depot/packages/networking/ipfs-cluster/cmd/ipfs-cluster-ctl/main.go

// The ipfs-cluster-ctl application.
package main

import (
	"context"
	"encoding/hex"
	"errors"
	"fmt"
	"io"
	"os"
	"strings"
	"sync"
	"time"

	"github.com/ipfs-cluster/ipfs-cluster/api"
	"github.com/ipfs-cluster/ipfs-cluster/api/rest/client"

	logging "github.com/ipfs/go-log/v2"
	peer "github.com/libp2p/go-libp2p/core/peer"
	ma "github.com/multiformats/go-multiaddr"

	uuid "github.com/google/uuid"
	cli "github.com/urfave/cli"
)

const programName = `ipfs-cluster-ctl`

// Version is the cluster-ctl tool version. It should match
// the IPFS cluster's version
const Version = "1.0.2"

var (
	defaultHost          = "/ip4/127.0.0.1/tcp/9094"
	defaultTimeout       = 0
	defaultWaitCheckFreq = time.Second
	defaultAddParams     = api.DefaultAddParams()
)

var logger = logging.Logger("cluster-ctl")

var globalClient client.Client

// Description provides a short summary of the functionality of this tool
var Description = fmt.Sprintf(`
%s is a tool to manage IPFS Cluster nodes.

Use "%s help" to list all available commands and
"%s help <command>" to get usage information for a
specific one.

%s uses the IPFS Cluster API to perform requests and
display responses in a user-readable format. The location of the IPFS
Cluster server is assumed to be %s, but can be configured
with the --host option. If several multiaddresses are specified 
(comma-separated), requests will be sent to the first one and fail-over 
over to the others. This also works for dns-based addresses which resolve
to multiple values.

To use the secure libp2p-http API endpoint, use "--host" with 
the full cluster libp2p listener address, including the "/p2p/<peerID>"
part, or a /dnsaddr that resolves to it. Provide the cluster secret with 
--secret as needed.

For feedback, bug reports or any additional information, visit
https://github.com/ipfs-cluster/ipfs-cluster.
`,
	programName,
	programName,
	programName,
	programName,
	defaultHost)

var (
	waitFlagDesc        = "Wait for the pin to reach the minimum replication factor before returning"
	waitTimeoutFlagDesc = "How long to --wait (in seconds). Default: forever"
)

// type peerAddBody struct {
// 	Addr string `json:"peer_multiaddress"`
// }

func out(m string, a ...interface{}) {
	fmt.Fprintf(os.Stderr, m, a...)
}

func checkErr(doing string, err error) {
	if err != nil {
		out("error %s: %s\n", doing, err)
		os.Exit(1)
	}
}

func main() {
	ctx := context.Background()

	app := cli.NewApp()
	app.Name = programName
	app.Usage = "CLI for IPFS Cluster"
	app.Description = Description
	app.Version = Version
	app.Flags = []cli.Flag{
		cli.StringFlag{
			Name:  "host, l",
			Value: defaultHost,
			Usage: `API endpoint multiaddresses (comma-separated)`,
		},
		cli.StringFlag{
			Name:  "secret",
			Value: "",
			Usage: "cluster secret (32 byte pnet-key) as needed. Only when using the LibP2P endpoint",
		},
		cli.BoolFlag{
			Name:  "https, s",
			Usage: "use https to connect to the API",
		},
		cli.BoolFlag{
			Name:  "no-check-certificate",
			Usage: "do not verify server TLS certificate. only valid with --https flag",
		},
		cli.StringFlag{
			Name:  "encoding, enc",
			Value: "text",
			Usage: "output format encoding [text, json]",
		},
		cli.IntFlag{
			Name:  "timeout, t",
			Value: defaultTimeout,
			Usage: "number of seconds to wait before timing out a request",
		},
		cli.BoolFlag{
			Name:  "debug, d",
			Usage: "set debug log level",
		},
		cli.StringFlag{
			Name: "basic-auth",
			Usage: `<username>[:<password>] specify BasicAuth credentials for server that
requires authorization. implies --https, which you can disable with --force-http`,
			EnvVar: "CLUSTER_CREDENTIALS",
		},
		cli.BoolFlag{
			Name:  "force-http, f",
			Usage: "force HTTP. only valid when using BasicAuth",
		},
	}

	app.Before = func(c *cli.Context) error {
		cfg := &client.Config{}

		if c.Bool("debug") {
			logging.SetLogLevel("cluster-ctl", "debug")
			logging.SetLogLevel("apitypes", "debug")
			cfg.LogLevel = "debug"
			logger.Debug("debug level enabled")
		}

		if hexSecret := c.String("secret"); hexSecret != "" {
			secret, err := hex.DecodeString(hexSecret)
			checkErr("parsing secret", err)
			cfg.ProtectorKey = secret
		}

		cfg.Timeout = time.Duration(c.Int("timeout")) * time.Second

		cfg.SSL = c.Bool("https")
		cfg.NoVerifyCert = c.Bool("no-check-certificate")
		user, pass := parseCredentials(c.String("basic-auth"))
		cfg.Username = user
		cfg.Password = pass
		if user != "" && !cfg.SSL && !c.Bool("force-http") {
			logger.Warn("SSL automatically enabled with basic auth credentials. Set \"force-http\" to disable")
			cfg.SSL = true
		}

		enc := c.String("encoding")
		if enc != "text" && enc != "json" {
			checkErr("", errors.New("unsupported encoding"))
		}

		var configs []*client.Config
		var err error
		for _, addr := range strings.Split(c.String("host"), ",") {
			multiaddr, err := ma.NewMultiaddr(addr)
			checkErr("parsing host multiaddress", err)

			if client.IsPeerAddress(multiaddr) && c.Bool("https") {
				logger.Warn("Using libp2p-http for %s. The https flag will be ignored for this connection", addr)
			}

			var cfgs []*client.Config

			// We can auto round-robin on DNS records when using
			// libp2p-http or not using SSL. When using SSL we
			// cannot use the resolve-IPs directly.
			if client.IsPeerAddress(multiaddr) || !cfg.SSL {
				cfgs, err = cfg.AsTemplateForResolvedAddress(ctx, multiaddr)
			} else {
				cfgs = cfg.AsTemplateFor([]ma.Multiaddr{multiaddr})
			}
			checkErr("creating configs", err)
			configs = append(configs, cfgs...)
		}

		retries := len(configs)
		globalClient, err = client.NewLBClient(&client.Failover{}, configs, retries)
		checkErr("creating API client", err)

		// TODO: need to figure out best way to configure tracing for ctl
		// leaving the following as it is still useful for local debugging.
		// tracingCfg := &observations.Config{}
		// tracingCfg.Default()
		// tracingCfg.EnableTracing = true
		// tracingCfg.TracingServiceName = "cluster-ctl"
		// tracingCfg.TracingSamplingProb = 1
		// tracer = observations.SetupTracing(tracingCfg)
		return nil
	}
	app.After = func(c *cli.Context) error {
		// TODO: need to figure out best way to configure tracing for ctl
		// leaving the following as it is still useful for local debugging.
		// tracer.Flush()
		return nil
	}

	app.Commands = []cli.Command{
		{
			Name:  "id",
			Usage: "Retrieve peer information",
			Description: `
This command displays information about the peer that the tool is contacting
(usually running in localhost).
`,
			Flags: []cli.Flag{},
			Action: func(c *cli.Context) error {
				resp, cerr := globalClient.ID(ctx)
				formatResponse(c, resp, cerr)
				return nil
			},
		},
		{
			Name:        "peers",
			Usage:       "List and manage IPFS Cluster peers",
			Description: "List and manage IPFS Cluster peers",
			Subcommands: []cli.Command{
				{
					Name:  "ls",
					Usage: "list the nodes participating in the IPFS Cluster",
					Description: `
This command provides a list of the ID information of all the peers in the Cluster.
`,
					Flags:     []cli.Flag{},
					ArgsUsage: " ",
					Action: func(c *cli.Context) error {
						out := make(chan api.ID, 1024)
						errCh := make(chan error, 1)
						go func() {
							defer close(errCh)
							errCh <- globalClient.Peers(ctx, out)
						}()
						formatResponse(c, out, nil)
						err := <-errCh
						formatResponse(c, nil, err)
						return nil
					},
				},
				{
					Name:  "rm",
					Usage: "remove a peer from the Cluster",
					Description: `
This command removes a peer from the cluster. If the peer is online, it will
automatically shut down. All other cluster peers should be online for the
operation to succeed, otherwise some nodes may be left with an outdated list of
cluster peers.
`,
					ArgsUsage: "<peer ID>",
					Flags:     []cli.Flag{},
					Action: func(c *cli.Context) error {
						pid := c.Args().First()
						p, err := peer.Decode(pid)
						checkErr("parsing peer ID", err)
						cerr := globalClient.PeerRm(ctx, p)
						formatResponse(c, nil, cerr)
						return nil
					},
				},
			},
		},
		{
			Name:      "add",
			Usage:     "Add a file or directory to ipfs and pin it in the cluster",
			ArgsUsage: "<path>",
			Description: `
Add allows to add and replicate content to several ipfs daemons, performing
a Cluster Pin operation on success. It takes elements from local paths as
well as from web URLs (accessed with a GET request). Providing several
arguments will automatically set --wrap-in-directory.

Cluster "add" works, by default, just like "ipfs add" and has similar options
in terms of DAG layout, chunker, hash function etc. It also supports adding
CAR files directly (--format car), as long as they have a single root. When
adding CAR files, all the options related to dag-building are ignored.

Added content will be allocated and sent block by block to the peers that
should pin it (among which may not necessarily be the local ipfs daemon).
Once all the blocks have arrived, they will be "cluster-pinned". This makes 
cluster add slower than a local ipfs add, but the result is a fully replicated
on completion. If you prefer faster adding, use the --local flag to add 
directly to the local IPFS node and pin in the destinations after that. 
Note that the local IPFS node may not be among the destinations, which will
leave the unpinned content in it.

Optional replication-min and replication-max factors can be provided: -1 means
"pin everywhere" and 0 means use cluster's default setting (i.e., replication
factor set in config). Positive values indicate how many peers should pin this
content.
`,
			/*
				Cluster Add supports handling huge files and sharding the resulting DAG among
				several ipfs daemons (--shard). In this case, a single ipfs daemon will not
				contain the full dag, but only parts of it (shards). Desired shard size can
				be provided with the --shard-size flag.

				We recommend setting a --name for sharded pins. Otherwise, it will be
				automatically generated.
			*/
			Flags: []cli.Flag{
				cli.BoolFlag{
					Name:  "recursive, r",
					Usage: "Add directory paths recursively",
				},
				cli.BoolFlag{
					Name:  "quiet, q",
					Usage: "Write only hashes to output (one per line)",
				},
				cli.BoolFlag{
					Name:  "quieter, Q",
					Usage: "Write only final hash to output",
				},
				cli.BoolFlag{
					Name:  "no-stream",
					Usage: "Buffer output locally. Produces a valid JSON array with --enc=json.",
				},
				cli.BoolFlag{
					Name:  "local",
					Usage: "Add to local peer but pin normally",
				},
				cli.StringFlag{
					Name:  "name, n",
					Value: defaultAddParams.Name,
					Usage: "Sets a name for this pin",
				},
				cli.IntFlag{
					Name:  "replication-min, rmin",
					Value: defaultAddParams.ReplicationFactorMin,
					Usage: "Sets the minimum replication factor for pinning this file",
				},
				cli.IntFlag{
					Name:  "replication-max, rmax",
					Value: defaultAddParams.ReplicationFactorMax,
					Usage: "Sets the maximum replication factor for pinning this file",
				},
				cli.StringFlag{
					Name:  "expire-in",
					Usage: "Duration after which the pin should be unpinned automatically",
				},
				cli.StringSliceFlag{
					Name:  "metadata",
					Usage: "Pin metadata: key=value. Can be added multiple times",
				},
				cli.StringFlag{
					Name:  "allocations, allocs",
					Usage: "Optional comma-separated list of peer IDs",
				},
				cli.BoolFlag{
					Name:  "wait",
					Usage: waitFlagDesc,
				},
				cli.DurationFlag{
					Name:  "wait-timeout, wt",
					Value: 0,
					Usage: waitTimeoutFlagDesc,
				},

				cli.BoolFlag{
					Name:  "wrap-with-directory, w",
					Usage: "Wrap a with a directory object",
				},

				cli.StringFlag{
					Name:  "format",
					Value: defaultAddParams.Format,
					Usage: "'unixfs' (add as unixfs DAG), 'car' (import CAR file)",
				},

				cli.StringFlag{
					Name:  "layout",
					Value: defaultAddParams.Layout,
					Usage: "Dag layout to use for dag generation: balanced or trickle",
				},
				cli.BoolFlag{
					Name:  "hidden, H",
					Usage: "Include files that are hidden.  Only takes effect on recursive add",
				},
				cli.StringFlag{
					Name:  "chunker, s",
					Usage: "'size-<size>' or 'rabin-<min>-<avg>-<max>'",
					Value: defaultAddParams.Chunker,
				},
				cli.BoolFlag{
					Name:  "raw-leaves",
					Usage: "Use raw blocks for leaves (experimental)",
				},
				cli.IntFlag{
					Name:  "cid-version",
					Usage: "CID version. Non default implies raw-leaves",
					Value: defaultAddParams.CidVersion,
				},
				cli.StringFlag{
					Name:  "hash",
					Usage: "Hash function to use. Implies cid-version=1",
					Value: defaultAddParams.HashFun,
				},
				cli.BoolFlag{
					Name:  "nocopy",
					Usage: "Add the URL using filestore. Implies raw-leaves. (experimental)",
				},

				// TODO: Uncomment when sharding is supported.
				// cli.BoolFlag{
				//	Name:  "shard",
				//	Usage: "Break the file into pieces (shards) and distributed among peers",
				// },
				// cli.Uint64Flag{
				//	Name:  "shard-size",
				//	Value: defaultAddParams.ShardSize,
				//	Usage: "Sets the maximum replication factor for pinning this file",
				// },
				// TODO: Figure progress over total bar.
				// cli.BoolFlag{
				//	Name:  "progress, p",
				//	Usage: "Stream progress data",
				// },

			},
			Action: func(c *cli.Context) error {
				shard := c.Bool("shard")
				name := c.String("name")
				if shard && name == "" {
					randName, err := uuid.NewRandom()
					if err != nil {
						return err
					}
					// take only first letters
					name = "sharded-" + strings.Split(randName.String(), "-")[0]
				}

				// Read arguments (paths)
				paths := make([]string, c.NArg())
				for i, path := range c.Args() {
					paths[i] = path
				}

				if len(paths) == 0 {
					checkErr("", errors.New("need at least one path"))
				}

				// Setup AddParams
				p := api.DefaultAddParams()
				p.ReplicationFactorMin = c.Int("replication-min")
				p.ReplicationFactorMax = c.Int("replication-max")
				if expireIn := c.String("expire-in"); expireIn != "" {
					d, err := time.ParseDuration(expireIn)
					checkErr("parsing expire-in", err)
					p.ExpireAt = time.Now().Add(d)
				}

				p.Metadata = parseMetadata(c.StringSlice("metadata"))
				p.Name = name
				if c.String("allocations") != "" {
					p.UserAllocations = api.StringsToPeers(strings.Split(c.String("allocations"), ","))
				}
				p.Format = c.String("format")
				//p.Shard = shard
				//p.ShardSize = c.Uint64("shard-size")
				p.Shard = false
				p.Recursive = c.Bool("recursive")
				p.Local = c.Bool("local")
				p.Layout = c.String("layout")
				p.Chunker = c.String("chunker")
				p.RawLeaves = c.Bool("raw-leaves")
				p.Hidden = c.Bool("hidden")
				p.Wrap = c.Bool("wrap-with-directory") || len(paths) > 1
				p.CidVersion = c.Int("cid-version")
				p.HashFun = c.String("hash")
				if p.HashFun != defaultAddParams.HashFun {
					p.CidVersion = 1
				}
				if p.CidVersion > 0 {
					p.RawLeaves = true
				}
				p.NoCopy = c.Bool("nocopy")
				if p.NoCopy {
					p.RawLeaves = true
				}

				// Prevent footgun
				if p.Wrap && p.Format == "car" {
					checkErr("", errors.New("only a single CAR file can be added and wrap-with-directory is not supported"))
				}

				out := make(chan api.AddedOutput, 1)
				var wg sync.WaitGroup
				wg.Add(1)
				go func() {
					defer wg.Done()

					var buffered []addedOutputQuiet
					var lastBuf addedOutputQuiet
					var qq = c.Bool("quieter")
					var q = c.Bool("quiet") || qq
					var bufferResults = c.Bool("no-stream")
					for v := range out {
						added := addedOutputQuiet{
							AddedOutput: v,
							quiet:       q,
						}
						lastBuf = added
						if bufferResults {
							buffered = append(buffered, added)
							continue
						}
						if !qq { // print things
							formatResponse(c, added, nil)
						}
					}
					if !lastBuf.AddedOutput.Cid.Defined() {
						return // no elements at all
					}
					if bufferResults { // we buffered.
						if qq { // [last elem]
							formatResponse(c, []addedOutputQuiet{lastBuf}, nil)
						} else { // [all elems]
							formatResponse(c, buffered, nil)
						}
					} else if qq { // we already printed unless Quieter
						formatResponse(c, lastBuf, nil)
					}
					if c.Bool("wait") {
						// In order to wait we need to get the allocation's replication factor.
						// If it errors, we use whatever we set on the request. If we set 0 or -1, then
						// no limit applies so we will wait for all.
						rplMin := p.ReplicationFactorMin
						alloc, err := globalClient.Allocation(ctx, lastBuf.AddedOutput.Cid)
						if err == nil {
							rplMin = alloc.ReplicationFactorMin
						}

						_, werr := waitFor(lastBuf.AddedOutput.Cid, api.TrackerStatusPinned, c.Duration("wait-timeout"), rplMin)
						checkErr("waiting for pin status", werr)
					}
				}()

				cerr := globalClient.Add(ctx, paths, p, out)
				wg.Wait()
				formatResponse(c, nil, cerr)
				return cerr
			},
		},
		{
			Name:        "pin",
			Usage:       "Pin and unpin and list items in IPFS Cluster",
			Description: "Pin and unpin and list items in IPFS Cluster",
			Subcommands: []cli.Command{
				{
					Name:  "add",
					Usage: "Pin an item in the cluster",
					Description: `
This command tells IPFS Cluster to start managing a CID. Depending on
the pinning strategy, this will trigger IPFS pin requests. The CID will
become part of the Cluster's state and will tracked from this point.

When the request has succeeded, the command returns the status of the CID
in the cluster and should be part of the list offered by "pin ls".

An optional replication factor can be provided: -1 means "pin everywhere"
and 0 means use cluster's default setting (i.e., replication factor set in
config). Positive values indicate how many peers should pin this content.

An optional allocations argument can be provided, allocations should be a
comma-separated list of peer IDs on which we want to pin. Peers in allocations
are prioritized over automatically-determined ones, but replication factors
would still be respected.
`,
					ArgsUsage: "<CID|Path>",
					Flags: []cli.Flag{
						cli.IntFlag{
							Name:  "replication, r",
							Value: 0,
							Usage: "Sets a custom replication factor (overrides -rmax and -rmin)",
						},
						cli.IntFlag{
							Name:  "replication-min, rmin",
							Value: 0,
							Usage: "Sets the minimum replication factor for this pin",
						},
						cli.IntFlag{
							Name:  "replication-max, rmax",
							Value: 0,
							Usage: "Sets the maximum replication factor for this pin",
						},
						cli.StringFlag{
							Name:  "allocations, allocs",
							Usage: "Optional comma-separated list of peer IDs",
						},
						cli.StringFlag{
							Name:  "name, n",
							Value: "",
							Usage: "Sets a name for this pin",
						},
						cli.StringFlag{
							Name:  "mode",
							Value: "recursive",
							Usage: "Select a way to pin: recursive or direct",
						},
						cli.StringFlag{
							Name:  "expire-in",
							Usage: "Duration after which pin should be unpinned automatically",
						},
						cli.StringSliceFlag{
							Name:  "metadata",
							Usage: "Pin metadata: key=value. Can be added multiple times",
						},
						cli.BoolFlag{
							Name:  "no-status, ns",
							Usage: "Prevents fetching pin status after pinning (faster, quieter)",
						},
						cli.BoolFlag{
							Name:  "wait, w",
							Usage: waitFlagDesc,
						},
						cli.DurationFlag{
							Name:  "wait-timeout, wt",
							Value: 0,
							Usage: waitTimeoutFlagDesc,
						},
					},
					Action: func(c *cli.Context) error {
						arg := c.Args().First()
						rpl := c.Int("replication")
						rplMin := c.Int("replication-min")
						rplMax := c.Int("replication-max")
						if rpl != 0 {
							rplMin = rpl
							rplMax = rpl
						}

						var userAllocs []peer.ID
						if c.String("allocations") != "" {
							allocs := strings.Split(c.String("allocations"), ",")
							for i := range allocs {
								allocs[i] = strings.TrimSpace(allocs[i])
							}
							userAllocs = api.StringsToPeers(allocs)
							if len(userAllocs) != len(allocs) {
								checkErr("decoding allocations", errors.New("some peer IDs could not be decoded"))
							}
						}
						var expireAt time.Time
						if expireIn := c.String("expire-in"); expireIn != "" {
							d, err := time.ParseDuration(expireIn)
							checkErr("parsing expire-in", err)
							expireAt = time.Now().Add(d)
						}

						opts := api.PinOptions{
							ReplicationFactorMin: rplMin,
							ReplicationFactorMax: rplMax,
							Name:                 c.String("name"),
							Mode:                 api.PinModeFromString(c.String("mode")),
							UserAllocations:      userAllocs,
							ExpireAt:             expireAt,
							Metadata:             parseMetadata(c.StringSlice("metadata")),
						}

						pin, cerr := globalClient.PinPath(ctx, arg, opts)
						if cerr != nil {
							formatResponse(c, nil, cerr)
							return nil
						}
						handlePinResponseFormatFlags(
							ctx,
							c,
							pin,
							api.TrackerStatusPinned,
						)
						return nil
					},
				},
				{
					Name:  "rm",
					Usage: "Unpin an item from the cluster",
					Description: `
This command tells IPFS Cluster to no longer manage a CID. This will
trigger unpinning operations in all the IPFS nodes holding the content.

When the request has succeeded, the command returns the status of the CID
in the cluster. The CID should disappear from the list offered by "pin ls",
although unpinning operations in the cluster may take longer or fail.
`,
					ArgsUsage: "<CID|Path>",
					Flags: []cli.Flag{
						cli.BoolFlag{
							Name:  "no-status, ns",
							Usage: "Prevents fetching pin status after unpinning (faster, quieter)",
						},
						cli.BoolFlag{
							Name:  "wait, w",
							Usage: waitFlagDesc,
						},
						cli.DurationFlag{
							Name:  "wait-timeout, wt",
							Value: 0,
							Usage: waitTimeoutFlagDesc,
						},
					},
					Action: func(c *cli.Context) error {
						arg := c.Args().First()
						pin, cerr := globalClient.UnpinPath(ctx, arg)
						if cerr != nil {
							formatResponse(c, nil, cerr)
							return nil
						}
						handlePinResponseFormatFlags(
							ctx,
							c,
							pin,
							api.TrackerStatusUnpinned,
						)
						return nil
					},
				},
				{
					Name:  "update",
					Usage: "Pin a new item based on an existing one",
					Description: `
This command will add a new pin to the cluster taking all the options from an
existing one, including name. This means that the new pin will bypass the
allocation process and will be allocated to the same peers as the existing
one.

The cluster peers will try to Pin the new item on IPFS using the "pin update"
command. This is especially efficient when the content of two pins (their DAGs)
are similar.

Unlike the "pin update" command in the ipfs daemon, this will not unpin the
existing item from the cluster. Please run "pin rm" for that.
`,
					ArgsUsage: "<existing-CID> <new-CID|Path>",
					Flags: []cli.Flag{
						cli.StringFlag{
							Name:  "name, n",
							Value: "",
							Usage: "Sets a name for this updated pin",
						},
						cli.StringFlag{
							Name:  "expire-in",
							Usage: "Duration after which the pin should be unpinned automatically after updating",
						},
						cli.BoolFlag{
							Name:  "no-status, ns",
							Usage: "Prevents fetching pin status after updating (faster, quieter)",
						},
						cli.BoolFlag{
							Name:  "wait, w",
							Usage: waitFlagDesc,
						},
						cli.DurationFlag{
							Name:  "wait-timeout, wt",
							Value: 0,
							Usage: waitTimeoutFlagDesc,
						},
					},
					Action: func(c *cli.Context) error {
						from := c.Args().Get(0)
						to := c.Args().Get(1)

						fromCid, err := api.DecodeCid(from)
						checkErr("parsing from Cid", err)

						var expireAt time.Time
						if expireIn := c.String("expire-in"); expireIn != "" {
							d, err := time.ParseDuration(expireIn)
							checkErr("parsing expire-in", err)
							expireAt = time.Now().Add(d)
						}

						opts := api.PinOptions{
							PinUpdate: fromCid,
							Name:      c.String("name"),
							ExpireAt:  expireAt,
						}

						pin, cerr := globalClient.PinPath(ctx, to, opts)
						if cerr != nil {
							formatResponse(c, nil, cerr)
							return nil
						}
						handlePinResponseFormatFlags(
							ctx,
							c,
							pin,
							api.TrackerStatusPinned,
						)
						return nil
					},
				},
				{
					Name:  "ls",
					Usage: "List items in the cluster pinset",
					Description: `
This command will list the CIDs which are tracked by IPFS Cluster and to
which peers they are currently allocated. This list does not include
any monitoring information about the IPFS status of the CIDs, it
merely represents the list of pins which are part of the shared state of
the cluster. For IPFS-status information about the pins, use "status".

The filter only takes effect when listing all pins. The possible values are:
  - all (default)
  - pin (normal pins, recursive or direct)
  - meta-pin (sharded pins)
  - clusterdag-pin (sharding-dag root pins)
  - shard-pin (individual shard pins)
`,
					ArgsUsage: "[CID]",
					Flags: []cli.Flag{
						cli.StringFlag{
							Name:  "filter",
							Usage: "Comma separated list of pin types. See help above.",
							Value: "all",
						},
					},
					Action: func(c *cli.Context) error {
						cidStr := c.Args().First()
						if cidStr != "" {
							ci, err := api.DecodeCid(cidStr)
							checkErr("parsing cid", err)
							resp, cerr := globalClient.Allocation(ctx, ci)
							formatResponse(c, resp, cerr)
						} else {
							var filter api.PinType
							strFilter := strings.Split(c.String("filter"), ",")
							for _, f := range strFilter {
								filter |= api.PinTypeFromString(f)
							}

							allocs := make(chan api.Pin, 1024)
							errCh := make(chan error, 1)
							go func() {
								defer close(errCh)
								errCh <- globalClient.Allocations(ctx, filter, allocs)
							}()
							formatResponse(c, allocs, nil)
							err := <-errCh
							formatResponse(c, nil, err)
						}
						return nil
					},
				},
			},
		},
		{
			Name:  "status",
			Usage: "Retrieve the status of tracked items",
			Description: `
This command retrieves the status of the CIDs tracked by IPFS
Cluster, including which member is pinning them and any errors.
If one of several CIDs are provided, the status will be only fetched 
for a single item.  Metadata CIDs are included in the status response.

When the --local flag is passed, it will only fetch the status from the
contacted cluster peer. By default, status will be fetched from all peers.

When the --filter flag is passed, it will only fetch the peer information
where status of the pin matches at least one of the filter values (a comma
separated list). The following are valid status values:

` + trackerStatusAllString(),
			ArgsUsage: "[CID1] [CID2]...",
			Flags: []cli.Flag{
				localFlag(),
				cli.StringFlag{
					Name:  "filter",
					Usage: "comma-separated list of filters",
				},
			},
			Action: func(c *cli.Context) error {
				cidsStr := c.Args()
				cids := make([]api.Cid, len(cidsStr))
				for i, cStr := range cidsStr {
					ci, err := api.DecodeCid(cStr)
					checkErr("parsing cid", err)
					cids[i] = ci
				}
				out := make(chan api.GlobalPinInfo, 1024)
				chErr := make(chan error, 1)
				go func() {
					defer close(chErr)

					if len(cids) == 1 {
						resp, cerr := globalClient.Status(ctx, cids[0], c.Bool("local"))
						out <- resp
						chErr <- cerr
						close(out)
					} else if len(cids) > 1 {
						chErr <- globalClient.StatusCids(ctx, cids, c.Bool("local"), out)
					} else {
						filterFlag := c.String("filter")
						filter := api.TrackerStatusFromString(c.String("filter"))
						if filter == api.TrackerStatusUndefined && filterFlag != "" {
							checkErr("parsing filter flag", errors.New("invalid filter name"))
						}
						chErr <- globalClient.StatusAll(ctx, filter, c.Bool("local"), out)
					}
				}()

				formatResponse(c, out, nil)
				err := <-chErr
				formatResponse(c, nil, err)
				return nil
			},
		},
		{
			Name:  "recover",
			Usage: "Recover tracked items in error state",
			Description: `
This command asks Cluster peers to re-track or re-forget CIDs in
error state, usually because the IPFS pin or unpin operation has failed.

The command will wait for any operations to succeed and will return the status
of the item upon completion. Note that, when running on the full sets of tracked
CIDs (without argument), it may take a considerably long time.

When the --local flag is passed, it will only trigger recover
operations on the contacted peer (as opposed to on every peer).
`,
			ArgsUsage: "[CID]",
			Flags: []cli.Flag{
				localFlag(),
			},
			Action: func(c *cli.Context) error {
				cidStr := c.Args().First()
				if cidStr != "" {
					ci, err := api.DecodeCid(cidStr)
					checkErr("parsing cid", err)
					resp, cerr := globalClient.Recover(ctx, ci, c.Bool("local"))
					formatResponse(c, resp, cerr)
				} else {
					out := make(chan api.GlobalPinInfo, 1024)
					errCh := make(chan error, 1)
					go func() {
						defer close(errCh)
						errCh <- globalClient.RecoverAll(ctx, c.Bool("local"), out)
					}()
					formatResponse(c, out, nil)
					err := <-errCh
					formatResponse(c, nil, err)
				}
				return nil
			},
		},

		{
			Name:  "version",
			Usage: "Retrieve cluster version",
			Description: `
This command retrieves the IPFS Cluster version and can be used
to check that it matches the CLI version (shown by -v).
`,
			ArgsUsage: " ",
			Flags:     []cli.Flag{},
			Action: func(c *cli.Context) error {
				resp, cerr := globalClient.Version(ctx)
				formatResponse(c, resp, cerr)
				return nil
			},
		},
		{
			Name:        "health",
			Usage:       "Cluster monitoring information",
			Description: "Cluster monitoring information",
			Subcommands: []cli.Command{
				{
					Name:  "graph",
					Usage: "create a graph displaying connectivity of cluster peers",
					Description: `
This command queries all connected cluster peers and their ipfs peers to generate a
graph of the connections.  Output is a dot file encoding the cluster's connection state.
`,
					Flags: []cli.Flag{
						cli.StringFlag{
							Name:  "file, f",
							Value: "",
							Usage: "sets an output dot-file for the connectivity graph",
						},
						cli.BoolFlag{
							Name:  "all-ipfs-peers",
							Usage: "causes the graph to mark nodes for ipfs peers not directly in the cluster",
						},
					},
					Action: func(c *cli.Context) error {
						resp, cerr := globalClient.GetConnectGraph(ctx)
						if cerr != nil {
							formatResponse(c, resp, cerr)
							return nil
						}
						var w io.WriteCloser
						var err error
						outputPath := c.String("file")
						if outputPath == "" {
							w = os.Stdout
						} else {
							w, err = os.Create(outputPath)
							checkErr("creating output file", err)
						}
						defer w.Close()
						err = makeDot(resp, w, c.Bool("all-ipfs-peers"))
						checkErr("printing graph", err)

						return nil
					},
				},
				{
					Name:  "metrics",
					Usage: "List latest metrics logged by this peer",
					Description: `
This commands displays the latest valid metrics of the given type logged
by this peer for all current cluster peers.

If no argument is provided, the command retrieves all currently existing metric types.

Currently supported metrics depend on the informer component used,
but usually are:

- freespace
- ping
`,
					ArgsUsage: "<metric name>",
					Action: func(c *cli.Context) error {
						metric := c.Args().First()
						if metric == "" {
							resp, cerr := globalClient.MetricNames(ctx)
							formatResponse(c, resp, cerr)
							return nil
						}

						resp, cerr := globalClient.Metrics(ctx, metric)
						formatResponse(c, resp, cerr)
						return nil
					},
				},
				{
					Name:  "alerts",
					Usage: "List the latest expired metric alerts",
					Description: `
This command provides a list of "alerts" that the cluster has seen.

An alert is triggered when one of the metrics seen for a peer expires, and no
new metrics have been received.

Different alerts may be handled in different ways. i.e. ping alerts may
trigger automatic repinnings if configured.
`,
					Action: func(c *cli.Context) error {
						resp, cerr := globalClient.Alerts(ctx)
						formatResponse(c, resp, cerr)
						return nil
					},
				},
			},
		},
		{
			Name:        "ipfs",
			Usage:       "Manage IPFS daemon",
			Description: "Manage IPFS daemon",
			Subcommands: []cli.Command{
				{
					Name:  "gc",
					Usage: "run garbage collection on IPFS repos of cluster peers",
					Description: `
This command will instruct current Cluster peers to run "repo gc" on their
respective IPFS daemons.

When --local flag is passed, it will garbage collect only on the local IPFS
deamon, otherwise on all IPFS daemons.
`,
					Flags: []cli.Flag{
						localFlag(),
					},
					Action: func(c *cli.Context) error {
						resp, cerr := globalClient.RepoGC(ctx, c.Bool("local"))
						formatResponse(c, resp, cerr)
						return nil
					},
				},
			},
		},
		{
			Name:      "commands",
			Usage:     "List all commands",
			ArgsUsage: " ",
			Hidden:    true,
			Action: func(c *cli.Context) error {
				walkCommands(c.App.Commands, "ipfs-cluster-ctl")
				return nil
			},
		},
	}

	err := app.Run(os.Args)
	if err != nil {
		os.Exit(1)
	}
}

func localFlag() cli.BoolFlag {
	return cli.BoolFlag{
		Name:  "local",
		Usage: "run operation only on the contacted peer",
	}
}

func walkCommands(cmds []cli.Command, parentHelpName string) {
	for _, c := range cmds {
		h := c.HelpName
		// Sometimes HelpName is empty
		if h == "" {
			h = fmt.Sprintf("%s %s", parentHelpName, c.FullName())
		}
		fmt.Println(h)
		walkCommands(c.Subcommands, h)
	}
}

func formatResponse(c *cli.Context, resp interface{}, err error) {
	enc := c.GlobalString("encoding")
	if resp == nil && err == nil {
		return
	}

	if err != nil {
		cerr, ok := err.(api.Error)
		if !ok {
			checkErr("", err)
		}
		switch enc {
		case "text":
			textFormatPrintError(cerr)
		case "json":
			jsonFormatPrint(cerr)
		default:
			checkErr("", errors.New("unsupported encoding selected"))
		}
		if cerr.Code == 0 {
			os.Exit(1) // problem with the call
		} else {
			os.Exit(2) // call went fine, response has an error
		}
	}

	switch enc {
	case "text":
		textFormatObject(resp)
	case "json":
		jsonFormatObject(resp)
	default:
		checkErr("", errors.New("unsupported encoding selected"))
	}
}

func parseCredentials(userInput string) (string, string) {
	credentials := strings.SplitN(userInput, ":", 2)
	switch len(credentials) {
	case 1:
		// only username passed in (with no trailing `:`), return empty password
		return credentials[0], ""
	case 2:
		return credentials[0], credentials[1]
	default:
		err := fmt.Errorf("invalid <username>[:<password>] input")
		checkErr("parsing credentials", err)
		return "", ""
	}
}

func handlePinResponseFormatFlags(
	ctx context.Context,
	c *cli.Context,
	pin api.Pin,
	target api.TrackerStatus,
) {

	var status api.GlobalPinInfo
	var cerr error

	if c.Bool("wait") {
		limit := 0
		if target == api.TrackerStatusPinned {
			limit = pin.ReplicationFactorMin
		}
		status, cerr = waitFor(pin.Cid, target, c.Duration("wait-timeout"), limit)
		checkErr("waiting for pin status", cerr)
	}

	if c.Bool("no-status") {
		formatResponse(c, pin, nil)
		return
	}

	if !status.Defined() { // no status from "wait"
		time.Sleep(time.Second)
		status, cerr = globalClient.Status(ctx, pin.Cid, false)
	}
	formatResponse(c, status, cerr)
}

func waitFor(
	ci api.Cid,
	target api.TrackerStatus,
	timeout time.Duration,
	limit int,
) (api.GlobalPinInfo, error) {

	ctx := context.Background()

	if timeout > defaultWaitCheckFreq {
		var cancel context.CancelFunc
		ctx, cancel = context.WithTimeout(ctx, timeout)
		defer cancel()
	}

	fp := client.StatusFilterParams{
		Cid:       ci,
		Local:     false,
		Target:    target,
		CheckFreq: defaultWaitCheckFreq,
		Limit:     limit,
	}

	return client.WaitFor(ctx, globalClient, fp)
}

func parseMetadata(metadata []string) map[string]string {
	metadataMap := make(map[string]string)
	for _, str := range metadata {
		parts := strings.SplitN(str, "=", 2)
		if len(parts) != 2 {
			checkErr("parsing metadata", errors.New("metadata were not in the format key=value"))
		}
		metadataMap[parts[0]] = parts[1]
	}

	return metadataMap
}

// func setupTracing(config tracingConfig) {
// 	if !config.Enable {
// 		return
// 	}

// 	agentEndpointURI := "0.0.0.0:6831"
// 	collectorEndpointURI := "http://0.0.0.0:14268"

// 	if config.JaegerAgentEndpoint != "" {
// 		agentEndpointURI = config.JaegerAgentEndpoint
// 	}
// 	if config.JaegerCollectorEndpoint != "" {
// 		collectorEndpointURI = config.JaegerCollectorEndpoint
// 	}

// 	je, err := jaeger.NewExporter(jaeger.Options{
// 		AgentEndpoint:     agentEndpointURI,
// 		CollectorEndpoint: collectorEndpointURI,
// 		ServiceName:       "ipfs-cluster-ctl",
// 	})
// 	if err != nil {
// 		log.Fatalf("Failed to create the Jaeger exporter: %v", err)
// 	}
// 	// Register/enable the trace exporter
// 	trace.RegisterExporter(je)

// 	// For demo purposes, set the trace sampling probability to be high
// 	trace.ApplyConfig(trace.Config{DefaultSampler: trace.ProbabilitySampler(1.0)})
// }