storj/storagenode/peer.go

801 lines
22 KiB
Go
Raw Normal View History

2019-01-10 13:13:27 +00:00
// Copyright (C) 2019 Storj Labs, Inc.
// See LICENSE for copying information.
package storagenode
import (
"context"
"errors"
"fmt"
"net"
"net/http"
"path/filepath"
"time"
2019-01-10 13:13:27 +00:00
"github.com/spacemonkeygo/monkit/v3"
2019-01-10 13:13:27 +00:00
"github.com/zeebo/errs"
"go.uber.org/zap"
"golang.org/x/sync/errgroup"
"storj.io/common/identity"
"storj.io/common/pb"
"storj.io/common/peertls/extensions"
"storj.io/common/peertls/tlsopts"
"storj.io/common/rpc"
"storj.io/common/signing"
"storj.io/common/storj"
"storj.io/private/debug"
"storj.io/private/version"
2019-01-10 13:13:27 +00:00
"storj.io/storj/pkg/server"
"storj.io/storj/private/lifecycle"
"storj.io/storj/private/version/checker"
2019-01-10 13:13:27 +00:00
"storj.io/storj/storage"
"storj.io/storj/storage/filestore"
"storj.io/storj/storagenode/bandwidth"
2019-05-08 12:11:59 +01:00
"storj.io/storj/storagenode/collector"
"storj.io/storj/storagenode/console"
"storj.io/storj/storagenode/console/consoleassets"
"storj.io/storj/storagenode/console/consoleserver"
"storj.io/storj/storagenode/contact"
"storj.io/storj/storagenode/gracefulexit"
"storj.io/storj/storagenode/inspector"
"storj.io/storj/storagenode/monitor"
"storj.io/storj/storagenode/nodestats"
"storj.io/storj/storagenode/notifications"
"storj.io/storj/storagenode/orders"
"storj.io/storj/storagenode/payout"
"storj.io/storj/storagenode/payout/estimatedpayout"
"storj.io/storj/storagenode/pieces"
"storj.io/storj/storagenode/piecestore"
"storj.io/storj/storagenode/piecestore/usedserials"
"storj.io/storj/storagenode/piecetransfer"
"storj.io/storj/storagenode/preflight"
"storj.io/storj/storagenode/pricing"
"storj.io/storj/storagenode/reputation"
"storj.io/storj/storagenode/retain"
"storj.io/storj/storagenode/satellites"
"storj.io/storj/storagenode/secret"
"storj.io/storj/storagenode/storagenodedb"
"storj.io/storj/storagenode/storageusage"
"storj.io/storj/storagenode/trust"
version2 "storj.io/storj/storagenode/version"
2019-01-10 13:13:27 +00:00
)
var (
mon = monkit.Package()
)
2019-01-10 13:13:27 +00:00
// DB is the master database for Storage Node
2019-09-10 14:24:16 +01:00
//
// architecture: Master Database
2019-01-10 13:13:27 +00:00
type DB interface {
// MigrateToLatest initializes the database
MigrateToLatest(ctx context.Context) error
2019-01-24 20:28:06 +00:00
// Close closes the database
Close() error
Pieces() storage.Blobs
Orders() orders.DB
V0PieceInfo() pieces.V0PieceInfoDB
PieceExpirationDB() pieces.PieceExpirationDB
PieceSpaceUsedDB() pieces.PieceSpaceUsedDB
Bandwidth() bandwidth.DB
Reputation() reputation.DB
StorageUsage() storageusage.DB
Satellites() satellites.DB
Notifications() notifications.DB
Payout() payout.DB
Pricing() pricing.DB
Secret() secret.DB
Preflight(ctx context.Context) error
2019-01-10 13:13:27 +00:00
}
// Config is all the configuration parameters for a Storage Node.
2019-01-10 13:13:27 +00:00
type Config struct {
Identity identity.Config
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
Server server.Config
Debug debug.Config
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
Preflight preflight.Config
Contact contact.Config
Operator OperatorConfig
2019-05-08 12:11:59 +01:00
// TODO: flatten storage config and only keep the new one
Storage piecestore.OldConfig
Storage2 piecestore.Config
Collector collector.Config
Add Versioning Server (#1576) * 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 * Move vars private and set minimum default versions for testing builds * Remove VersionedClient * Better Error Handling and naked return removal * Straighten the Regex and string conversion * Change Check to allows testplanet and storj-sim to run without the need to pass an LDFlag * Cosmetic Change to Dashboard * Cleanup Returns and remove commented code * Remove Version Check if no build options are passed in * Pass in Config Values instead of Pointers * Handle missed Error * Update Endpoint URL * Change Type of Release Flag * Add additional Logging * Remove Versions Logging of other Services * minor fixes Change-Id: I5cc04a410ea6b2008d14dffd63eb5f36dd348a8b
2019-04-03 20:13:39 +01:00
Filestore filestore.Config
Pieces pieces.Config
Retain retain.Config
Nodestats nodestats.Config
Console consoleserver.Config
Version checker.Config
Bandwidth bandwidth.Config
GracefulExit gracefulexit.Config
2019-01-10 13:13:27 +00:00
}
// DatabaseConfig returns the storagenodedb.Config that should be used with this Config.
func (config *Config) DatabaseConfig() storagenodedb.Config {
dbdir := config.Storage2.DatabaseDir
if dbdir == "" {
dbdir = config.Storage.Path
}
return storagenodedb.Config{
Storage: config.Storage.Path,
Info: filepath.Join(dbdir, "piecestore.db"),
Info2: filepath.Join(dbdir, "info.db"),
Pieces: config.Storage.Path,
Filestore: config.Filestore,
}
}
2019-01-10 13:13:27 +00:00
// Verify verifies whether configuration is consistent and acceptable.
func (config *Config) Verify(log *zap.Logger) error {
err := config.Operator.Verify(log)
if err != nil {
return err
}
if config.Contact.ExternalAddress != "" {
err := isAddressValid(config.Contact.ExternalAddress)
if err != nil {
return errs.New("invalid contact.external-address: %v", err)
}
}
if config.Server.Address != "" {
err := isAddressValid(config.Server.Address)
if err != nil {
return errs.New("invalid server.address: %v", err)
}
}
return nil
}
func isAddressValid(addrstring string) error {
addr, port, err := net.SplitHostPort(addrstring)
if err != nil || port == "" {
return errs.New("split host-port %q failed: %+v", addrstring, err)
}
if addr == "" {
return nil
}
resolvedhosts, err := net.LookupHost(addr)
if err != nil || len(resolvedhosts) == 0 {
return errs.New("lookup %q failed: %+v", addr, err)
}
return nil
2019-01-10 13:13:27 +00:00
}
// Peer is the representation of a Storage Node.
2019-09-10 14:24:16 +01:00
//
// architecture: Peer
2019-01-10 13:13:27 +00:00
type Peer struct {
// core dependencies
Log *zap.Logger
Identity *identity.FullIdentity
DB DB
UsedSerials *usedserials.Table
OrdersStore *orders.FileStore
2019-01-10 13:13:27 +00:00
Servers *lifecycle.Group
Services *lifecycle.Group
Dialer rpc.Dialer
Server *server.Server
2019-01-10 13:13:27 +00:00
Version struct {
Chore *version2.Chore
Service *checker.Service
}
Add Versioning Server (#1576) * 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 * Move vars private and set minimum default versions for testing builds * Remove VersionedClient * Better Error Handling and naked return removal * Straighten the Regex and string conversion * Change Check to allows testplanet and storj-sim to run without the need to pass an LDFlag * Cosmetic Change to Dashboard * Cleanup Returns and remove commented code * Remove Version Check if no build options are passed in * Pass in Config Values instead of Pointers * Handle missed Error * Update Endpoint URL * Change Type of Release Flag * Add additional Logging * Remove Versions Logging of other Services * minor fixes Change-Id: I5cc04a410ea6b2008d14dffd63eb5f36dd348a8b
2019-04-03 20:13:39 +01:00
Debug struct {
Listener net.Listener
Server *debug.Server
}
2019-01-10 13:13:27 +00:00
// services and endpoints
// TODO: similar grouping to satellite.Core
2019-01-10 13:13:27 +00:00
Preflight struct {
LocalTime *preflight.LocalTime
}
Contact struct {
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
Service *contact.Service
Chore *contact.Chore
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
Endpoint *contact.Endpoint
PingStats *contact.PingStats
}
Estimation struct {
Service *estimatedpayout.Service
}
Storage2 struct {
2019-05-08 12:11:59 +01:00
// TODO: lift things outside of it to organize better
Trust *trust.Pool
Store *pieces.Store
TrashChore *pieces.TrashChore
BlobsCache *pieces.BlobsUsageCache
CacheService *pieces.CacheService
RetainService *retain.Service
PieceDeleter *pieces.Deleter
Endpoint *piecestore.Endpoint
Inspector *inspector.Endpoint
Monitor *monitor.Service
Orders *orders.Service
}
2019-05-08 12:11:59 +01:00
Collector *collector.Service
NodeStats struct {
Service *nodestats.Service
Cache *nodestats.Cache
}
// Web server with web UI
Console struct {
Listener net.Listener
Service *console.Service
Endpoint *consoleserver.Server
}
PieceTransfer struct {
Service piecetransfer.Service
}
GracefulExit struct {
Service gracefulexit.Service
Endpoint *gracefulexit.Endpoint
Chore *gracefulexit.Chore
BlobsCleaner *gracefulexit.BlobsCleaner
}
Notifications struct {
Service *notifications.Service
}
Payout struct {
Service *payout.Service
Endpoint *payout.Endpoint
}
Bandwidth *bandwidth.Service
2019-01-10 13:13:27 +00:00
}
// New creates a new Storage Node.
func New(log *zap.Logger, full *identity.FullIdentity, db DB, revocationDB extensions.RevocationDB, config Config, versionInfo version.Info, atomicLogLevel *zap.AtomicLevel) (*Peer, error) {
2019-01-10 13:13:27 +00:00
peer := &Peer{
Log: log,
Identity: full,
DB: db,
Servers: lifecycle.NewGroup(log.Named("servers")),
Services: lifecycle.NewGroup(log.Named("services")),
2019-01-10 13:13:27 +00:00
}
{ // setup notification service.
peer.Notifications.Service = notifications.NewService(peer.Log, peer.DB.Notifications())
}
{ // setup debug
var err error
if config.Debug.Address != "" {
peer.Debug.Listener, err = net.Listen("tcp", config.Debug.Address)
if err != nil {
withoutStack := errors.New(err.Error())
peer.Log.Debug("failed to start debug endpoints", zap.Error(withoutStack))
err = nil
}
}
debugConfig := config.Debug
debugConfig.ControlTitle = "Storage Node"
peer.Debug.Server = debug.NewServerWithAtomicLevel(log.Named("debug"), peer.Debug.Listener, monkit.Default, debugConfig, atomicLogLevel)
peer.Servers.Add(lifecycle.Item{
Name: "debug",
Run: peer.Debug.Server.Run,
Close: peer.Debug.Server.Close,
})
}
2019-01-10 13:13:27 +00:00
var err error
{ // version setup
if !versionInfo.IsZero() {
peer.Log.Debug("Version info",
zap.Stringer("Version", versionInfo.Version.Version),
zap.String("Commit Hash", versionInfo.CommitHash),
zap.Stringer("Build Timestamp", versionInfo.Timestamp),
zap.Bool("Release Build", versionInfo.Release),
)
Add Versioning Server (#1576) * 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 * Move vars private and set minimum default versions for testing builds * Remove VersionedClient * Better Error Handling and naked return removal * Straighten the Regex and string conversion * Change Check to allows testplanet and storj-sim to run without the need to pass an LDFlag * Cosmetic Change to Dashboard * Cleanup Returns and remove commented code * Remove Version Check if no build options are passed in * Pass in Config Values instead of Pointers * Handle missed Error * Update Endpoint URL * Change Type of Release Flag * Add additional Logging * Remove Versions Logging of other Services * minor fixes Change-Id: I5cc04a410ea6b2008d14dffd63eb5f36dd348a8b
2019-04-03 20:13:39 +01:00
}
peer.Version.Service = checker.NewService(log.Named("version"), config.Version, versionInfo, "Storagenode")
versionCheckInterval := 12 * time.Hour
peer.Version.Chore = version2.NewChore(peer.Log.Named("version:chore"), peer.Version.Service, peer.Notifications.Service, peer.Identity.ID, versionCheckInterval)
peer.Services.Add(lifecycle.Item{
Name: "version",
Run: peer.Version.Chore.Run,
})
Add Versioning Server (#1576) * 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 * Move vars private and set minimum default versions for testing builds * Remove VersionedClient * Better Error Handling and naked return removal * Straighten the Regex and string conversion * Change Check to allows testplanet and storj-sim to run without the need to pass an LDFlag * Cosmetic Change to Dashboard * Cleanup Returns and remove commented code * Remove Version Check if no build options are passed in * Pass in Config Values instead of Pointers * Handle missed Error * Update Endpoint URL * Change Type of Release Flag * Add additional Logging * Remove Versions Logging of other Services * minor fixes Change-Id: I5cc04a410ea6b2008d14dffd63eb5f36dd348a8b
2019-04-03 20:13:39 +01:00
}
2019-01-10 13:13:27 +00:00
{ // setup listener and server
sc := config.Server
tlsOptions, err := tlsopts.NewOptions(peer.Identity, sc.Config, revocationDB)
2019-01-10 13:13:27 +00:00
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
peer.Dialer = rpc.NewDefaultDialer(tlsOptions)
2019-01-10 13:13:27 +00:00
peer.Server, err = server.New(log.Named("server"), tlsOptions, sc.Address, sc.PrivateAddress)
2019-01-10 13:13:27 +00:00
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
peer.Servers.Add(lifecycle.Item{
Name: "server",
Run: func(ctx context.Context) error {
// Don't change the format of this comment, it is used to figure out the node id.
peer.Log.Info(fmt.Sprintf("Node %s started", peer.Identity.ID))
peer.Log.Info(fmt.Sprintf("Public server started on %s", peer.Addr()))
peer.Log.Info(fmt.Sprintf("Private server started on %s", peer.PrivateAddr()))
return peer.Server.Run(ctx)
},
Close: peer.Server.Close,
})
2019-01-10 13:13:27 +00:00
}
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
{ // setup trust pool
peer.Storage2.Trust, err = trust.NewPool(log.Named("trust"), trust.Dialer(peer.Dialer), config.Storage2.Trust)
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
peer.Services.Add(lifecycle.Item{
Name: "trust",
Run: peer.Storage2.Trust.Run,
})
}
{
peer.Preflight.LocalTime = preflight.NewLocalTime(peer.Log.Named("preflight:localtime"), config.Preflight, peer.Storage2.Trust, peer.Dialer)
}
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
{ // setup contact service
c := config.Contact
if c.ExternalAddress == "" {
c.ExternalAddress = peer.Addr()
2019-01-10 13:13:27 +00:00
}
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
pbVersion, err := versionInfo.Proto()
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
self := contact.NodeInfo{
ID: peer.ID(),
Address: c.ExternalAddress,
2019-04-22 10:07:50 +01:00
Operator: pb.NodeOperator{
Email: config.Operator.Email,
Wallet: config.Operator.Wallet,
2019-01-10 13:13:27 +00:00
},
2019-04-22 10:07:50 +01:00
Version: *pbVersion,
2019-01-10 13:13:27 +00:00
}
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
peer.Contact.PingStats = new(contact.PingStats)
peer.Contact.Service = contact.NewService(peer.Log.Named("contact:service"), peer.Dialer, self, peer.Storage2.Trust)
peer.Contact.Chore = contact.NewChore(peer.Log.Named("contact:chore"), config.Contact.Interval, peer.Contact.Service)
peer.Services.Add(lifecycle.Item{
Name: "contact:chore",
Run: peer.Contact.Chore.Run,
Close: peer.Contact.Chore.Close,
})
peer.Contact.Endpoint = contact.NewEndpoint(peer.Log.Named("contact:endpoint"), peer.Contact.PingStats)
if err := pb.DRPCRegisterContact(peer.Server.DRPC(), peer.Contact.Endpoint); err != nil {
return nil, errs.Combine(err, peer.Close())
}
}
2019-05-08 12:11:59 +01:00
{ // setup storage
peer.Storage2.BlobsCache = pieces.NewBlobsUsageCache(peer.Log.Named("blobscache"), peer.DB.Pieces())
peer.Storage2.Store = pieces.NewStore(peer.Log.Named("pieces"),
peer.Storage2.BlobsCache,
peer.DB.V0PieceInfo(),
peer.DB.PieceExpirationDB(),
peer.DB.PieceSpaceUsedDB(),
config.Pieces,
)
peer.Storage2.PieceDeleter = pieces.NewDeleter(log.Named("piecedeleter"), peer.Storage2.Store, config.Storage2.DeleteWorkers, config.Storage2.DeleteQueueSize)
peer.Services.Add(lifecycle.Item{
Name: "PieceDeleter",
Run: peer.Storage2.PieceDeleter.Run,
Close: peer.Storage2.PieceDeleter.Close,
})
peer.Storage2.TrashChore = pieces.NewTrashChore(
log.Named("pieces:trash"),
24*time.Hour, // choreInterval: how often to run the chore
7*24*time.Hour, // trashExpiryInterval: when items in the trash should be deleted
peer.Storage2.Trust,
peer.Storage2.Store,
)
peer.Services.Add(lifecycle.Item{
Name: "pieces:trash",
Run: peer.Storage2.TrashChore.Run,
Close: peer.Storage2.TrashChore.Close,
})
peer.Storage2.CacheService = pieces.NewService(
log.Named("piecestore:cache"),
peer.Storage2.BlobsCache,
peer.Storage2.Store,
config.Storage2.CacheSyncInterval,
)
peer.Services.Add(lifecycle.Item{
Name: "piecestore:cache",
Run: peer.Storage2.CacheService.Run,
Close: peer.Storage2.CacheService.Close,
})
peer.Debug.Server.Panel.Add(
debug.Cycle("Piecestore Cache", peer.Storage2.CacheService.Loop))
peer.Storage2.Monitor = monitor.NewService(
log.Named("piecestore:monitor"),
peer.Storage2.Store,
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
peer.Contact.Service,
peer.DB.Bandwidth(),
config.Storage.AllocatedDiskSpace.Int64(),
// TODO: use config.Storage.Monitor.Interval, but for some reason is not set
config.Storage.KBucketRefreshInterval,
peer.Contact.Chore.Trigger,
config.Storage2.Monitor,
)
peer.Services.Add(lifecycle.Item{
Name: "piecestore:monitor",
Run: peer.Storage2.Monitor.Run,
Close: peer.Storage2.Monitor.Close,
})
peer.Debug.Server.Panel.Add(
debug.Cycle("Piecestore Monitor", peer.Storage2.Monitor.Loop))
peer.Storage2.RetainService = retain.NewService(
peer.Log.Named("retain"),
peer.Storage2.Store,
config.Retain,
)
peer.Services.Add(lifecycle.Item{
Name: "retain",
Run: peer.Storage2.RetainService.Run,
Close: peer.Storage2.RetainService.Close,
})
peer.UsedSerials = usedserials.NewTable(config.Storage2.MaxUsedSerialsSize)
peer.OrdersStore, err = orders.NewFileStore(
peer.Log.Named("ordersfilestore"),
config.Storage2.Orders.Path,
config.Storage2.OrderLimitGracePeriod,
)
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
peer.Storage2.Endpoint, err = piecestore.NewEndpoint(
peer.Log.Named("piecestore"),
signing.SignerFromFullIdentity(peer.Identity),
peer.Storage2.Trust,
peer.Storage2.Monitor,
peer.Storage2.RetainService,
peer.Contact.PingStats,
peer.Storage2.Store,
peer.Storage2.PieceDeleter,
peer.OrdersStore,
peer.DB.Bandwidth(),
peer.UsedSerials,
config.Storage2,
)
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
if err := pb.DRPCRegisterPiecestore(peer.Server.DRPC(), peer.Storage2.Endpoint); err != nil {
return nil, errs.Combine(err, peer.Close())
}
// TODO workaround for custom timeout for order sending request (read/write)
sc := config.Server
tlsOptions, err := tlsopts.NewOptions(peer.Identity, sc.Config, revocationDB)
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
dialer := rpc.NewDefaultDialer(tlsOptions)
dialer.DialTimeout = config.Storage2.Orders.SenderDialTimeout
peer.Storage2.Orders = orders.NewService(
log.Named("orders"),
dialer,
peer.OrdersStore,
peer.DB.Orders(),
peer.Storage2.Trust,
config.Storage2.Orders,
)
peer.Services.Add(lifecycle.Item{
Name: "orders",
Run: peer.Storage2.Orders.Run,
Close: peer.Storage2.Orders.Close,
})
peer.Debug.Server.Panel.Add(
debug.Cycle("Orders Sender", peer.Storage2.Orders.Sender))
peer.Debug.Server.Panel.Add(
debug.Cycle("Orders Cleanup", peer.Storage2.Orders.Cleanup))
}
{ // setup payout service.
peer.Payout.Service = payout.NewService(
peer.Log.Named("payout:service"),
peer.DB.Payout(),
peer.DB.Reputation(),
peer.DB.Satellites(),
peer.Storage2.Trust,
)
peer.Payout.Endpoint = payout.NewEndpoint(
peer.Log.Named("payout:endpoint"),
peer.Dialer,
peer.Storage2.Trust,
)
}
{ // setup node stats service
peer.NodeStats.Service = nodestats.NewService(
peer.Log.Named("nodestats:service"),
peer.Dialer,
peer.Storage2.Trust,
)
peer.NodeStats.Cache = nodestats.NewCache(
peer.Log.Named("nodestats:cache"),
config.Nodestats,
nodestats.CacheStorage{
Reputation: peer.DB.Reputation(),
StorageUsage: peer.DB.StorageUsage(),
Payout: peer.DB.Payout(),
Pricing: peer.DB.Pricing(),
Satellites: peer.DB.Satellites(),
},
peer.NodeStats.Service,
peer.Payout.Endpoint,
peer.Payout.Service,
peer.Storage2.Trust,
)
peer.Services.Add(lifecycle.Item{
Name: "nodestats:cache",
Run: peer.NodeStats.Cache.Run,
Close: peer.NodeStats.Cache.Close,
})
peer.Debug.Server.Panel.Add(
debug.Cycle("Node Stats Cache Reputation", peer.NodeStats.Cache.Reputation))
peer.Debug.Server.Panel.Add(
debug.Cycle("Node Stats Cache Storage", peer.NodeStats.Cache.Storage))
}
{ // setup estimation service
peer.Estimation.Service = estimatedpayout.NewService(
peer.DB.Bandwidth(),
peer.DB.Reputation(),
peer.DB.StorageUsage(),
peer.DB.Pricing(),
peer.DB.Satellites(),
peer.Storage2.Trust,
)
}
{ // setup storage node operator dashboard
peer.Console.Service, err = console.NewService(
peer.Log.Named("console:service"),
peer.DB.Bandwidth(),
peer.Storage2.Store,
peer.Version.Service,
config.Storage.AllocatedDiskSpace,
config.Operator.Wallet,
versionInfo,
peer.Storage2.Trust,
peer.DB.Reputation(),
peer.DB.StorageUsage(),
peer.DB.Pricing(),
peer.DB.Satellites(),
peer.Contact.PingStats,
peer.Contact.Service,
peer.Estimation.Service,
peer.Storage2.BlobsCache,
)
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
peer.Console.Listener, err = net.Listen("tcp", config.Console.Address)
if err != nil {
return nil, errs.Combine(err, peer.Close())
}
assets := consoleassets.FileSystem
if config.Console.StaticDir != "" {
// a specific directory has been configured. use it
assets = http.Dir(config.Console.StaticDir)
}
peer.Console.Endpoint = consoleserver.NewServer(
peer.Log.Named("console:endpoint"),
assets,
peer.Notifications.Service,
peer.Console.Service,
peer.Payout.Service,
peer.Console.Listener,
)
peer.Services.Add(lifecycle.Item{
Name: "console:endpoint",
Run: peer.Console.Endpoint.Run,
Close: peer.Console.Endpoint.Close,
})
}
{ // setup storage inspector
peer.Storage2.Inspector = inspector.NewEndpoint(
peer.Log.Named("pieces:inspector"),
peer.Storage2.Store,
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
peer.Contact.Service,
peer.Contact.PingStats,
peer.DB.Bandwidth(),
config.Storage,
peer.Console.Listener.Addr(),
Remove Kademlia dependencies from Satellite and Storagenode (#2966) What: cmd/inspector/main.go: removes kad commands internal/testplanet/planet.go: Waits for contact chore to finish satellite/contact/nodesservice.go: creates an empty nodes service implementation satellite/contact/service.go: implements Local and FetchInfo methods & adds external address config value satellite/discovery/service.go: replaces kad.FetchInfo with contact.FetchInfo in Refresh() & removes Discover() satellite/peer.go: sets up contact service and endpoints storagenode/console/service.go: replaces nodeID with contact.Local() storagenode/contact/chore.go: replaces routing table with contact service storagenode/contact/nodesservice.go: creates empty implementation for ping and request info nodes service & implements RequestInfo method storagenode/contact/service.go: creates a service to return the local node and update its own capacity storagenode/monitor/monitor.go: uses contact service in place of routing table storagenode/operator.go: moves operatorconfig from kad into its own setup storagenode/peer.go: sets up contact service, chore, pingstats and endpoints satellite/overlay/config.go: changes NodeSelectionConfig.OnlineWindow default to 4hr to allow for accurate repair selection Removes kademlia setups in: cmd/storagenode/main.go cmd/storj-sim/network.go internal/testplane/planet.go internal/testplanet/satellite.go internal/testplanet/storagenode.go satellite/peer.go scripts/test-sim-backwards.sh scripts/testdata/satellite-config.yaml.lock storagenode/inspector/inspector.go storagenode/peer.go storagenode/storagenodedb/database.go Why: Replacing Kademlia Please describe the tests: • internal/testplanet/planet_test.go: TestBasic: assert that the storagenode can check in with the satellite without any errors TestContact: test that all nodes get inserted into both satellites' overlay cache during testplanet setup • satellite/contact/contact_test.go: TestFetchInfo: Tests that the FetchInfo method returns the correct info • storagenode/contact/contact_test.go: TestNodeInfoUpdated: tests that the contact chore updates the node information TestRequestInfoEndpoint: tests that the Request info endpoint returns the correct info Please describe the performance impact: Node discovery should be at least slightly more performant since each node connects directly to each satellite and no longer needs to wait for bootstrapping. It probably won't be faster in real time on start up since each node waits a random amount of time (less than 1 hr) to initialize its first connection (jitter).
2019-09-19 20:56:34 +01:00
config.Contact.ExternalAddress,
)
if err := pb.DRPCRegisterPieceStoreInspector(peer.Server.PrivateDRPC(), peer.Storage2.Inspector); err != nil {
return nil, errs.Combine(err, peer.Close())
}
}
{ // setup piecetransfer service
peer.PieceTransfer.Service = piecetransfer.NewService(
peer.Log.Named("piecetransfer"),
peer.Storage2.Store,
peer.Storage2.Trust,
peer.Dialer,
// using GracefulExit config here for historical reasons
config.GracefulExit.MinDownloadTimeout,
config.GracefulExit.MinBytesPerSecond,
)
}
{ // setup graceful exit service
peer.GracefulExit.Service = gracefulexit.NewService(
peer.Log.Named("gracefulexit:service"),
peer.Storage2.Store,
peer.Storage2.Trust,
peer.DB.Satellites(),
peer.Dialer,
config.GracefulExit,
)
peer.GracefulExit.Endpoint = gracefulexit.NewEndpoint(
peer.Log.Named("gracefulexit:endpoint"),
peer.Storage2.Trust,
peer.DB.Satellites(),
peer.Dialer,
peer.Storage2.BlobsCache,
)
if err := pb.DRPCRegisterNodeGracefulExit(peer.Server.PrivateDRPC(), peer.GracefulExit.Endpoint); err != nil {
return nil, errs.Combine(err, peer.Close())
}
peer.GracefulExit.Chore = gracefulexit.NewChore(
peer.Log.Named("gracefulexit:chore"),
peer.GracefulExit.Service,
peer.PieceTransfer.Service,
peer.Dialer,
config.GracefulExit,
)
peer.GracefulExit.BlobsCleaner = gracefulexit.NewBlobsCleaner(
peer.Log.Named("gracefulexit:blobscleaner"),
peer.Storage2.Store,
peer.Storage2.Trust,
peer.DB.Satellites(),
)
// Runs once on node start to clean blobs from trash that left after successful GE.
peer.Services.Add(lifecycle.Item{
Name: "gracefulexit:blobscleaner",
Run: peer.GracefulExit.BlobsCleaner.RemoveBlobs,
})
peer.Services.Add(lifecycle.Item{
Name: "gracefulexit:chore",
Run: peer.GracefulExit.Chore.Run,
Close: peer.GracefulExit.Chore.Close,
})
peer.Debug.Server.Panel.Add(
debug.Cycle("Graceful Exit", peer.GracefulExit.Chore.Loop))
}
peer.Collector = collector.NewService(peer.Log.Named("collector"), peer.Storage2.Store, peer.UsedSerials, config.Collector)
peer.Services.Add(lifecycle.Item{
Name: "collector",
Run: peer.Collector.Run,
Close: peer.Collector.Close,
})
peer.Debug.Server.Panel.Add(
debug.Cycle("Collector", peer.Collector.Loop))
2019-05-08 12:11:59 +01:00
peer.Bandwidth = bandwidth.NewService(peer.Log.Named("bandwidth"), peer.DB.Bandwidth(), config.Bandwidth)
peer.Services.Add(lifecycle.Item{
Name: "bandwidth",
Run: peer.Bandwidth.Run,
Close: peer.Bandwidth.Close,
})
peer.Debug.Server.Panel.Add(
debug.Cycle("Bandwidth", peer.Bandwidth.Loop))
2019-01-10 13:13:27 +00:00
return peer, nil
}
// Run runs storage node until it's either closed or it errors.
func (peer *Peer) Run(ctx context.Context) (err error) {
defer mon.Task()(&ctx)(&err)
// Refresh the trust pool first. It will be updated periodically via
// Run() below.
if err := peer.Storage2.Trust.Refresh(ctx); err != nil {
return err
}
if err := peer.Preflight.LocalTime.Check(ctx); err != nil {
peer.Log.Error("Failed preflight check.", zap.Error(err))
return err
}
2019-02-06 13:19:14 +00:00
group, ctx := errgroup.WithContext(ctx)
peer.Servers.Run(ctx, group)
peer.Services.Run(ctx, group)
2019-01-10 13:13:27 +00:00
return group.Wait()
}
// Close closes all the resources.
func (peer *Peer) Close() error {
return errs.Combine(
peer.Servers.Close(),
peer.Services.Close(),
)
2019-01-10 13:13:27 +00:00
}
// ID returns the peer ID.
func (peer *Peer) ID() storj.NodeID { return peer.Identity.ID }
// Addr returns the public address.
func (peer *Peer) Addr() string { return peer.Server.Addr().String() }
// URL returns the storj.NodeURL.
func (peer *Peer) URL() storj.NodeURL { return storj.NodeURL{ID: peer.ID(), Address: peer.Addr()} }
// PrivateAddr returns the private address.
func (peer *Peer) PrivateAddr() string { return peer.Server.PrivateAddr().String() }