error) {
defer mon.Task()(&ctx)(&err)
return service.db.DisqualifyNode(ctx, nodeID, time.Now().UTC(), reason)
}
// ResolveIPAndNetwork resolves the target address and determines its IP and /24 subnet IPv4 or /64 subnet IPv6.
func ResolveIPAndNetwork(ctx context.Context, target string) (ipPort, network string, err error) {
2019-06-24 16:33:18 +01:00
defer mon.Task()(&ctx)(&err)
host, port, err := net.SplitHostPort(target)
if err != nil {
return "", "", err
}
ipAddr, err := net.ResolveIPAddr("ip", host)
if err != nil {
return "", "", err
}
2019-06-24 16:33:18 +01:00
// If addr can be converted to 4byte notation, it is an IPv4 address, else its an IPv6 address
if ipv4 := ipAddr.IP.To4(); ipv4 != nil {
// Filter all IPv4 Addresses into /24 Subnet's
2019-06-24 16:33:18 +01:00
mask := net.CIDRMask(24, 32)
return net.JoinHostPort(ipAddr.String(), port), ipv4.Mask(mask).String(), nil
2019-06-24 16:33:18 +01:00
}
if ipv6 := ipAddr.IP.To16(); ipv6 != nil {
// Filter all IPv6 Addresses into /64 Subnet's
2019-06-24 16:33:18 +01:00
mask := net.CIDRMask(64, 128)
return net.JoinHostPort(ipAddr.String(), port), ipv6.Mask(mask).String(), nil
2019-06-24 16:33:18 +01:00
}
return "", "", errors.New("unable to get network for address " + ipAddr.String())
}
// TestVetNode directly sets a node's vetted_at timestamp to make testing easier.
func (service *Service) TestVetNode(ctx context.Context, nodeID storj.NodeID) (vettedTime *time.Time, err error) {
vettedTime, err = service.db.TestVetNode(ctx, nodeID)
service.log.Warn("node vetted", zap.Stringer("node ID", nodeID), zap.Stringer("vetted time", vettedTime))
if err != nil {
service.log.Warn("error vetting node", zap.Stringer("node ID", nodeID))
return nil, err
}
err = service.UploadSelectionCache.Refresh(ctx)
service.log.Warn("nodecache refresh err", zap.Error(err))
return vettedTime, err
}
// TestUnvetNode directly sets a node's vetted_at timestamp to null to make testing easier.
func (service *Service) TestUnvetNode(ctx context.Context, nodeID storj.NodeID) (err error) {
err = service.db.TestUnvetNode(ctx, nodeID)
if err != nil {
service.log.Warn("error unvetting node", zap.Stringer("node ID", nodeID), zap.Error(err))
return err
}
err = service.UploadSelectionCache.Refresh(ctx)
service.log.Warn("nodecache refresh err", zap.Error(err))
return err
}
// TestNodeCountryCode directly sets a node's vetted_at timestamp to null to make testing easier.
func (service *Service) TestNodeCountryCode(ctx context.Context, nodeID storj.NodeID, countryCode string) (err error) {
err = service.db.TestNodeCountryCode(ctx, nodeID, countryCode)
if err != nil {
service.log.Warn("error updating node", zap.Stringer("node ID", nodeID), zap.Error(err))
return err
}
return nil
}