lotus/lib/consensus/raft/config.go
2022-09-12 16:10:15 -04:00

320 lines
11 KiB
Go

package consensus
import (
"io/ioutil"
"time"
hraft "github.com/hashicorp/raft"
"github.com/libp2p/go-libp2p/core/peer"
)
// ConfigKey is the default configuration key for holding this component's
// configuration section.
var configKey = "raft"
var envConfigKey = "cluster_raft"
// Configuration defaults
var (
DefaultDataSubFolder = "raft"
DefaultWaitForLeaderTimeout = 15 * time.Second
DefaultCommitRetries = 1
DefaultNetworkTimeout = 100 * time.Second
DefaultCommitRetryDelay = 200 * time.Millisecond
DefaultBackupsRotate = 6
DefaultDatastoreNamespace = "/r" // from "/raft"
)
// Config allows to configure the Raft Consensus component for ipfs-cluster.
// The component's configuration section is represented by ConfigJSON.
// Config implements the ComponentConfig interface.
type Config struct {
//config.Saver
//
//// will shutdown libp2p host on shutdown. Useful for testing
hostShutdown bool
// A folder to store Raft's data.
DataFolder string
// InitPeerset provides the list of initial cluster peers for new Raft
// peers (with no prior state). It is ignored when Raft was already
// initialized or when starting in staging mode.
InitPeerset []peer.ID
// LeaderTimeout specifies how long to wait for a leader before
// failing an operation.
WaitForLeaderTimeout time.Duration
// NetworkTimeout specifies how long before a Raft network
// operation is timed out
NetworkTimeout time.Duration
// CommitRetries specifies how many times we retry a failed commit until
// we give up.
CommitRetries int
// How long to wait between retries
CommitRetryDelay time.Duration
// BackupsRotate specifies the maximum number of Raft's DataFolder
// copies that we keep as backups (renaming) after cleanup.
BackupsRotate int
// Namespace to use when writing keys to the datastore
DatastoreNamespace string
// A Hashicorp Raft's configuration object.
RaftConfig *hraft.Config
// Tracing enables propagation of contexts across binary boundaries.
Tracing bool
}
// ConfigJSON represents a human-friendly Config
// object which can be saved to JSON. Most configuration keys are converted
// into simple types like strings, and key names aim to be self-explanatory
// for the user.
// Check https://godoc.org/github.com/hashicorp/raft#Config for extended
// description on all Raft-specific keys.
//type jsonConfig struct {
// // Storage folder for snapshots, log store etc. Used by
// // the Raft.
// DataFolder string `json:"data_folder,omitempty"`
//
// // InitPeerset provides the list of initial cluster peers for new Raft
// // peers (with no prior state). It is ignored when Raft was already
// // initialized or when starting in staging mode.
// InitPeerset []string `json:"init_peerset"`
//
// // How long to wait for a leader before failing
// WaitForLeaderTimeout string `json:"wait_for_leader_timeout"`
//
// // How long to wait before timing out network operations
// NetworkTimeout string `json:"network_timeout"`
//
// // How many retries to make upon a failed commit
// CommitRetries int `json:"commit_retries"`
//
// // How long to wait between commit retries
// CommitRetryDelay string `json:"commit_retry_delay"`
//
// // BackupsRotate specifies the maximum number of Raft's DataFolder
// // copies that we keep as backups (renaming) after cleanup.
// BackupsRotate int `json:"backups_rotate"`
//
// DatastoreNamespace string `json:"datastore_namespace,omitempty"`
//
// // HeartbeatTimeout specifies the time in follower state without
// // a leader before we attempt an election.
// HeartbeatTimeout string `json:"heartbeat_timeout,omitempty"`
//
// // ElectionTimeout specifies the time in candidate state without
// // a leader before we attempt an election.
// ElectionTimeout string `json:"election_timeout,omitempty"`
//
// // CommitTimeout controls the time without an Apply() operation
// // before we heartbeat to ensure a timely commit.
// CommitTimeout string `json:"commit_timeout,omitempty"`
//
// // MaxAppendEntries controls the maximum number of append entries
// // to send at once.
// MaxAppendEntries int `json:"max_append_entries,omitempty"`
//
// // TrailingLogs controls how many logs we leave after a snapshot.
// TrailingLogs uint64 `json:"trailing_logs,omitempty"`
//
// // SnapshotInterval controls how often we check if we should perform
// // a snapshot.
// SnapshotInterval string `json:"snapshot_interval,omitempty"`
//
// // SnapshotThreshold controls how many outstanding logs there must be
// // before we perform a snapshot.
// SnapshotThreshold uint64 `json:"snapshot_threshold,omitempty"`
//
// // LeaderLeaseTimeout is used to control how long the "lease" lasts
// // for being the leader without being able to contact a quorum
// // of nodes. If we reach this interval without contact, we will
// // step down as leader.
// LeaderLeaseTimeout string `json:"leader_lease_timeout,omitempty"`
//
// // The unique ID for this server across all time. When running with
// // ProtocolVersion < 3, you must set this to be the same as the network
// // address of your transport.
// // LocalID string `json:local_id`
//}
// ConfigKey returns a human-friendly indentifier for this Config.
func (cfg *Config) ConfigKey() string {
return configKey
}
//// Validate checks that this configuration has working values,
//// at least in appearance.
//func (cfg *Config) Validate() error {
// if cfg.RaftConfig == nil {
// return errors.New("no hashicorp/raft.Config")
// }
// if cfg.WaitForLeaderTimeout <= 0 {
// return errors.New("wait_for_leader_timeout <= 0")
// }
//
// if cfg.NetworkTimeout <= 0 {
// return errors.New("network_timeout <= 0")
// }
//
// if cfg.CommitRetries < 0 {
// return errors.New("commit_retries is invalid")
// }
//
// if cfg.CommitRetryDelay <= 0 {
// return errors.New("commit_retry_delay is invalid")
// }
//
// if cfg.BackupsRotate <= 0 {
// return errors.New("backups_rotate should be larger than 0")
// }
//
// return hraft.ValidateConfig(cfg.RaftConfig)
//}
// LoadJSON parses a json-encoded configuration (see jsonConfig).
// The Config will have default values for all fields not explicited
// in the given json object.
//func (cfg *Config) LoadJSON(raw []byte) error {
// jcfg := &jsonConfig{}
// err := json.Unmarshal(raw, jcfg)
// if err != nil {
// logger.Error("Error unmarshaling raft config")
// return err
// }
//
// cfg.Default()
//
// return cfg.applyJSONConfig(jcfg)
//}
//
//func (cfg *Config) applyJSONConfig(jcfg *jsonConfig) error {
// parseDuration := func(txt string) time.Duration {
// d, _ := time.ParseDuration(txt)
// if txt != "" && d == 0 {
// logger.Warnf("%s is not a valid duration. Default will be used", txt)
// }
// return d
// }
//
// // Parse durations. We ignore errors as 0 will take Default values.
// waitForLeaderTimeout := parseDuration(jcfg.WaitForLeaderTimeout)
// networkTimeout := parseDuration(jcfg.NetworkTimeout)
// commitRetryDelay := parseDuration(jcfg.CommitRetryDelay)
// heartbeatTimeout := parseDuration(jcfg.HeartbeatTimeout)
// electionTimeout := parseDuration(jcfg.ElectionTimeout)
// commitTimeout := parseDuration(jcfg.CommitTimeout)
// snapshotInterval := parseDuration(jcfg.SnapshotInterval)
// leaderLeaseTimeout := parseDuration(jcfg.LeaderLeaseTimeout)
//
// // Set all values in config. For some, take defaults if they are 0.
// // Set values from jcfg if they are not 0 values
//
// // Own values
// config.SetIfNotDefault(jcfg.DataFolder, &cfg.DataFolder)
// config.SetIfNotDefault(waitForLeaderTimeout, &cfg.WaitForLeaderTimeout)
// config.SetIfNotDefault(networkTimeout, &cfg.NetworkTimeout)
// cfg.CommitRetries = jcfg.CommitRetries
// config.SetIfNotDefault(commitRetryDelay, &cfg.CommitRetryDelay)
// config.SetIfNotDefault(jcfg.BackupsRotate, &cfg.BackupsRotate)
//
// // Raft values
// config.SetIfNotDefault(heartbeatTimeout, &cfg.RaftConfig.HeartbeatTimeout)
// config.SetIfNotDefault(electionTimeout, &cfg.RaftConfig.ElectionTimeout)
// config.SetIfNotDefault(commitTimeout, &cfg.RaftConfig.CommitTimeout)
// config.SetIfNotDefault(jcfg.MaxAppendEntries, &cfg.RaftConfig.MaxAppendEntries)
// config.SetIfNotDefault(jcfg.TrailingLogs, &cfg.RaftConfig.TrailingLogs)
// config.SetIfNotDefault(snapshotInterval, &cfg.RaftConfig.SnapshotInterval)
// config.SetIfNotDefault(jcfg.SnapshotThreshold, &cfg.RaftConfig.SnapshotThreshold)
// config.SetIfNotDefault(leaderLeaseTimeout, &cfg.RaftConfig.LeaderLeaseTimeout)
//
// cfg.InitPeerset = api.StringsToPeers(jcfg.InitPeerset)
// return cfg.Validate()
//}
//
//// ToJSON returns the pretty JSON representation of a Config.
//func (cfg *Config) ToJSON() ([]byte, error) {
// jcfg := cfg.toJSONConfig()
//
// return config.DefaultJSONMarshal(jcfg)
//}
//
//func (cfg *Config) toJSONConfig() *jsonConfig {
// jcfg := &jsonConfig{
// DataFolder: cfg.DataFolder,
// InitPeerset: api.PeersToStrings(cfg.InitPeerset),
// WaitForLeaderTimeout: cfg.WaitForLeaderTimeout.String(),
// NetworkTimeout: cfg.NetworkTimeout.String(),
// CommitRetries: cfg.CommitRetries,
// CommitRetryDelay: cfg.CommitRetryDelay.String(),
// BackupsRotate: cfg.BackupsRotate,
// HeartbeatTimeout: cfg.RaftConfig.HeartbeatTimeout.String(),
// ElectionTimeout: cfg.RaftConfig.ElectionTimeout.String(),
// CommitTimeout: cfg.RaftConfig.CommitTimeout.String(),
// MaxAppendEntries: cfg.RaftConfig.MaxAppendEntries,
// TrailingLogs: cfg.RaftConfig.TrailingLogs,
// SnapshotInterval: cfg.RaftConfig.SnapshotInterval.String(),
// SnapshotThreshold: cfg.RaftConfig.SnapshotThreshold,
// LeaderLeaseTimeout: cfg.RaftConfig.LeaderLeaseTimeout.String(),
// }
// if cfg.DatastoreNamespace != DefaultDatastoreNamespace {
// jcfg.DatastoreNamespace = cfg.DatastoreNamespace
// // otherwise leave empty so it gets omitted.
// }
// return jcfg
//}
//
// Default initializes this configuration with working defaults.
func (cfg *Config) Default() error {
cfg.DataFolder = "" // empty so it gets omitted
cfg.InitPeerset = []peer.ID{}
cfg.WaitForLeaderTimeout = DefaultWaitForLeaderTimeout
cfg.NetworkTimeout = DefaultNetworkTimeout
cfg.CommitRetries = DefaultCommitRetries
cfg.CommitRetryDelay = DefaultCommitRetryDelay
cfg.BackupsRotate = DefaultBackupsRotate
cfg.DatastoreNamespace = DefaultDatastoreNamespace
cfg.RaftConfig = hraft.DefaultConfig()
// These options are imposed over any Default Raft Config.
cfg.RaftConfig.ShutdownOnRemove = false
cfg.RaftConfig.LocalID = "will_be_set_automatically"
// Set up logging
cfg.RaftConfig.LogOutput = ioutil.Discard
//cfg.RaftConfig.Logger = &hcLogToLogger{}
return nil
}
func NewDefaultConfig() *Config {
var cfg Config
cfg.Default()
return &cfg
}
//
//// ApplyEnvVars fills in any Config fields found
//// as environment variables.
//func (cfg *Config) ApplyEnvVars() error {
// jcfg := cfg.toJSONConfig()
//
// err := envconfig.Process(envConfigKey, jcfg)
// if err != nil {
// return err
// }
//
// return cfg.applyJSONConfig(jcfg)
//}
//
//// GetDataFolder returns the Raft data folder that we are using.
//func (cfg *Config) GetDataFolder() string {
// if cfg.DataFolder == "" {
// return filepath.Join(cfg.BaseDir, DefaultDataSubFolder)
// }
// return cfg.DataFolder
//}
//
//// ToDisplayJSON returns JSON config as a string.
//func (cfg *Config) ToDisplayJSON() ([]byte, error) {
// return config.DisplayJSON(cfg.toJSONConfig())
//}