commit a1f1d4a14a0341ad215cbe15859791d4f92b6fb5 · edavis.dev/monarch

+82

cmd/backfiller/backend.go

···

       1
       +
       package main

     

       2
       +
       

     

       3
       +
       import (

     

       4
       +
       	"context"

     

       5
       +
       	"errors"

     

       6
       +
       	"fmt"

     

       7
       +
       	"log/slog"

     

       8
       +
       	"sync"

     

       9
       +
       	"time"

     

       10
       +
       

     

       11
       +
       	comatproto "github.com/bluesky-social/indigo/api/atproto"

     

       12
       +
       	"github.com/bluesky-social/indigo/backfill"

     

       13
       +
       	"gorm.io/gorm"

     

       14
       +
       )

     

       15
       +
       

     

       16
       +
       type Backend struct {

     

       17
       +
       	state   *gorm.DB

     

       18
       +
       	bf      *backfill.Backfiller

     

       19
       +
       	seqLk   sync.Mutex

     

       20
       +
       	lastSeq int64

     

       21
       +
       }

     

       22
       +
       

     

       23
       +
       func NewBackend(state *gorm.DB, bf *backfill.Backfiller) *Backend {

     

       24
       +
       	return &Backend{

     

       25
       +
       		state: state,

     

       26
       +
       		bf:    bf,

     

       27
       +
       	}

     

       28
       +
       }

     

       29
       +
       

     

       30
       +
       func (b *Backend) RepoCommitHandler(ctx context.Context, evt *comatproto.SyncSubscribeRepos_Commit) error {

     

       31
       +
       	b.seqLk.Lock()

     

       32
       +
       	if evt.Seq > b.lastSeq {

     

       33
       +
       		b.lastSeq = evt.Seq

     

       34
       +
       	}

     

       35
       +
       	b.seqLk.Unlock()

     

       36
       +
       

     

       37
       +
       	job, err := b.bf.Store.GetJob(ctx, evt.Repo)

     

       38
       +
       	if job == nil {

     

       39
       +
       		if errors.Is(err, backfill.ErrJobNotFound) {

     

       40
       +
       			return nil

     

       41
       +
       		} else {

     

       42
       +
       			return fmt.Errorf("error getting job: %w", err)

     

       43
       +
       		}

     

       44
       +
       	} else {

     

       45
       +
       		return b.bf.HandleEvent(ctx, evt)

     

       46
       +
       	}

     

       47
       +
       }

     

       48
       +
       

     

       49
       +
       func (b *Backend) LoadCursor() (int, error) {

     

       50
       +
       	var rec cursorRecord

     

       51
       +
       	if err := b.state.Find(&rec, "id = 1").Error; err != nil {

     

       52
       +
       		return 0, err

     

       53
       +
       	}

     

       54
       +
       

     

       55
       +
       	if rec.ID == 0 {

     

       56
       +
       		if err := b.state.Create(&cursorRecord{ID: 1}).Error; err != nil {

     

       57
       +
       			return 0, err

     

       58
       +
       		}

     

       59
       +
       	}

     

       60
       +
       

     

       61
       +
       	return rec.Val, nil

     

       62
       +
       }

     

       63
       +
       

     

       64
       +
       func (b *Backend) FlushCursor() error {

     

       65
       +
       	b.seqLk.Lock()

     

       66
       +
       	v := b.lastSeq

     

       67
       +
       	b.seqLk.Unlock()

     

       68
       +
       

     

       69
       +
       	if err := b.state.Model(cursorRecord{}).Where("id = 1 and val < ?", v).Update("val", v).Error; err != nil {

     

       70
       +
       		return err

     

       71
       +
       	}

     

       72
       +
       

     

       73
       +
       	return nil

     

       74
       +
       }

     

       75
       +
       

     

       76
       +
       func (b *Backend) syncCursorRoutine() {

     

       77
       +
       	for range time.Tick(time.Second * 5) {

     

       78
       +
       		if err := b.FlushCursor(); err != nil {

     

       79
       +
       			slog.Error("failed to flush cursor", "err", err)

     

       80
       +
       		}

     

       81
       +
       	}

     

       82
       +
       }

+33 -24

cmd/backfiller/backfiller.go

···

       2
        
       

     

       3
        
       import (

     

       4
        
       	"context"

     

       5
       -
       	"errors"

     

       6
        
       	"fmt"

     

       7
        
       	"log"

     

       8
        
       	"log/slog"

     

       9
        
       	"net/http"

     

       10
        
       	"os"

     

       11
        
       	"os/signal"

     

       0
        
       
     

       12
        
       	"syscall"

     

       13
        
       	"time"

     

       14
        
       

     

       15
        
       	"github.com/bluesky-social/indigo/api/atproto"

     

       16
       -
       	comatproto "github.com/bluesky-social/indigo/api/atproto"

     

       17
        
       	"github.com/bluesky-social/indigo/backfill"

     

       18
        
       	"github.com/bluesky-social/indigo/events"

     

       19
        
       	"github.com/bluesky-social/indigo/events/schedulers/parallel"

     
···

       25
        
       	"gorm.io/gorm"

     

       26
        
       	"gorm.io/gorm/logger"

     

       27
        
       )

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       28
        
       

     

       29
        
       func handleCreate(ctx context.Context, repo, rev, path string, rec *[]byte, cid *cid.Cid) error {

     

       30
        
       	return nil

     
···

       47
        
       	newLogger := logger.New(

     

       48
        
       		log.New(os.Stdout, "\n", log.LstdFlags),

     

       49
        
       		logger.Config{

     

       50
       -
       			SlowThreshold: 5*time.Second,

     

       51
       -
       			Colorful: false,

     

       52
        
       		},

     

       53
        
       	)

     

       54
        
       	db, err := gorm.Open(sqlite.Open("state.db"), &gorm.Config{

     
···

       58
        
       		sl.Error("failed to connect to database", "err", err)

     

       59
        
       	}

     

       60
        
       	db.AutoMigrate(&backfill.GormDBJob{})

     

       0
        
       
     

       61
        
       	store := backfill.NewGormstore(db)

     

       62
        
       

     

       63
       -
       	// connect to the relay

     

       64
       -
       	con, _, err := websocket.DefaultDialer.Dial("wss://bsky.network/xrpc/com.atproto.sync.subscribeRepos", http.Header{

     

       65
       -
       		"User-Agent": []string{"backfiller/0.1 (@edavis.dev)"},

     

       66
       -
       	})

     

       67
       -
       	if err != nil {

     

       68
       -
       		sl.Error("failed to connect to relay", "err", err)

     

       69
       -
       	}

     

       70
       -
       

     

       71
       -
       	// start backfilling

     

       72
        
       	opts := &backfill.BackfillOptions{

     

       73
        
       		ParallelBackfills:     10,

     

       74
        
       		ParallelRecordCreates: 100,

     
···

       78
        
       	bf := backfill.NewBackfiller("backfills", store, handleCreate, handleUpdate, handleDelete, opts)

     

       79
        
       	go bf.Start()

     

       80
        
       

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       81
        
       	// pump repos

     

       82
        
       	go func(ctx context.Context, bf *backfill.Backfiller) {

     

       83
        
       		if err := pumpRepos(ctx, bf); err != nil {

     
···

       87
        
       

     

       88
        
       	// read from the firehose

     

       89
        
       	rsc := events.RepoStreamCallbacks{

     

       90
       -
       		RepoCommit: func(evt *comatproto.SyncSubscribeRepos_Commit) error {

     

       91
       -
       			job, err := bf.Store.GetJob(ctx, evt.Repo)

     

       92
       -
       			if job == nil {

     

       93
       -
       				if errors.Is(err, backfill.ErrJobNotFound) {

     

       94
       -
       					return nil

     

       95
       -
       				} else {

     

       96
       -
       					return fmt.Errorf("error getting job: %w", err)

     

       97
       -
       				}

     

       98
       -
       			} else {

     

       99
       -
       				return bf.HandleEvent(ctx, evt)

     

       100
       -
       			}

     

       101
        
       		},

     

       102
        
       	}

     

       103
        
       	sched := parallel.NewScheduler(16, 100, "firehose", rsc.EventHandler)

···

       2
        
       

     

       3
        
       import (

     

       4
        
       	"context"

     

       0
        
       
     

       5
        
       	"fmt"

     

       6
        
       	"log"

     

       7
        
       	"log/slog"

     

       8
        
       	"net/http"

     

       9
        
       	"os"

     

       10
        
       	"os/signal"

     

       11
       +
       	"strconv"

     

       12
        
       	"syscall"

     

       13
        
       	"time"

     

       14
        
       

     

       15
        
       	"github.com/bluesky-social/indigo/api/atproto"

     

       0
        
       
     

       16
        
       	"github.com/bluesky-social/indigo/backfill"

     

       17
        
       	"github.com/bluesky-social/indigo/events"

     

       18
        
       	"github.com/bluesky-social/indigo/events/schedulers/parallel"

     
···

       24
        
       	"gorm.io/gorm"

     

       25
        
       	"gorm.io/gorm/logger"

     

       26
        
       )

     

       27
       +
       

     

       28
       +
       type cursorRecord struct {

     

       29
       +
       	ID  uint `gorm:"primaryKey"`

     

       30
       +
       	Val int

     

       31
       +
       }

     

       32
        
       

     

       33
        
       func handleCreate(ctx context.Context, repo, rev, path string, rec *[]byte, cid *cid.Cid) error {

     

       34
        
       	return nil

     
···

       51
        
       	newLogger := logger.New(

     

       52
        
       		log.New(os.Stdout, "\n", log.LstdFlags),

     

       53
        
       		logger.Config{

     

       54
       +
       			SlowThreshold: 5 * time.Second,

     

       55
       +
       			Colorful:      false,

     

       56
        
       		},

     

       57
        
       	)

     

       58
        
       	db, err := gorm.Open(sqlite.Open("state.db"), &gorm.Config{

     
···

       62
        
       		sl.Error("failed to connect to database", "err", err)

     

       63
        
       	}

     

       64
        
       	db.AutoMigrate(&backfill.GormDBJob{})

     

       65
       +
       	db.AutoMigrate(&cursorRecord{})

     

       66
        
       	store := backfill.NewGormstore(db)

     

       67
        
       

     

       68
       +
       	// create and start the backfiller

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       69
        
       	opts := &backfill.BackfillOptions{

     

       70
        
       		ParallelBackfills:     10,

     

       71
        
       		ParallelRecordCreates: 100,

     
···

       75
        
       	bf := backfill.NewBackfiller("backfills", store, handleCreate, handleUpdate, handleDelete, opts)

     

       76
        
       	go bf.Start()

     

       77
        
       

     

       78
       +
       	// set up the backend service

     

       79
       +
       	backend := NewBackend(db, bf)

     

       80
       +
       	go backend.syncCursorRoutine()

     

       81
       +
       

     

       82
       +
       	// connect to the relay

     

       83
       +
       	subscribeUrl := "wss://bsky.network/xrpc/com.atproto.sync.subscribeRepos"

     

       84
       +
       	curs, err := backend.LoadCursor()

     

       85
       +
       	if err != nil {

     

       86
       +
       		sl.Error("error loading firehose cursor", "err", err)

     

       87
       +
       	}

     

       88
       +
       	if curs > 0 {

     

       89
       +
       		subscribeUrl += "?cursor=" + strconv.Itoa(curs)

     

       90
       +
       	}

     

       91
       +
       

     

       92
       +
       	con, _, err := websocket.DefaultDialer.Dial(subscribeUrl, http.Header{

     

       93
       +
       		"User-Agent": []string{"backfiller/0.1 (@edavis.dev)"},

     

       94
       +
       	})

     

       95
       +
       	if err != nil {

     

       96
       +
       		sl.Error("failed to connect to relay", "err", err)

     

       97
       +
       	}

     

       98
       +
       

     

       99
        
       	// pump repos

     

       100
        
       	go func(ctx context.Context, bf *backfill.Backfiller) {

     

       101
        
       		if err := pumpRepos(ctx, bf); err != nil {

     
···

       105
        
       

     

       106
        
       	// read from the firehose

     

       107
        
       	rsc := events.RepoStreamCallbacks{

     

       108
       +
       		RepoCommit: func(evt *atproto.SyncSubscribeRepos_Commit) error {

     

       109
       +
       			return backend.RepoCommitHandler(ctx, evt)

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       110
        
       		},

     

       111
        
       	}

     

       112
        
       	sched := parallel.NewScheduler(16, 100, "firehose", rsc.EventHandler)