Diag session routing (erigontech#8232)

Code to support react based UI for diagnostics:

* pprof, prometheus and diagnistics rationalized to use a single router
(i.e. they can all run in the same port)
* support_cmd updated to support node routing (was only first node)
* Multi content support in router tunnel (application/octet-stream &
appliaction/json)
* Routing requests changed from using http forms to rest + query params
* REST query requests can now be made against erigon base port and
diagnostics with the same url format/params

---------

Co-authored-by: dvovk <[email protected]>
Co-authored-by: Mark Holt <[email protected]>

Author: 3 people

cmd/abigen/main.go

@@ -19,13 +19,14 @@ package main
 import (
 	"encoding/json"
 	"fmt"
-	"github.com/ledgerwatch/erigon-lib/common"
 	"io"
 	"os"
 	"path/filepath"
 	"regexp"
 	"strings"
 
+	"github.com/ledgerwatch/erigon-lib/common"
+
 	"github.com/ledgerwatch/log/v3"
 	"github.com/urfave/cli/v2"
 

cmd/caplin-phase1/main.go

@@ -60,7 +60,7 @@ func runCaplinNode(cliCtx *cli.Context) error {
 	if err != nil {
 		log.Error("[Phase1] Could not initialize caplin", "err", err)
 	}
-	if _, err := debug.Setup(cliCtx, true /* root logger */); err != nil {
+	if _, _, err := debug.Setup(cliCtx, true /* root logger */); err != nil {
 		return err
 	}
 	log.Root().SetHandler(log.LvlFilterHandler(log.Lvl(cfg.LogLvl), log.StderrHandler))

cmd/devnet/devnet/node.go

@@ -4,12 +4,14 @@ import (
 	context "context"
 	"fmt"
 	"math/big"
+	"net/http"
 	"sync"
 
 	"github.com/c2h5oh/datasize"
 	"github.com/ledgerwatch/erigon/cmd/devnet/accounts"
 	"github.com/ledgerwatch/erigon/cmd/devnet/args"
 	"github.com/ledgerwatch/erigon/cmd/devnet/requests"
+	"github.com/ledgerwatch/erigon/diagnostics"
 	"github.com/ledgerwatch/erigon/eth/ethconfig"
 	"github.com/ledgerwatch/erigon/node/nodecfg"
 	"github.com/ledgerwatch/erigon/params"
@@ -129,6 +131,7 @@ func (n *node) ChainID() *big.Int {
 func (n *node) run(ctx *cli.Context) error {
 	var logger log.Logger
 	var err error
+	var metricsMux *http.ServeMux
 
 	defer n.done()
 	defer func() {
@@ -141,7 +144,7 @@ func (n *node) run(ctx *cli.Context) error {
 		n.Unlock()
 	}()
 
-	if logger, err = debug.Setup(ctx, false /* rootLogger */); err != nil {
+	if logger, metricsMux, err = debug.Setup(ctx, false /* rootLogger */); err != nil {
 		return err
 	}
 
@@ -166,6 +169,10 @@ func (n *node) run(ctx *cli.Context) error {
 
 	n.ethNode, err = enode.New(n.nodeCfg, n.ethCfg, logger)
 
+	if metricsMux != nil {
+		diagnostics.Setup(ctx, metricsMux, n.ethNode)
+	}
+
 	n.Lock()
 	if n.startErr != nil {
 		n.startErr <- err

cmd/devnet/main.go

@@ -107,7 +107,7 @@ var (
 	}
 
 	metricsURLsFlag = cli.StringSliceFlag{
-		Name:  "metrics.urls",
+		Name:  "debug.urls",
 		Usage: "internal flag",
 	}
 
@@ -199,7 +199,7 @@ func action(ctx *cli.Context) error {
 	if metrics {
 		// TODO should get this from the network as once we have multiple nodes we'll need to iterate the
 		// nodes and create a series of urls - for the moment only one is supported
-		ctx.Set("metrics.urls", fmt.Sprintf("http://localhost:%d/debug/metrics/", ctx.Int("metrics.port")))
+		ctx.Set("metrics.urls", fmt.Sprintf("http://localhost:%d/debug/", ctx.Int("metrics.port")))
 	}
 
 	// start the network with each node in a go routine

cmd/erigon/main.go

@@ -3,12 +3,14 @@ package main
 import (
 	"errors"
 	"fmt"
+	"net/http"
 	"os"
 	"path/filepath"
 	"reflect"
 	"strings"
 
 	"github.com/ledgerwatch/erigon-lib/common/dbg"
+	"github.com/ledgerwatch/erigon/diagnostics"
 	"github.com/ledgerwatch/erigon/metrics"
 	"github.com/ledgerwatch/log/v3"
 	"github.com/pelletier/go-toml"
@@ -55,7 +57,9 @@ func runErigon(cliCtx *cli.Context) error {
 
 	var logger log.Logger
 	var err error
-	if logger, err = debug.Setup(cliCtx, true /* root logger */); err != nil {
+	var metricsMux *http.ServeMux
+
+	if logger, metricsMux, err = debug.Setup(cliCtx, true /* root logger */); err != nil {
 		return err
 	}
 
@@ -73,6 +77,11 @@ func runErigon(cliCtx *cli.Context) error {
 		log.Error("Erigon startup", "err", err)
 		return err
 	}
+
+	if metricsMux != nil {
+		diagnostics.Setup(cliCtx, metricsMux, ethNode)
+	}
+
 	err = ethNode.Serve()
 	if err != nil {
 		log.Error("error while serving an Erigon node", "err", err)

cmd/utils/flags.go

@@ -813,10 +813,27 @@ var (
 		Usage: "Max allowed page size for search methods",
 		Value: 25,
 	}
+
+	DiagnosticsURLFlag = cli.StringFlag{
+		Name:  "diagnostics.url",
+		Usage: "URL of the diagnostics system provided by the support team",
+	}
+
+	DiagnosticsInsecureFlag = cli.BoolFlag{
+		Name:  "diagnostics.insecure",
+		Usage: "Allows communication with diagnostics system using self-signed TLS certificates",
+	}
+
+	DiagnosticsSessionsFlag = cli.StringSliceFlag{
+		Name:  "diagnostics.ids",
+		Usage: "Comma separated list of support session ids to connect to",
+	}
 )
 
 var MetricFlags = []cli.Flag{&MetricsEnabledFlag, &MetricsHTTPFlag, &MetricsPortFlag}
 
+var DiagnosticsFlags = []cli.Flag{&DiagnosticsURLFlag, &DiagnosticsURLFlag, &DiagnosticsSessionsFlag}
+
 // setNodeKey loads a node key from command line flags if provided,
 // otherwise it tries to load it from datadir,
 // otherwise it generates a new key in datadir.

cmd/utils/flags/flags.go

@@ -305,6 +305,10 @@ func (b *bigValue) Set(s string) error {
 	return nil
 }
 
+func (b *bigValue) Get() any {
+	return b.String()
+}
+
 // GlobalBig returns the value of a BigFlag from the global flag set.
 func GlobalBig(ctx *cli.Context, name string) *big.Int {
 	val := ctx.Generic(name)

diagnostics/block_body_download.go renamed to diagnostics/block_body_download_stats.go

@@ -10,7 +10,7 @@ import (
 )
 
 func SetupBlockBodyDownload(metricsMux *http.ServeMux) {
-	metricsMux.HandleFunc("/debug/metrics/block_body_download", func(w http.ResponseWriter, r *http.Request) {
+	metricsMux.HandleFunc("/block_body_download", func(w http.ResponseWriter, r *http.Request) {
 		w.Header().Set("Access-Control-Allow-Origin", "*")
 		writeBlockBodyDownload(w, r)
 	})

diagnostics/cmd_line.go

@@ -1,22 +1,29 @@
 package diagnostics
 
 import (
-	"fmt"
-	"io"
 	"net/http"
 	"os"
+	"strconv"
+	"strings"
 )
 
 func SetupCmdLineAccess(metricsMux *http.ServeMux) {
-	metricsMux.HandleFunc("/debug/metrics/cmdline", func(w http.ResponseWriter, r *http.Request) {
+	metricsMux.HandleFunc("/cmdline", func(w http.ResponseWriter, r *http.Request) {
 		w.Header().Set("Access-Control-Allow-Origin", "*")
-		writeCmdLine(w)
-	})
-}
+		w.Header().Set("Content-Type", "application/json")
+
+		var space []byte
 
-func writeCmdLine(w io.Writer) {
-	fmt.Fprintf(w, "SUCCESS\n")
-	for _, arg := range os.Args {
-		fmt.Fprintf(w, "%s\n", arg)
-	}
+		w.Write([]byte{'"'})
+		for _, arg := range os.Args {
+			if len(space) > 0 {
+				w.Write(space)
+			} else {
+				space = []byte(" ")
+			}
+
+			w.Write([]byte(strings.Trim(strconv.Quote(arg), `"`)))
+		}
+		w.Write([]byte{'"'})
+	})
 }