Backport index and ipni pdp tasks to PDPv0

cmd/curio/tasks/tasks.go

@@ -287,6 +287,7 @@ func StartTasks(ctx context.Context, dependencies *deps.Deps, shutdownChan chan
 			return nil, err
 		}
 		var sdeps cuhttp.ServiceDeps
+		idxMax := taskhelp.Max(cfg.Subsystems.IndexingMaxTasks)
 
 		if cfg.Subsystems.EnablePDP {
 			es := getSenderEth()
@@ -299,11 +300,11 @@ func StartTasks(ctx context.Context, dependencies *deps.Deps, shutdownChan chan
 			pdpNextProvingPeriodTask := pdp.NewNextProvingPeriodTask(db, must.One(dependencies.EthClient.Val()), dependencies.Chain, chainSched, es)
 			pdpInitProvingPeriodTask := pdp.NewInitProvingPeriodTask(db, must.One(dependencies.EthClient.Val()), dependencies.Chain, chainSched, es)
 			pdpNotifTask := pdp.NewPDPNotifyTask(db)
-			activeTasks = append(activeTasks, pdpNotifTask, pdpProveTask, pdpNextProvingPeriodTask, pdpInitProvingPeriodTask)
+			pdpIndexingTask := indexing.NewPDPIndexingTask(db, iStore, dependencies.CachedPieceReader, cfg, idxMax)
+			pdpIpniTask := indexing.NewPDPIPNITask(db, sc, dependencies.CachedPieceReader, cfg, idxMax)
+			activeTasks = append(activeTasks, pdpNotifTask, pdpProveTask, pdpNextProvingPeriodTask, pdpInitProvingPeriodTask, pdpIndexingTask, pdpIpniTask)
 		}
 
-		idxMax := taskhelp.Max(cfg.Subsystems.IndexingMaxTasks)
-
 		indexingTask := indexing.NewIndexingTask(db, sc, iStore, pp, cfg, idxMax)
 		ipniTask := indexing.NewIPNITask(db, sc, iStore, pp, cfg, idxMax)
 		activeTasks = append(activeTasks, ipniTask, indexingTask)

deps/config/types.go

@@ -114,7 +114,7 @@ func DefaultCurioConfig() *CurioConfig {
 				},
 				IPNI: IPNIConfig{
 					ServiceURL:         []string{"https://cid.contact"},
-					DirectAnnounceURLs: []string{"https://cid.contact/ingest/announce"},
+					DirectAnnounceURLs: []string{"https://cid.contact/ingest/announce", "https://filecoinpin.contact/announce"},
 				},
 			},
 		},

harmony/harmonydb/sql/20240823-ipni.sql

@@ -8,7 +8,7 @@ CREATE TABLE ipni_peerid (
 CREATE TABLE ipni (
     order_number BIGSERIAL PRIMARY KEY, -- Unique increasing order number
     ad_cid TEXT NOT NULL,
-    context_id BYTEA NOT NULL, -- abi.PieceInfo in Curio
+    context_id BYTEA NOT NULL, -- abi.PieceInfo || PDPIPNIContext in Curio
     -- metadata column in not required as Curio only supports one type of metadata(HTTP)
     is_rm BOOLEAN NOT NULL,
 

harmony/harmonydb/sql/20251004-pdp-indexing.sql

@@ -0,0 +1,5 @@
+-- fields tracking indexing and ipni jobs over pdp pieces 
+ALTER TABLE pdp_piecerefs ADD COLUMN indexing_task_id BIGINT DEFAULT NULL;
+ALTER TABLE pdp_piecerefs ADD COLUMN needs_indexing BOOLEAN DEFAULT FALSE;
+ALTER TABLE pdp_piecerefs ADD COLUMN ipni_task_id BIGINT DEFAULT NULL;
+ALTER TABLE pdp_piecerefs ADD COLUMN needs_ipni BOOLEAN DEFAULT FALSE;

market/ipni/ipni-provider/ipni-provider.go

@@ -49,8 +49,8 @@ const IPNIRoutePath = "/ipni-provider/"
 const IPNIPath = "/ipni/v1/ad/"
 
 // publishInterval represents the time interval between each publishing operation.
-// It is set to 10 minutes.
-const publishInterval = 10 * time.Minute
+// It is set to 30 seconds for the purposes of PDP index publishing
+const publishInterval = 30 * time.Second
 const publishProviderSpacing = 10 * time.Second
 
 var (
@@ -73,6 +73,7 @@ type Provider struct {
 	indexStore    *indexstore.IndexStore
 	sc            *chunker.ServeChunker
 	keys          map[string]*peerInfo // map[peerID String]Private_Key
+	latest        map[string]cid.Cid   // map[peerID String]last published head, used to avoid duplicate announce
 	// announceURLs enables sending direct announcements via HTTP. This is
 	// the list of indexer URLs to send direct HTTP announce messages to.
 	announceURLs []*url.URL
@@ -193,6 +194,7 @@ func NewProvider(d *deps.Deps) (*Provider, error) {
 		keys:                keyMap,
 		announceURLs:        announceURLs,
 		httpServerAddresses: httpServerAddresses,
+		latest:              make(map[string]cid.Cid),
 	}, nil
 }
 
@@ -364,7 +366,7 @@ func (p *Provider) handleGet(w http.ResponseWriter, r *http.Request) {
 	start := time.Now()
 
 	defer func() {
-		log.Infow("Served IPNI request", "path", r.URL.Path, "cid", reqCid, "providerId", providerID, "took", time.Since(start))
+		log.Infow("Served IPNI request", "path", r.URL.Path, "cid", reqCid, "providerId", providerID, "took", time.Since(start), "remote_addr", r.RemoteAddr)
 	}()
 
 	b, err := cid.Parse(reqCid)
@@ -477,7 +479,6 @@ func RemoveCidContact(slice []*url.URL) []*url.URL {
 // StartPublishing starts a poller which publishes the head for each provider every 10 minutes.
 func (p *Provider) StartPublishing(ctx context.Context) {
 	var ticker *time.Ticker
-
 	// A poller which publishes head for each provider
 	// every 10 minutes for mainnet build
 	if build.BuildType == build.BuildMainnet {
@@ -489,12 +490,22 @@ func (p *Provider) StartPublishing(ctx context.Context) {
 			return
 		}
 		log.Info("Starting IPNI provider publishing for testnet build")
+		ticker = time.NewTicker(publishInterval)
 		if build.BuildType != build.BuildCalibnet {
 			ticker = time.NewTicker(time.Second * 10)
 			log.Info("Resetting IPNI provider publishing ticker to 10 seconds for devnet build")
 		}
 	}
 
+	// Populated latest head cid from the ipni_head table
+	for provider := range p.keys {
+		c, err := p.getHeadCID(ctx, provider)
+		if err != nil {
+			log.Errorw("failed to get head CID", "provider", provider, "error", err)
+			continue
+		}
+		p.latest[provider] = c
+	}
 	go func(ticker *time.Ticker) {
 		for {
 			select {
@@ -544,10 +555,17 @@ func (p *Provider) publishHead(ctx context.Context) {
 			log.Errorw("failed to get head CID", "provider", provider, "error", err)
 			continue
 		}
+		if _, ok := p.latest[provider]; ok && p.latest[provider] == c {
+			log.Debugw("Skipping duplicate announce for provider", "provider", provider, "cid", c.String())
+			continue
+		}
+
 		log.Infow("Publishing head for provider", "provider", provider, "cid", c.String())
 		err = p.publishhttp(ctx, c, provider)
 		if err != nil {
 			log.Errorw("failed to publish head for provide", "provider", provider, "error", err)
+		} else {
+			p.latest[provider] = c
 		}
 
 		i++
@@ -567,6 +585,7 @@ func (p *Provider) publishProviderSpacingWait() {
 // It obtains the HTTP addresses for the peer and sends the announce message to those addresses.
 func (p *Provider) publishhttp(ctx context.Context, adCid cid.Cid, peer string) error {
 	// Create the http announce sender.
+	log.Infow("Creating http announce sender", "urls", p.announceURLs)
 	httpSender, err := httpsender.New(p.announceURLs, p.keys[peer].ID)
 	if err != nil {
 		return fmt.Errorf("cannot create http announce sender: %w", err)
@@ -577,7 +596,6 @@ func (p *Provider) publishhttp(ctx context.Context, adCid cid.Cid, peer string)
 		return fmt.Errorf("cannot create provider http addresses: %w", err)
 	}
 
-	log.Infow("Announcing advertisements over HTTP", "urls", p.announceURLs)
 	return announce.Send(ctx, adCid, addrs, httpSender)
 }
 

market/ipni/ipni-provider/spark.go

@@ -26,7 +26,10 @@ import (
 
 func (p *Provider) updateSparkContract(ctx context.Context) error {
 	for _, pInfo := range p.keys {
-		pInfo := pInfo
+		if pInfo.SPID <= 0 {
+			log.Debugf("spark does not yet support pdp data")
+			continue
+		}
 		mInfo, err := p.full.StateMinerInfo(ctx, pInfo.Miner, types.EmptyTSK)
 		if err != nil {
 			return err