inspect: resurrect the inspect command (tendermint#9655)

resurrect the inspect command from tendermint#6785

Co-authored-by: Sam Kleinman <garen@tychoish.com>
Co-authored-by: Thane Thomson <connect@thanethomson.com>
Co-authored-by: Callum Waters <cmwaters19@gmail.com>

CHANGELOG_PENDING.md

@@ -12,6 +12,8 @@
 
 - Go API
   - [p2p] \#9625 Remove unused p2p/trust package (@cmwaters)
+  - [rpc] \#9655 Remove global environment and replace with constructor. (@williambanfield,@tychoish)
+  - [node] \#9655 Move DBContext and DBProvider from the node package to the config package. (@williambanfield,@tychoish)
 
 - Blockchain Protocol
 
@@ -22,6 +24,7 @@
   - [tools/tm-signer-harness] \#6498 Set OS home dir to instead of the hardcoded PATH. (@JayT106)
   - [metrics] \#9682 move state-syncing and block-syncing metrics to their respective packages (@cmwaters)
     labels have moved from block_syncing -> blocksync_syncing and state_syncing -> statesync_syncing
+  - [inspect] \#9655 Add a new `inspect` command for introspecting the state and block store of a crashed tendermint node. (@williambanfield)
 
 ### FEATURES
 

cmd/tendermint/commands/inspect.go

@@ -0,0 +1,87 @@
+package commands
+
+import (
+	"context"
+	"os"
+	"os/signal"
+	"syscall"
+
+	"github.com/spf13/cobra"
+
+	cfg "github.com/tendermint/tendermint/config"
+	"github.com/tendermint/tendermint/inspect"
+	"github.com/tendermint/tendermint/state"
+	"github.com/tendermint/tendermint/state/indexer/block"
+	"github.com/tendermint/tendermint/store"
+	"github.com/tendermint/tendermint/types"
+)
+
+// InspectCmd is the command for starting an inspect server.
+var InspectCmd = &cobra.Command{
+	Use:   "inspect",
+	Short: "Run an inspect server for investigating Tendermint state",
+	Long: `
+	inspect runs a subset of Tendermint's RPC endpoints that are useful for debugging
+	issues with Tendermint.
+
+	When the Tendermint consensus engine detects inconsistent state, it will crash the
+	Tendermint process. Tendermint will not start up while in this inconsistent state. 
+	The inspect command can be used to query the block and state store using Tendermint
+	RPC calls to debug issues of inconsistent state.
+	`,
+
+	RunE: runInspect,
+}
+
+func init() {
+	InspectCmd.Flags().
+		String("rpc.laddr",
+			config.RPC.ListenAddress, "RPC listenener address. Port required")
+	InspectCmd.Flags().
+		String("db-backend",
+			config.DBBackend, "database backend: goleveldb | cleveldb | boltdb | rocksdb | badgerdb")
+	InspectCmd.Flags().
+		String("db-dir", config.DBPath, "database directory")
+}
+
+func runInspect(cmd *cobra.Command, args []string) error {
+	ctx, cancel := context.WithCancel(cmd.Context())
+	defer cancel()
+
+	c := make(chan os.Signal, 1)
+	signal.Notify(c, syscall.SIGTERM, syscall.SIGINT)
+	go func() {
+		<-c
+		cancel()
+	}()
+
+	blockStoreDB, err := cfg.DefaultDBProvider(&cfg.DBContext{ID: "blockstore", Config: config})
+	if err != nil {
+		return err
+	}
+	blockStore := store.NewBlockStore(blockStoreDB)
+	defer blockStore.Close()
+
+	stateDB, err := cfg.DefaultDBProvider(&cfg.DBContext{ID: "state", Config: config})
+	if err != nil {
+		return err
+	}
+	stateStore := state.NewStore(stateDB, state.StoreOptions{DiscardABCIResponses: false})
+	defer stateStore.Close()
+
+	genDoc, err := types.GenesisDocFromFile(config.GenesisFile())
+	if err != nil {
+		return err
+	}
+	txIndexer, blockIndexer, err := block.IndexerFromConfig(config, cfg.DefaultDBProvider, genDoc.ChainID)
+	if err != nil {
+		return err
+	}
+	ins := inspect.New(config.RPC, blockStore, stateStore, txIndexer, blockIndexer, logger)
+
+	logger.Info("starting inspect server")
+	if err := ins.Run(ctx); err != nil {
+		return err
+	}
+	return nil
+}

cmd/tendermint/main.go

@@ -30,6 +30,7 @@ func main() {
 		cmd.VersionCmd,
 		cmd.RollbackStateCmd,
 		cmd.CompactGoLevelDBCmd,
+		cmd.InspectCmd,
 		debug.DebugCmd,
 		cli.NewCompletionCmd(rootCmd, true),
 	)

config/db.go

@@ -0,0 +1,30 @@
+package config
+
+import (
+	"context"
+
+	dbm "github.com/tendermint/tm-db"
+
+	"github.com/tendermint/tendermint/libs/log"
+	"github.com/tendermint/tendermint/libs/service"
+)
+
+// ServiceProvider takes a config and a logger and returns a ready to go Node.
+type ServiceProvider func(context.Context, *Config, log.Logger) (service.Service, error)
+
+// DBContext specifies config information for loading a new DB.
+type DBContext struct {
+	ID     string
+	Config *Config
+}
+
+// DBProvider takes a DBContext and returns an instantiated DB.
+type DBProvider func(*DBContext) (dbm.DB, error)
+
+// DefaultDBProvider returns a database using the DBBackend and DBDir
+// specified in the Config.
+func DefaultDBProvider(ctx *DBContext) (dbm.DB, error) {
+	dbType := dbm.BackendType(ctx.Config.DBBackend)
+
+	return dbm.NewDB(ctx.ID, dbType, ctx.Config.DBDir())
+}

consensus/replay.go

@@ -54,7 +54,7 @@ func (cs *State) readReplayMessage(msg *TimedWALMessage, newStepSub types.Subscr
 				if m.Height != m2.Height || m.Round != m2.Round || m.Step != m2.Step {
 					return fmt.Errorf("roundState mismatch. Got %v; Expected %v", m2, m)
 				}
-			case <-newStepSub.Cancelled():
+			case <-newStepSub.Canceled():
 				return fmt.Errorf("failed to read off newStepSub.Out(). newStepSub was canceled")
 			case <-ticker:
 				return fmt.Errorf("failed to read off newStepSub.Out()")

consensus/replay_test.go

@@ -97,7 +97,7 @@ func startNewStateAndWaitForBlock(t *testing.T, consensusReplayConfig *cfg.Confi
 	require.NoError(t, err)
 	select {
 	case <-newBlockSub.Out():
-	case <-newBlockSub.Cancelled():
+	case <-newBlockSub.Canceled():
 		t.Fatal("newBlockSub was canceled")
 	case <-time.After(120 * time.Second):
 		t.Fatal("Timed out waiting for new block (see trace above)")
@@ -1198,6 +1198,7 @@ func (bs *mockBlockStore) PruneBlocks(height int64, state sm.State) (uint64, int
 }
 
 func (bs *mockBlockStore) DeleteLatestBlock() error { return nil }
+func (bs *mockBlockStore) Close() error             { return nil }
 
 //---------------------------------------
 // Test handshake/init chain

docs/tools/debugging.md

@@ -55,3 +55,47 @@ given destination directory. Each archive will contain:
 
 Note: goroutine.out and heap.out will only be written if a profile address is
 provided and is operational. This command is blocking and will log any error.
+
+## Tendermint Inspect
+
+Tendermint includes an `inspect` command for querying Tendermint's state store and block
+store over Tendermint RPC.
+
+When the Tendermint consensus engine detects inconsistent state, it will crash the
+entire Tendermint process. 
+While in this inconsistent state, a node running Tendermint's consensus engine will not start up. 
+The `inspect` command runs only a subset of Tendermint's RPC endpoints for querying the block store
+and state store. 
+`inspect` allows operators to query a read-only view of the stage.
+`inspect` does not run the consensus engine at all and can therefore be used to debug
+processes that have crashed due to inconsistent state. 
+
+### Running inspect
+
+Start up the `inspect` tool on the machine where Tendermint crashed using: 
+```bash
+tendermint inspect --home=</path/to/app.d>
+```
+
+`inspect` will use the data directory specified in your Tendermint configuration file.
+`inspect` will also run the RPC server at the address specified in your Tendermint configuration file.
+
+### Using inspect
+
+With the `inspect` server running, you can access RPC endpoints that are critically important
+for debugging.
+Calling the `/status`, `/consensus_state` and `/dump_consensus_state` RPC endpoint 
+will return useful information about the Tendermint consensus state.
+
+To start the `inspect` process, run
+```bash
+tendermint inspect
+```
+
+### RPC endpoints
+
+The list of available RPC endpoints can be found by making a request to the RPC port.
+For an `inspect` process running on `127.0.0.1:26657`, navigate your browser to 
+`http://127.0.0.1:26657/` to retrieve the list of enabled RPC endpoints.
+
+Additional information on the Tendermint RPC endpoints can be found in the [rpc documentation](https://docs.tendermint.com/master/rpc).

go.sum

@@ -1361,6 +1361,7 @@ golang.org/x/sync v0.0.0-20200317015054-43a5402ce75a/go.mod h1:RxMgew5VJxzue5/jJ
 golang.org/x/sync v0.0.0-20200625203802-6e8e738ad208/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20201207232520-09787c993a3a/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
+golang.org/x/sync v0.0.0-20210220032951-036812b2e83c h1:5KslGYwFpkhGh+Q16bwMP3cOontH8FOep7tGV86Y7SQ=
 golang.org/x/sync v0.0.0-20210220032951-036812b2e83c/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20220722155255-886fb9371eb4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.1.0 h1:wsuoTGHzEhffawBOhz5CYhcrV4IdKZbEyZjBMuTp12o=

inspect/doc.go

@@ -0,0 +1,36 @@
+/*
+Package inspect provides a tool for investigating the state of a
+failed Tendermint node.
+
+This package provides the Inspector type. The Inspector type runs a subset of the Tendermint
+RPC endpoints that are useful for debugging issues with Tendermint consensus.
+
+When a node running the Tendermint consensus engine detects an inconsistent consensus state,
+the entire node will crash. The Tendermint consensus engine cannot run in this
+inconsistent state so the node will not be able to start up again.
+
+The RPC endpoints provided by the Inspector type allow for a node operator to inspect
+the block store and state store to better understand what may have caused the inconsistent state.
+
+The Inspector type's lifecycle is controlled by a context.Context
+
+	ins := inspect.NewFromConfig(rpcConfig)
+	ctx, cancelFunc:= context.WithCancel(context.Background())
+
+	// Run blocks until the Inspector server is shut down.
+	go ins.Run(ctx)
+	...
+
+	// calling the cancel function will stop the running inspect server
+	cancelFunc()
+
+Inspector serves its RPC endpoints on the address configured in the RPC configuration
+
+	rpcConfig.ListenAddress = "tcp://127.0.0.1:26657"
+	ins := inspect.NewFromConfig(rpcConfig)
+	go ins.Run(ctx)
+
+The list of available RPC endpoints can then be viewed by navigating to
+http://127.0.0.1:26657/ in the web browser.
+*/
+package inspect