storj/cmd/satellite/main.go
Jeff Wendling 78c6d5bb32 satellite/satellitedb: reported_serials table for processing orders
this commit introduces the reported_serials table. its purpose is
to allow for blind writes into it as nodes report in so that we have
minimal contention. in order to continue to accurately account for
used bandwidth, though, we cannot immediately add the settled amount.
if we did, we would have to give up on blind writes.

the table's primary key is structured precisely so that we can quickly
find expired orders and so that we maximally benefit from rocksdb
path prefix compression. we do this by rounding the expires at time
forward to the next day, effectively giving us storagenode petnames
for free. and since there's no secondary index or foreign key
constraints, this design should use significantly less space than
the current used_serials table while also reducing contention.

after inserting the orders into the table, we have a chore that
periodically consumes all of the expired orders in it and inserts
them into the existing rollups tables. this is as if we changed
the nodes to report as the order expired rather than as soon as
possible, so the belief in correctness of the refactor is higher.

since we are able to process large batches of orders (typically
a day's worth), we can use the code to maximally batch inserts into
the rollup tables to make inserts as friendly as possible to
cockroach.

Change-Id: I25d609ca2679b8331979184f16c6d46d4f74c1a6
2020-01-15 19:21:21 -07:00

441 lines
13 KiB
Go

// Copyright (C) 2019 Storj Labs, Inc.
// See LICENSE for copying information.
package main
import (
"context"
"fmt"
"os"
"path/filepath"
"text/tabwriter"
"time"
"github.com/skyrings/skyring-common/tools/uuid"
"github.com/spf13/cobra"
"github.com/zeebo/errs"
"go.uber.org/zap"
"storj.io/common/fpath"
"storj.io/common/storj"
"storj.io/storj/cmd/satellite/reports"
"storj.io/storj/pkg/cache"
"storj.io/storj/pkg/cfgstruct"
"storj.io/storj/pkg/process"
"storj.io/storj/pkg/revocation"
"storj.io/storj/private/version"
"storj.io/storj/satellite"
"storj.io/storj/satellite/accounting/live"
"storj.io/storj/satellite/metainfo"
"storj.io/storj/satellite/satellitedb"
)
// Satellite defines satellite configuration
type Satellite struct {
Database string `help:"satellite database connection string" releaseDefault:"postgres://" devDefault:"postgres://"`
DatabaseOptions struct {
APIKeysCache struct {
Expiration time.Duration `help:"satellite database api key expiration" default:"60s"`
Capacity int `help:"satellite database api key lru capacity" default:"1000"`
}
}
satellite.Config
}
// APIKeysLRUOptions returns a cache.Options based on the APIKeys LRU config
func (s *Satellite) APIKeysLRUOptions() cache.Options {
return cache.Options{
Expiration: s.DatabaseOptions.APIKeysCache.Expiration,
Capacity: s.DatabaseOptions.APIKeysCache.Capacity,
}
}
var (
rootCmd = &cobra.Command{
Use: "satellite",
Short: "Satellite",
}
runCmd = &cobra.Command{
Use: "run",
Short: "Run the satellite",
RunE: cmdRun,
}
runMigrationCmd = &cobra.Command{
Use: "migration",
Short: "Run the satellite database migration",
RunE: cmdMigrationRun,
}
runAPICmd = &cobra.Command{
Use: "api",
Short: "Run the satellite API",
RunE: cmdAPIRun,
}
runRepairerCmd = &cobra.Command{
Use: "repair",
Short: "Run the repair service",
RunE: cmdRepairerRun,
}
setupCmd = &cobra.Command{
Use: "setup",
Short: "Create config files",
RunE: cmdSetup,
Annotations: map[string]string{"type": "setup"},
}
qdiagCmd = &cobra.Command{
Use: "qdiag",
Short: "Repair Queue Diagnostic Tool support",
RunE: cmdQDiag,
}
reportsCmd = &cobra.Command{
Use: "reports",
Short: "Generate a report",
}
nodeUsageCmd = &cobra.Command{
Use: "storagenode-usage [start] [end]",
Short: "Generate a node usage report for a given period to use for payments",
Long: "Generate a node usage report for a given period to use for payments. Format dates using YYYY-MM-DD. The end date is exclusive.",
Args: cobra.MinimumNArgs(2),
RunE: cmdNodeUsage,
}
partnerAttributionCmd = &cobra.Command{
Use: "partner-attribution [partner ID] [start] [end]",
Short: "Generate a partner attribution report for a given period to use for payments",
Long: "Generate a partner attribution report for a given period to use for payments. Format dates using YYYY-MM-DD. The end date is exclusive.",
Args: cobra.MinimumNArgs(3),
RunE: cmdValueAttribution,
}
gracefulExitCmd = &cobra.Command{
Use: "graceful-exit [start] [end]",
Short: "Generate a graceful exit report",
Long: "Generate a node usage report for a given period to use for payments. Format dates using YYYY-MM-DD. The end date is exclusive.",
Args: cobra.MinimumNArgs(2),
RunE: cmdGracefulExit,
}
verifyGracefulExitReceiptCmd = &cobra.Command{
Use: "verify-exit-receipt [storage node ID] [receipt]",
Short: "Verify a graceful exit receipt",
Long: "Verify a graceful exit receipt is valid.",
Args: cobra.MinimumNArgs(2),
RunE: cmdVerifyGracefulExitReceipt,
}
runCfg Satellite
setupCfg Satellite
qdiagCfg struct {
Database string `help:"satellite database connection string" releaseDefault:"postgres://" devDefault:"postgres://"`
QListLimit int `help:"maximum segments that can be requested" default:"1000"`
}
nodeUsageCfg struct {
Database string `help:"satellite database connection string" releaseDefault:"postgres://" devDefault:"postgres://"`
Output string `help:"destination of report output" default:""`
}
partnerAttribtionCfg struct {
Database string `help:"satellite database connection string" releaseDefault:"postgres://" devDefault:"postgres://"`
Output string `help:"destination of report output" default:""`
}
gracefulExitCfg struct {
Database string `help:"satellite database connection string" releaseDefault:"postgres://" devDefault:"postgres://"`
Output string `help:"destination of report output" default:""`
Completed bool `help:"whether to output (initiated and completed) or (initiated and not completed)" default:"false"`
}
verifyGracefulExitReceiptCfg struct {
}
confDir string
identityDir string
)
func init() {
defaultConfDir := fpath.ApplicationDir("storj", "satellite")
defaultIdentityDir := fpath.ApplicationDir("storj", "identity", "satellite")
cfgstruct.SetupFlag(zap.L(), rootCmd, &confDir, "config-dir", defaultConfDir, "main directory for satellite configuration")
cfgstruct.SetupFlag(zap.L(), rootCmd, &identityDir, "identity-dir", defaultIdentityDir, "main directory for satellite identity credentials")
defaults := cfgstruct.DefaultsFlag(rootCmd)
rootCmd.AddCommand(runCmd)
runCmd.AddCommand(runMigrationCmd)
runCmd.AddCommand(runAPICmd)
runCmd.AddCommand(runRepairerCmd)
rootCmd.AddCommand(setupCmd)
rootCmd.AddCommand(qdiagCmd)
rootCmd.AddCommand(reportsCmd)
reportsCmd.AddCommand(nodeUsageCmd)
reportsCmd.AddCommand(partnerAttributionCmd)
reportsCmd.AddCommand(gracefulExitCmd)
reportsCmd.AddCommand(verifyGracefulExitReceiptCmd)
process.Bind(runCmd, &runCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
process.Bind(runMigrationCmd, &runCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
process.Bind(runAPICmd, &runCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
process.Bind(runRepairerCmd, &runCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
process.Bind(setupCmd, &setupCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir), cfgstruct.SetupMode())
process.Bind(qdiagCmd, &qdiagCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
process.Bind(nodeUsageCmd, &nodeUsageCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
process.Bind(gracefulExitCmd, &gracefulExitCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
process.Bind(verifyGracefulExitReceiptCmd, &verifyGracefulExitReceiptCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
process.Bind(partnerAttributionCmd, &partnerAttribtionCfg, defaults, cfgstruct.ConfDir(confDir), cfgstruct.IdentityDir(identityDir))
}
func cmdRun(cmd *cobra.Command, args []string) (err error) {
// inert constructors only ====
ctx, _ := process.Ctx(cmd)
log := zap.L()
identity, err := runCfg.Identity.Load()
if err != nil {
zap.S().Fatal(err)
}
db, err := satellitedb.New(log.Named("db"), runCfg.Database, satellitedb.Options{
ReportedRollupsReadBatchSize: runCfg.Orders.SettlementBatchSize,
})
if err != nil {
return errs.New("Error starting master database on satellite: %+v", err)
}
defer func() {
err = errs.Combine(err, db.Close())
}()
pointerDB, err := metainfo.NewStore(log.Named("pointerdb"), runCfg.Metainfo.DatabaseURL)
if err != nil {
return errs.New("Error creating revocation database: %+v", err)
}
defer func() {
err = errs.Combine(err, db.Close())
}()
revocationDB, err := revocation.NewDBFromCfg(runCfg.Server.Config)
if err != nil {
return errs.New("Error creating revocation database: %+v", err)
}
defer func() {
err = errs.Combine(err, revocationDB.Close())
}()
liveAccounting, err := live.NewCache(log.Named("live-accounting"), runCfg.LiveAccounting)
if err != nil {
return errs.New("Error creating live accounting cache: %+v", err)
}
defer func() {
err = errs.Combine(err, liveAccounting.Close())
}()
peer, err := satellite.New(log, identity, db, pointerDB, revocationDB, liveAccounting, version.Build, &runCfg.Config)
if err != nil {
return err
}
// okay, start doing stuff ====
err = peer.Version.CheckVersion(ctx)
if err != nil {
return err
}
if err := process.InitMetricsWithCertPath(ctx, log, nil, runCfg.Identity.CertPath); err != nil {
zap.S().Warn("Failed to initialize telemetry batcher: ", err)
}
err = db.CheckVersion(ctx)
if err != nil {
zap.S().Fatal("failed satellite database version check: ", err)
return errs.New("Error checking version for satellitedb: %+v", err)
}
runError := peer.Run(ctx)
closeError := peer.Close()
return errs.Combine(runError, closeError)
}
func cmdMigrationRun(cmd *cobra.Command, args []string) (err error) {
ctx, _ := process.Ctx(cmd)
log := zap.L()
db, err := satellitedb.New(log.Named("migration"), runCfg.Database, satellitedb.Options{})
if err != nil {
return errs.New("Error creating new master database connection for satellitedb migration: %+v", err)
}
defer func() {
err = errs.Combine(err, db.Close())
}()
err = db.CreateTables(ctx)
if err != nil {
return errs.New("Error creating tables for master database on satellite: %+v", err)
}
// There should be an explicit CreateTables call for the pointerdb as well.
// This is tracked in jira ticket #3337.
pdb, err := metainfo.NewStore(log.Named("migration"), runCfg.Metainfo.DatabaseURL)
if err != nil {
return errs.New("Error creating tables for pointer database on satellite: %+v", err)
}
defer func() {
err = errs.Combine(err, pdb.Close())
}()
return nil
}
func cmdSetup(cmd *cobra.Command, args []string) (err error) {
setupDir, err := filepath.Abs(confDir)
if err != nil {
return err
}
valid, _ := fpath.IsValidSetupDir(setupDir)
if !valid {
return fmt.Errorf("satellite configuration already exists (%v)", setupDir)
}
err = os.MkdirAll(setupDir, 0700)
if err != nil {
return err
}
return process.SaveConfig(cmd, filepath.Join(setupDir, "config.yaml"))
}
func cmdQDiag(cmd *cobra.Command, args []string) (err error) {
// open the master db
database, err := satellitedb.New(zap.L().Named("db"), qdiagCfg.Database, satellitedb.Options{})
if err != nil {
return errs.New("error connecting to master database on satellite: %+v", err)
}
defer func() {
err := database.Close()
if err != nil {
fmt.Printf("error closing connection to master database on satellite: %+v\n", err)
}
}()
list, err := database.RepairQueue().SelectN(context.Background(), qdiagCfg.QListLimit)
if err != nil {
return err
}
// initialize the table header (fields)
const padding = 3
w := tabwriter.NewWriter(os.Stdout, 0, 0, padding, ' ', tabwriter.AlignRight|tabwriter.Debug)
fmt.Fprintln(w, "Path\tLost Pieces\t")
// populate the row fields
for _, v := range list {
fmt.Fprint(w, v.GetPath(), "\t", v.GetLostPieces(), "\t")
}
// display the data
return w.Flush()
}
func cmdVerifyGracefulExitReceipt(cmd *cobra.Command, args []string) (err error) {
ctx, _ := process.Ctx(cmd)
identity, err := runCfg.Identity.Load()
if err != nil {
zap.S().Fatal(err)
}
// Check the node ID is valid
nodeID, err := storj.NodeIDFromString(args[0])
if err != nil {
return errs.Combine(err, errs.New("Invalid node ID."))
}
return verifyGracefulExitReceipt(ctx, identity, nodeID, args[1])
}
func cmdGracefulExit(cmd *cobra.Command, args []string) (err error) {
ctx, _ := process.Ctx(cmd)
start, end, err := reports.ParseRange(args[0], args[1])
if err != nil {
return err
}
// send output to stdout
if gracefulExitCfg.Output == "" {
return generateGracefulExitCSV(ctx, gracefulExitCfg.Completed, start, end, os.Stdout)
}
// send output to file
file, err := os.Create(gracefulExitCfg.Output)
if err != nil {
return err
}
defer func() {
err = errs.Combine(err, file.Close())
}()
return generateGracefulExitCSV(ctx, gracefulExitCfg.Completed, start, end, file)
}
func cmdNodeUsage(cmd *cobra.Command, args []string) (err error) {
ctx, _ := process.Ctx(cmd)
start, end, err := reports.ParseRange(args[0], args[1])
if err != nil {
return err
}
// send output to stdout
if nodeUsageCfg.Output == "" {
return generateNodeUsageCSV(ctx, start, end, os.Stdout)
}
// send output to file
file, err := os.Create(nodeUsageCfg.Output)
if err != nil {
return err
}
defer func() {
err = errs.Combine(err, file.Close())
}()
return generateNodeUsageCSV(ctx, start, end, file)
}
func cmdValueAttribution(cmd *cobra.Command, args []string) (err error) {
ctx, _ := process.Ctx(cmd)
log := zap.L().Named("satellite-cli")
// Parse the UUID
partnerID, err := uuid.Parse(args[0])
if err != nil {
return errs.Combine(errs.New("Invalid Partner ID format. %s", args[0]), err)
}
start, end, err := reports.ParseRange(args[1], args[2])
if err != nil {
return err
}
// send output to stdout
if partnerAttribtionCfg.Output == "" {
return reports.GenerateAttributionCSV(ctx, partnerAttribtionCfg.Database, *partnerID, start, end, os.Stdout)
}
// send output to file
file, err := os.Create(partnerAttribtionCfg.Output)
if err != nil {
return err
}
defer func() {
err = errs.Combine(err, file.Close())
if err != nil {
log.Sugar().Errorf("error closing the file %v after retrieving partner value attribution data: %+v", partnerAttribtionCfg.Output, err)
}
}()
return reports.GenerateAttributionCSV(ctx, partnerAttribtionCfg.Database, *partnerID, start, end, file)
}
func main() {
process.Exec(rootCmd)
}