feat: overhaul hp_agent lifecycle handling

* Added backoff and liveness probes for better management
* Switched IPC to a simple text based system
* Lookups don't directly touch the agent now
* Use the database as a source of truth

internal/tsnet/peers.go

@@ -3,10 +3,14 @@ package tsnet
 import (
 	"context"
 	"encoding/hex"
+	"encoding/json"
 	"fmt"
 	"strings"
+	"sync"
+	"time"
 
 	"github.com/tale/headplane/internal/util"
+	"tailscale.com/ipn/ipnstate"
 	"tailscale.com/tailcfg"
 	"tailscale.com/types/key"
 
@@ -64,3 +68,81 @@ func (s *TSAgent) GetStatusForPeer(id string) (*tailcfg.HostinfoView, error) {
 	log.Debug("Got whois for peer %s: %v", id, whois)
 	return &whois.Node.Hostinfo, nil
 }
+
+// Dispatches ALL the HostInfo entries in our Tailnet to the master
+func (s *TSAgent) DispatchHostInfo(ctx context.Context) error {
+	log := util.GetLogger()
+
+	stat, err := s.Lc.Status(ctx)
+	if err != nil {
+		log.Debug("Failed to get status: %s", err)
+		return fmt.Errorf("failed to get status: %w", err)
+	}
+
+	// Do lookups for all peers with a hint of parallelism for speed!
+	const maxParallel = 8
+	sema := make(chan struct{}, maxParallel)
+	var wg sync.WaitGroup
+	var mu sync.Mutex
+
+	nodeMap := make(map[key.NodePublic]*ipnstate.PeerStatus)
+	nodeMap[stat.Self.PublicKey] = stat.Self
+	for nodeKey, peer := range stat.Peer {
+		if peer == nil {
+			log.Debug("Skipping nil peer for node key: %s", nodeKey)
+			continue
+		}
+
+		nodeMap[nodeKey] = peer
+	}
+
+	for nodeKey, peer := range nodeMap {
+		idBytes, err := nodeKey.MarshalText()
+		if err != nil {
+			log.Debug("Failed to marshal node key: %s", err)
+			continue
+		}
+
+		nodeID := string(idBytes)
+		wg.Add(1)
+		sema <- struct{}{}
+
+		go func() {
+			defer wg.Done()
+			defer func() { <-sema }()
+
+			wctx, cancel := context.WithTimeout(ctx, 3*time.Second)
+			defer cancel()
+
+			ip := peer.TailscaleIPs[0].String()
+			if len(ip) == 0 {
+				log.Debug("Peer %s has no Tailscale IPs", nodeID)
+				return
+			}
+
+			whois, err := s.Lc.WhoIs(wctx, ip)
+			if err != nil {
+				log.Debug("WhoIs failed for %s (%s): %s", nodeID, ip, err)
+				return
+			}
+
+			if whois == nil || whois.Node == nil {
+				log.Debug("WhoIs returned nil node for %s (%s)", nodeID, ip)
+				return
+			}
+
+			data, err := json.Marshal(whois.Node.Hostinfo)
+			if err != nil {
+				log.Debug("Failed to marshal hostinfo for %s (%s): %s", nodeID, ip, err)
+				return
+			}
+
+			mu.Lock()
+			fmt.Println("HOSTINFO " + nodeID + " " + string(data))
+			mu.Unlock()
+		}()
+	}
+
+	wg.Wait()
+	return nil
+}