d8dfun/virtual-kubelet
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

Add HashiCorp Nomad provider (#483)

Browse Source 
* provider: adding Nomad provider

* updating CONTRIBUTING.md with Nomad provider

* updated README.md by adding the Nomad provider

* fix typo

* adding nomad/api and nomad/testutil deps

* adding Nomad binary dependency for provider tests

* fixed the nomad binary download command step and added tolerations to the nomad provider.

* adding nomad provider demo gif

* adding my name to authors

* adding two missing go-rootcerts files after dep ensure

* delete pod comment
This commit is contained in:
Anubhav Mishra
2019-01-08 01:18:11 +05:30
committed by Robbie Zhang
parent 5796be449b
commit a46e1dd2ce
332 changed files with 126455 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

354
vendor/github.com/hashicorp/raft/LICENSE generated vendored Normal file
View File

@@ -0,0 +1,354 @@
Mozilla Public License, version 2.0
1. Definitions
1.1. “Contributor”
means each individual or legal entity that creates, contributes to the
creation of, or owns Covered Software.
1.2. “Contributor Version”
means the combination of the Contributions of others (if any) used by a
Contributor and that particular Contributor’s Contribution.
1.3. “Contribution”
means Covered Software of a particular Contributor.
1.4. “Covered Software”
means Source Code Form to which the initial Contributor has attached the
notice in Exhibit A, the Executable Form of such Source Code Form, and
Modifications of such Source Code Form, in each case including portions
thereof.
1.5. “Incompatible With Secondary Licenses”
means
a. that the initial Contributor has attached the notice described in
Exhibit B to the Covered Software; or
b. that the Covered Software was made available under the terms of version
1.1 or earlier of the License, but not also under the terms of a
Secondary License.
1.6. “Executable Form”
means any form of the work other than Source Code Form.
1.7. “Larger Work”
means a work that combines Covered Software with other material, in a separate
file or files, that is not Covered Software.
1.8. “License”
means this document.
1.9. “Licensable”
means having the right to grant, to the maximum extent possible, whether at the
time of the initial grant or subsequently, any and all of the rights conveyed by
this License.
1.10. “Modifications”
means any of the following:
a. any file in Source Code Form that results from an addition to, deletion
from, or modification of the contents of Covered Software; or
b. any new file in Source Code Form that contains any Covered Software.
1.11. “Patent Claims” of a Contributor
means any patent claim(s), including without limitation, method, process,
and apparatus claims, in any patent Licensable by such Contributor that
would be infringed, but for the grant of the License, by the making,
using, selling, offering for sale, having made, import, or transfer of
either its Contributions or its Contributor Version.
1.12. “Secondary License”
means either the GNU General Public License, Version 2.0, the GNU Lesser
General Public License, Version 2.1, the GNU Affero General Public
License, Version 3.0, or any later versions of those licenses.
1.13. “Source Code Form”
means the form of the work preferred for making modifications.
1.14. “You” (or “Your”)
means an individual or a legal entity exercising rights under this
License. For legal entities, “You” includes any entity that controls, is
controlled by, or is under common control with You. For purposes of this
definition, “control” means (a) the power, direct or indirect, to cause
the direction or management of such entity, whether by contract or
otherwise, or (b) ownership of more than fifty percent (50%) of the
outstanding shares or beneficial ownership of such entity.
2. License Grants and Conditions
2.1. Grants
Each Contributor hereby grants You a world-wide, royalty-free,
non-exclusive license:
a. under intellectual property rights (other than patent or trademark)
Licensable by such Contributor to use, reproduce, make available,
modify, display, perform, distribute, and otherwise exploit its
Contributions, either on an unmodified basis, with Modifications, or as
part of a Larger Work; and
b. under Patent Claims of such Contributor to make, use, sell, offer for
sale, have made, import, and otherwise transfer either its Contributions
or its Contributor Version.
2.2. Effective Date
The licenses granted in Section 2.1 with respect to any Contribution become
effective for each Contribution on the date the Contributor first distributes
such Contribution.
2.3. Limitations on Grant Scope
The licenses granted in this Section 2 are the only rights granted under this
License. No additional rights or licenses will be implied from the distribution
or licensing of Covered Software under this License. Notwithstanding Section
2.1(b) above, no patent license is granted by a Contributor:
a. for any code that a Contributor has removed from Covered Software; or
b. for infringements caused by: (i) Your and any other third party’s
modifications of Covered Software, or (ii) the combination of its
Contributions with other software (except as part of its Contributor
Version); or
c. under Patent Claims infringed by Covered Software in the absence of its
Contributions.
This License does not grant any rights in the trademarks, service marks, or
logos of any Contributor (except as may be necessary to comply with the
notice requirements in Section 3.4).
2.4. Subsequent Licenses
No Contributor makes additional grants as a result of Your choice to
distribute the Covered Software under a subsequent version of this License
(see Section 10.2) or under the terms of a Secondary License (if permitted
under the terms of Section 3.3).
2.5. Representation
Each Contributor represents that the Contributor believes its Contributions
are its original creation(s) or it has sufficient rights to grant the
rights to its Contributions conveyed by this License.
2.6. Fair Use
This License is not intended to limit any rights You have under applicable
copyright doctrines of fair use, fair dealing, or other equivalents.
2.7. Conditions
Sections 3.1, 3.2, 3.3, and 3.4 are conditions of the licenses granted in
Section 2.1.
3. Responsibilities
3.1. Distribution of Source Form
All distribution of Covered Software in Source Code Form, including any
Modifications that You create or to which You contribute, must be under the
terms of this License. You must inform recipients that the Source Code Form
of the Covered Software is governed by the terms of this License, and how
they can obtain a copy of this License. You may not attempt to alter or
restrict the recipients’ rights in the Source Code Form.
3.2. Distribution of Executable Form
If You distribute Covered Software in Executable Form then:
a. such Covered Software must also be made available in Source Code Form,
as described in Section 3.1, and You must inform recipients of the
Executable Form how they can obtain a copy of such Source Code Form by
reasonable means in a timely manner, at a charge no more than the cost
of distribution to the recipient; and
b. You may distribute such Executable Form under the terms of this License,
or sublicense it under different terms, provided that the license for
the Executable Form does not attempt to limit or alter the recipients’
rights in the Source Code Form under this License.
3.3. Distribution of a Larger Work
You may create and distribute a Larger Work under terms of Your choice,
provided that You also comply with the requirements of this License for the
Covered Software. If the Larger Work is a combination of Covered Software
with a work governed by one or more Secondary Licenses, and the Covered
Software is not Incompatible With Secondary Licenses, this License permits
You to additionally distribute such Covered Software under the terms of
such Secondary License(s), so that the recipient of the Larger Work may, at
their option, further distribute the Covered Software under the terms of
either this License or such Secondary License(s).
3.4. Notices
You may not remove or alter the substance of any license notices (including
copyright notices, patent notices, disclaimers of warranty, or limitations
of liability) contained within the Source Code Form of the Covered
Software, except that You may alter any license notices to the extent
required to remedy known factual inaccuracies.
3.5. Application of Additional Terms
You may choose to offer, and to charge a fee for, warranty, support,
indemnity or liability obligations to one or more recipients of Covered
Software. However, You may do so only on Your own behalf, and not on behalf
of any Contributor. You must make it absolutely clear that any such
warranty, support, indemnity, or liability obligation is offered by You
alone, and You hereby agree to indemnify every Contributor for any
liability incurred by such Contributor as a result of warranty, support,
indemnity or liability terms You offer. You may include additional
disclaimers of warranty and limitations of liability specific to any
jurisdiction.
4. Inability to Comply Due to Statute or Regulation
If it is impossible for You to comply with any of the terms of this License
with respect to some or all of the Covered Software due to statute, judicial
order, or regulation then You must: (a) comply with the terms of this License
to the maximum extent possible; and (b) describe the limitations and the code
they affect. Such description must be placed in a text file included with all
distributions of the Covered Software under this License. Except to the
extent prohibited by statute or regulation, such description must be
sufficiently detailed for a recipient of ordinary skill to be able to
understand it.
5. Termination
5.1. The rights granted under this License will terminate automatically if You
fail to comply with any of its terms. However, if You become compliant,
then the rights granted under this License from a particular Contributor
are reinstated (a) provisionally, unless and until such Contributor
explicitly and finally terminates Your grants, and (b) on an ongoing basis,
if such Contributor fails to notify You of the non-compliance by some
reasonable means prior to 60 days after You have come back into compliance.
Moreover, Your grants from a particular Contributor are reinstated on an
ongoing basis if such Contributor notifies You of the non-compliance by
some reasonable means, this is the first time You have received notice of
non-compliance with this License from such Contributor, and You become
compliant prior to 30 days after Your receipt of the notice.
5.2. If You initiate litigation against any entity by asserting a patent
infringement claim (excluding declaratory judgment actions, counter-claims,
and cross-claims) alleging that a Contributor Version directly or
indirectly infringes any patent, then the rights granted to You by any and
all Contributors for the Covered Software under Section 2.1 of this License
shall terminate.
5.3. In the event of termination under Sections 5.1 or 5.2 above, all end user
license agreements (excluding distributors and resellers) which have been
validly granted by You or Your distributors under this License prior to
termination shall survive termination.
6. Disclaimer of Warranty
Covered Software is provided under this License on an “as is” basis, without
warranty of any kind, either expressed, implied, or statutory, including,
without limitation, warranties that the Covered Software is free of defects,
merchantable, fit for a particular purpose or non-infringing. The entire
risk as to the quality and performance of the Covered Software is with You.
Should any Covered Software prove defective in any respect, You (not any
Contributor) assume the cost of any necessary servicing, repair, or
correction. This disclaimer of warranty constitutes an essential part of this
License. No use of any Covered Software is authorized under this License
except under this disclaimer.
7. Limitation of Liability
Under no circumstances and under no legal theory, whether tort (including
negligence), contract, or otherwise, shall any Contributor, or anyone who
distributes Covered Software as permitted above, be liable to You for any
direct, indirect, special, incidental, or consequential damages of any
character including, without limitation, damages for lost profits, loss of
goodwill, work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses, even if such party shall have been
informed of the possibility of such damages. This limitation of liability
shall not apply to liability for death or personal injury resulting from such
party’s negligence to the extent applicable law prohibits such limitation.
Some jurisdictions do not allow the exclusion or limitation of incidental or
consequential damages, so this exclusion and limitation may not apply to You.
8. Litigation
Any litigation relating to this License may be brought only in the courts of
a jurisdiction where the defendant maintains its principal place of business
and such litigation shall be governed by laws of that jurisdiction, without
reference to its conflict-of-law provisions. Nothing in this Section shall
prevent a party’s ability to bring cross-claims or counter-claims.
9. Miscellaneous
This License represents the complete agreement concerning the subject matter
hereof. If any provision of this License is held to be unenforceable, such
provision shall be reformed only to the extent necessary to make it
enforceable. Any law or regulation which provides that the language of a
contract shall be construed against the drafter shall not be used to construe
this License against a Contributor.
10. Versions of the License
10.1. New Versions
Mozilla Foundation is the license steward. Except as provided in Section
10.3, no one other than the license steward has the right to modify or
publish new versions of this License. Each version will be given a
distinguishing version number.
10.2. Effect of New Versions
You may distribute the Covered Software under the terms of the version of
the License under which You originally received the Covered Software, or
under the terms of any subsequent version published by the license
steward.
10.3. Modified Versions
If you create software not governed by this License, and you want to
create a new license for such software, you may create and use a modified
version of this License if you rename the license and remove any
references to the name of the license steward (except to note that such
modified license differs from this License).
10.4. Distributing Source Code Form that is Incompatible With Secondary Licenses
If You choose to distribute Source Code Form that is Incompatible With
Secondary Licenses under the terms of this version of the License, the
notice described in Exhibit B of this License must be attached.
Exhibit A - Source Code Form License Notice
This Source Code Form is subject to the
terms of the Mozilla Public License, v.
2.0. If a copy of the MPL was not
distributed with this file, You can
obtain one at
http://mozilla.org/MPL/2.0/.
If it is not possible or desirable to put the notice in a particular file, then
You may include the notice in a location (such as a LICENSE file in a relevant
directory) where a recipient would be likely to look for such a notice.
You may add additional accurate notices of copyright ownership.
Exhibit B - “Incompatible With Secondary Licenses” Notice
This Source Code Form is “Incompatible
With Secondary Licenses”, as defined by
the Mozilla Public License, v. 2.0.

1008
vendor/github.com/hashicorp/raft/api.go generated vendored Normal file
View File

File diff suppressed because it is too large Load Diff

151
vendor/github.com/hashicorp/raft/commands.go generated vendored Normal file
View File

@@ -0,0 +1,151 @@
package raft
// RPCHeader is a common sub-structure used to pass along protocol version and
// other information about the cluster. For older Raft implementations before
// versioning was added this will default to a zero-valued structure when read
// by newer Raft versions.
type RPCHeader struct {
// ProtocolVersion is the version of the protocol the sender is
// speaking.
ProtocolVersion ProtocolVersion
}
// WithRPCHeader is an interface that exposes the RPC header.
type WithRPCHeader interface {
GetRPCHeader() RPCHeader
}
// AppendEntriesRequest is the command used to append entries to the
// replicated log.
type AppendEntriesRequest struct {
RPCHeader
// Provide the current term and leader
Term uint64
Leader []byte
// Provide the previous entries for integrity checking
PrevLogEntry uint64
PrevLogTerm uint64
// New entries to commit
Entries []*Log
// Commit index on the leader
LeaderCommitIndex uint64
}
// See WithRPCHeader.
func (r *AppendEntriesRequest) GetRPCHeader() RPCHeader {
return r.RPCHeader
}
// AppendEntriesResponse is the response returned from an
// AppendEntriesRequest.
type AppendEntriesResponse struct {
RPCHeader
// Newer term if leader is out of date
Term uint64
// Last Log is a hint to help accelerate rebuilding slow nodes
LastLog uint64
// We may not succeed if we have a conflicting entry
Success bool
// There are scenarios where this request didn't succeed
// but there's no need to wait/back-off the next attempt.
NoRetryBackoff bool
}
// See WithRPCHeader.
func (r *AppendEntriesResponse) GetRPCHeader() RPCHeader {
return r.RPCHeader
}
// RequestVoteRequest is the command used by a candidate to ask a Raft peer
// for a vote in an election.
type RequestVoteRequest struct {
RPCHeader
// Provide the term and our id
Term uint64
Candidate []byte
// Used to ensure safety
LastLogIndex uint64
LastLogTerm uint64
}
// See WithRPCHeader.
func (r *RequestVoteRequest) GetRPCHeader() RPCHeader {
return r.RPCHeader
}
// RequestVoteResponse is the response returned from a RequestVoteRequest.
type RequestVoteResponse struct {
RPCHeader
// Newer term if leader is out of date.
Term uint64
// Peers is deprecated, but required by servers that only understand
// protocol version 0. This is not populated in protocol version 2
// and later.
Peers []byte
// Is the vote granted.
Granted bool
}
// See WithRPCHeader.
func (r *RequestVoteResponse) GetRPCHeader() RPCHeader {
return r.RPCHeader
}
// InstallSnapshotRequest is the command sent to a Raft peer to bootstrap its
// log (and state machine) from a snapshot on another peer.
type InstallSnapshotRequest struct {
RPCHeader
SnapshotVersion SnapshotVersion
Term uint64
Leader []byte
// These are the last index/term included in the snapshot
LastLogIndex uint64
LastLogTerm uint64
// Peer Set in the snapshot. This is deprecated in favor of Configuration
// but remains here in case we receive an InstallSnapshot from a leader
// that's running old code.
Peers []byte
// Cluster membership.
Configuration []byte
// Log index where 'Configuration' entry was originally written.
ConfigurationIndex uint64
// Size of the snapshot
Size int64
}
// See WithRPCHeader.
func (r *InstallSnapshotRequest) GetRPCHeader() RPCHeader {
return r.RPCHeader
}
// InstallSnapshotResponse is the response returned from an
// InstallSnapshotRequest.
type InstallSnapshotResponse struct {
RPCHeader
Term uint64
Success bool
}
// See WithRPCHeader.
func (r *InstallSnapshotResponse) GetRPCHeader() RPCHeader {
return r.RPCHeader
}

101
vendor/github.com/hashicorp/raft/commitment.go generated vendored Normal file
View File

@@ -0,0 +1,101 @@
package raft
import (
"sort"
"sync"
)
// Commitment is used to advance the leader's commit index. The leader and
// replication goroutines report in newly written entries with Match(), and
// this notifies on commitCh when the commit index has advanced.
type commitment struct {
// protectes matchIndexes and commitIndex
sync.Mutex
// notified when commitIndex increases
commitCh chan struct{}
// voter ID to log index: the server stores up through this log entry
matchIndexes map[ServerID]uint64
// a quorum stores up through this log entry. monotonically increases.
commitIndex uint64
// the first index of this leader's term: this needs to be replicated to a
// majority of the cluster before this leader may mark anything committed
// (per Raft's commitment rule)
startIndex uint64
}
// newCommitment returns an commitment struct that notifies the provided
// channel when log entries have been committed. A new commitment struct is
// created each time this server becomes leader for a particular term.
// 'configuration' is the servers in the cluster.
// 'startIndex' is the first index created in this term (see
// its description above).
func newCommitment(commitCh chan struct{}, configuration Configuration, startIndex uint64) *commitment {
matchIndexes := make(map[ServerID]uint64)
for _, server := range configuration.Servers {
if server.Suffrage == Voter {
matchIndexes[server.ID] = 0
}
}
return &commitment{
commitCh: commitCh,
matchIndexes: matchIndexes,
commitIndex: 0,
startIndex: startIndex,
}
}
// Called when a new cluster membership configuration is created: it will be
// used to determine commitment from now on. 'configuration' is the servers in
// the cluster.
func (c *commitment) setConfiguration(configuration Configuration) {
c.Lock()
defer c.Unlock()
oldMatchIndexes := c.matchIndexes
c.matchIndexes = make(map[ServerID]uint64)
for _, server := range configuration.Servers {
if server.Suffrage == Voter {
c.matchIndexes[server.ID] = oldMatchIndexes[server.ID] // defaults to 0
}
}
c.recalculate()
}
// Called by leader after commitCh is notified
func (c *commitment) getCommitIndex() uint64 {
c.Lock()
defer c.Unlock()
return c.commitIndex
}
// Match is called once a server completes writing entries to disk: either the
// leader has written the new entry or a follower has replied to an
// AppendEntries RPC. The given server's disk agrees with this server's log up
// through the given index.
func (c *commitment) match(server ServerID, matchIndex uint64) {
c.Lock()
defer c.Unlock()
if prev, hasVote := c.matchIndexes[server]; hasVote && matchIndex > prev {
c.matchIndexes[server] = matchIndex
c.recalculate()
}
}
// Internal helper to calculate new commitIndex from matchIndexes.
// Must be called with lock held.
func (c *commitment) recalculate() {
if len(c.matchIndexes) == 0 {
return
}
matched := make([]uint64, 0, len(c.matchIndexes))
for _, idx := range c.matchIndexes {
matched = append(matched, idx)
}
sort.Sort(uint64Slice(matched))
quorumMatchIndex := matched[(len(matched)-1)/2]
if quorumMatchIndex > c.commitIndex && quorumMatchIndex >= c.startIndex {
c.commitIndex = quorumMatchIndex
asyncNotifyCh(c.commitCh)
}
}

258
vendor/github.com/hashicorp/raft/config.go generated vendored Normal file
View File

@@ -0,0 +1,258 @@
package raft
import (
"fmt"
"io"
"log"
"time"
)
// These are the versions of the protocol (which includes RPC messages as
// well as Raft-specific log entries) that this server can _understand_. Use
// the ProtocolVersion member of the Config object to control the version of
// the protocol to use when _speaking_ to other servers. Note that depending on
// the protocol version being spoken, some otherwise understood RPC messages
// may be refused. See dispositionRPC for details of this logic.
//
// There are notes about the upgrade path in the description of the versions
// below. If you are starting a fresh cluster then there's no reason not to
// jump right to the latest protocol version. If you need to interoperate with
// older, version 0 Raft servers you'll need to drive the cluster through the
// different versions in order.
//
// The version details are complicated, but here's a summary of what's required
// to get from a version 0 cluster to version 3:
//
// 1. In version N of your app that starts using the new Raft library with
// versioning, set ProtocolVersion to 1.
// 2. Make version N+1 of your app require version N as a prerequisite (all
// servers must be upgraded). For version N+1 of your app set ProtocolVersion
// to 2.
// 3. Similarly, make version N+2 of your app require version N+1 as a
// prerequisite. For version N+2 of your app, set ProtocolVersion to 3.
//
// During this upgrade, older cluster members will still have Server IDs equal
// to their network addresses. To upgrade an older member and give it an ID, it
// needs to leave the cluster and re-enter:
//
// 1. Remove the server from the cluster with RemoveServer, using its network
// address as its ServerID.
// 2. Update the server's config to a better ID (restarting the server).
// 3. Add the server back to the cluster with AddVoter, using its new ID.
//
// You can do this during the rolling upgrade from N+1 to N+2 of your app, or
// as a rolling change at any time after the upgrade.
//
// Version History
//
// 0: Original Raft library before versioning was added. Servers running this
// version of the Raft library use AddPeerDeprecated/RemovePeerDeprecated
// for all configuration changes, and have no support for LogConfiguration.
// 1: First versioned protocol, used to interoperate with old servers, and begin
// the migration path to newer versions of the protocol. Under this version
// all configuration changes are propagated using the now-deprecated
// RemovePeerDeprecated Raft log entry. This means that server IDs are always
// set to be the same as the server addresses (since the old log entry type
// cannot transmit an ID), and only AddPeer/RemovePeer APIs are supported.
// Servers running this version of the protocol can understand the new
// LogConfiguration Raft log entry but will never generate one so they can
// remain compatible with version 0 Raft servers in the cluster.
// 2: Transitional protocol used when migrating an existing cluster to the new
// server ID system. Server IDs are still set to be the same as server
// addresses, but all configuration changes are propagated using the new
// LogConfiguration Raft log entry type, which can carry full ID information.
// This version supports the old AddPeer/RemovePeer APIs as well as the new
// ID-based AddVoter/RemoveServer APIs which should be used when adding
// version 3 servers to the cluster later. This version sheds all
// interoperability with version 0 servers, but can interoperate with newer
// Raft servers running with protocol version 1 since they can understand the
// new LogConfiguration Raft log entry, and this version can still understand
// their RemovePeerDeprecated Raft log entries. We need this protocol version
// as an intermediate step between 1 and 3 so that servers will propagate the
// ID information that will come from newly-added (or -rolled) servers using
// protocol version 3, but since they are still using their address-based IDs
// from the previous step they will still be able to track commitments and
// their own voting status properly. If we skipped this step, servers would
// be started with their new IDs, but they wouldn't see themselves in the old
// address-based configuration, so none of the servers would think they had a
// vote.
// 3: Protocol adding full support for server IDs and new ID-based server APIs
// (AddVoter, AddNonvoter, etc.), old AddPeer/RemovePeer APIs are no longer
// supported. Version 2 servers should be swapped out by removing them from
// the cluster one-by-one and re-adding them with updated configuration for
// this protocol version, along with their server ID. The remove/add cycle
// is required to populate their server ID. Note that removing must be done
// by ID, which will be the old server's address.
type ProtocolVersion int
const (
ProtocolVersionMin ProtocolVersion = 0
ProtocolVersionMax = 3
)
// These are versions of snapshots that this server can _understand_. Currently,
// it is always assumed that this server generates the latest version, though
// this may be changed in the future to include a configurable version.
//
// Version History
//
// 0: Original Raft library before versioning was added. The peers portion of
// these snapshots is encoded in the legacy format which requires decodePeers
// to parse. This version of snapshots should only be produced by the
// unversioned Raft library.
// 1: New format which adds support for a full configuration structure and its
// associated log index, with support for server IDs and non-voting server
// modes. To ease upgrades, this also includes the legacy peers structure but
// that will never be used by servers that understand version 1 snapshots.
// Since the original Raft library didn't enforce any versioning, we must
// include the legacy peers structure for this version, but we can deprecate
// it in the next snapshot version.
type SnapshotVersion int
const (
SnapshotVersionMin SnapshotVersion = 0
SnapshotVersionMax = 1
)
// Config provides any necessary configuration for the Raft server.
type Config struct {
// ProtocolVersion allows a Raft server to inter-operate with older
// Raft servers running an older version of the code. This is used to
// version the wire protocol as well as Raft-specific log entries that
// the server uses when _speaking_ to other servers. There is currently
// no auto-negotiation of versions so all servers must be manually
// configured with compatible versions. See ProtocolVersionMin and
// ProtocolVersionMax for the versions of the protocol that this server
// can _understand_.
ProtocolVersion ProtocolVersion
// HeartbeatTimeout specifies the time in follower state without
// a leader before we attempt an election.
HeartbeatTimeout time.Duration
// ElectionTimeout specifies the time in candidate state without
// a leader before we attempt an election.
ElectionTimeout time.Duration
// CommitTimeout controls the time without an Apply() operation
// before we heartbeat to ensure a timely commit. Due to random
// staggering, may be delayed as much as 2x this value.
CommitTimeout time.Duration
// MaxAppendEntries controls the maximum number of append entries
// to send at once. We want to strike a balance between efficiency
// and avoiding waste if the follower is going to reject because of
// an inconsistent log.
MaxAppendEntries int
// If we are a member of a cluster, and RemovePeer is invoked for the
// local node, then we forget all peers and transition into the follower state.
// If ShutdownOnRemove is is set, we additional shutdown Raft. Otherwise,
// we can become a leader of a cluster containing only this node.
ShutdownOnRemove bool
// TrailingLogs controls how many logs we leave after a snapshot. This is
// used so that we can quickly replay logs on a follower instead of being
// forced to send an entire snapshot.
TrailingLogs uint64
// SnapshotInterval controls how often we check if we should perform a snapshot.
// We randomly stagger between this value and 2x this value to avoid the entire
// cluster from performing a snapshot at once.
SnapshotInterval time.Duration
// SnapshotThreshold controls how many outstanding logs there must be before
// we perform a snapshot. This is to prevent excessive snapshots when we can
// just replay a small set of logs.
SnapshotThreshold uint64
// 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 time.Duration
// StartAsLeader forces Raft to start in the leader state. This should
// never be used except for testing purposes, as it can cause a split-brain.
StartAsLeader bool
// 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 ServerID
// NotifyCh is used to provide a channel that will be notified of leadership
// changes. Raft will block writing to this channel, so it should either be
// buffered or aggressively consumed.
NotifyCh chan<- bool
// LogOutput is used as a sink for logs, unless Logger is specified.
// Defaults to os.Stderr.
LogOutput io.Writer
// Logger is a user-provided logger. If nil, a logger writing to LogOutput
// is used.
Logger *log.Logger
}
// DefaultConfig returns a Config with usable defaults.
func DefaultConfig() *Config {
return &Config{
ProtocolVersion: ProtocolVersionMax,
HeartbeatTimeout: 1000 * time.Millisecond,
ElectionTimeout: 1000 * time.Millisecond,
CommitTimeout: 50 * time.Millisecond,
MaxAppendEntries: 64,
ShutdownOnRemove: true,
TrailingLogs: 10240,
SnapshotInterval: 120 * time.Second,
SnapshotThreshold: 8192,
LeaderLeaseTimeout: 500 * time.Millisecond,
}
}
// ValidateConfig is used to validate a sane configuration
func ValidateConfig(config *Config) error {
// We don't actually support running as 0 in the library any more, but
// we do understand it.
protocolMin := ProtocolVersionMin
if protocolMin == 0 {
protocolMin = 1
}
if config.ProtocolVersion < protocolMin ||
config.ProtocolVersion > ProtocolVersionMax {
return fmt.Errorf("Protocol version %d must be >= %d and <= %d",
config.ProtocolVersion, protocolMin, ProtocolVersionMax)
}
if len(config.LocalID) == 0 {
return fmt.Errorf("LocalID cannot be empty")
}
if config.HeartbeatTimeout < 5*time.Millisecond {
return fmt.Errorf("Heartbeat timeout is too low")
}
if config.ElectionTimeout < 5*time.Millisecond {
return fmt.Errorf("Election timeout is too low")
}
if config.CommitTimeout < time.Millisecond {
return fmt.Errorf("Commit timeout is too low")
}
if config.MaxAppendEntries <= 0 {
return fmt.Errorf("MaxAppendEntries must be positive")
}
if config.MaxAppendEntries > 1024 {
return fmt.Errorf("MaxAppendEntries is too large")
}
if config.SnapshotInterval < 5*time.Millisecond {
return fmt.Errorf("Snapshot interval is too low")
}
if config.LeaderLeaseTimeout < 5*time.Millisecond {
return fmt.Errorf("Leader lease timeout is too low")
}
if config.LeaderLeaseTimeout > config.HeartbeatTimeout {
return fmt.Errorf("Leader lease timeout cannot be larger than heartbeat timeout")
}
if config.ElectionTimeout < config.HeartbeatTimeout {
return fmt.Errorf("Election timeout must be equal or greater than Heartbeat Timeout")
}
return nil
}

343
vendor/github.com/hashicorp/raft/configuration.go generated vendored Normal file
View File

@@ -0,0 +1,343 @@
package raft
import "fmt"
// ServerSuffrage determines whether a Server in a Configuration gets a vote.
type ServerSuffrage int
// Note: Don't renumber these, since the numbers are written into the log.
const (
// Voter is a server whose vote is counted in elections and whose match index
// is used in advancing the leader's commit index.
Voter ServerSuffrage = iota
// Nonvoter is a server that receives log entries but is not considered for
// elections or commitment purposes.
Nonvoter
// Staging is a server that acts like a nonvoter with one exception: once a
// staging server receives enough log entries to be sufficiently caught up to
// the leader's log, the leader will invoke a membership change to change
// the Staging server to a Voter.
Staging
)
func (s ServerSuffrage) String() string {
switch s {
case Voter:
return "Voter"
case Nonvoter:
return "Nonvoter"
case Staging:
return "Staging"
}
return "ServerSuffrage"
}
// ServerID is a unique string identifying a server for all time.
type ServerID string
// ServerAddress is a network address for a server that a transport can contact.
type ServerAddress string
// Server tracks the information about a single server in a configuration.
type Server struct {
// Suffrage determines whether the server gets a vote.
Suffrage ServerSuffrage
// ID is a unique string identifying this server for all time.
ID ServerID
// Address is its network address that a transport can contact.
Address ServerAddress
}
// Configuration tracks which servers are in the cluster, and whether they have
// votes. This should include the local server, if it's a member of the cluster.
// The servers are listed no particular order, but each should only appear once.
// These entries are appended to the log during membership changes.
type Configuration struct {
Servers []Server
}
// Clone makes a deep copy of a Configuration.
func (c *Configuration) Clone() (copy Configuration) {
copy.Servers = append(copy.Servers, c.Servers...)
return
}
// ConfigurationChangeCommand is the different ways to change the cluster
// configuration.
type ConfigurationChangeCommand uint8
const (
// AddStaging makes a server Staging unless its Voter.
AddStaging ConfigurationChangeCommand = iota
// AddNonvoter makes a server Nonvoter unless its Staging or Voter.
AddNonvoter
// DemoteVoter makes a server Nonvoter unless its absent.
DemoteVoter
// RemoveServer removes a server entirely from the cluster membership.
RemoveServer
// Promote is created automatically by a leader; it turns a Staging server
// into a Voter.
Promote
)
func (c ConfigurationChangeCommand) String() string {
switch c {
case AddStaging:
return "AddStaging"
case AddNonvoter:
return "AddNonvoter"
case DemoteVoter:
return "DemoteVoter"
case RemoveServer:
return "RemoveServer"
case Promote:
return "Promote"
}
return "ConfigurationChangeCommand"
}
// configurationChangeRequest describes a change that a leader would like to
// make to its current configuration. It's used only within a single server
// (never serialized into the log), as part of `configurationChangeFuture`.
type configurationChangeRequest struct {
command ConfigurationChangeCommand
serverID ServerID
serverAddress ServerAddress // only present for AddStaging, AddNonvoter
// prevIndex, if nonzero, is the index of the only configuration upon which
// this change may be applied; if another configuration entry has been
// added in the meantime, this request will fail.
prevIndex uint64
}
// configurations is state tracked on every server about its Configurations.
// Note that, per Diego's dissertation, there can be at most one uncommitted
// configuration at a time (the next configuration may not be created until the
// prior one has been committed).
//
// One downside to storing just two configurations is that if you try to take a
// snahpsot when your state machine hasn't yet applied the committedIndex, we
// have no record of the configuration that would logically fit into that
// snapshot. We disallow snapshots in that case now. An alternative approach,
// which LogCabin uses, is to track every configuration change in the
// log.
type configurations struct {
// committed is the latest configuration in the log/snapshot that has been
// committed (the one with the largest index).
committed Configuration
// committedIndex is the log index where 'committed' was written.
committedIndex uint64
// latest is the latest configuration in the log/snapshot (may be committed
// or uncommitted)
latest Configuration
// latestIndex is the log index where 'latest' was written.
latestIndex uint64
}
// Clone makes a deep copy of a configurations object.
func (c *configurations) Clone() (copy configurations) {
copy.committed = c.committed.Clone()
copy.committedIndex = c.committedIndex
copy.latest = c.latest.Clone()
copy.latestIndex = c.latestIndex
return
}
// hasVote returns true if the server identified by 'id' is a Voter in the
// provided Configuration.
func hasVote(configuration Configuration, id ServerID) bool {
for _, server := range configuration.Servers {
if server.ID == id {
return server.Suffrage == Voter
}
}
return false
}
// checkConfiguration tests a cluster membership configuration for common
// errors.
func checkConfiguration(configuration Configuration) error {
idSet := make(map[ServerID]bool)
addressSet := make(map[ServerAddress]bool)
var voters int
for _, server := range configuration.Servers {
if server.ID == "" {
return fmt.Errorf("Empty ID in configuration: %v", configuration)
}
if server.Address == "" {
return fmt.Errorf("Empty address in configuration: %v", server)
}
if idSet[server.ID] {
return fmt.Errorf("Found duplicate ID in configuration: %v", server.ID)
}
idSet[server.ID] = true
if addressSet[server.Address] {
return fmt.Errorf("Found duplicate address in configuration: %v", server.Address)
}
addressSet[server.Address] = true
if server.Suffrage == Voter {
voters++
}
}
if voters == 0 {
return fmt.Errorf("Need at least one voter in configuration: %v", configuration)
}
return nil
}
// nextConfiguration generates a new Configuration from the current one and a
// configuration change request. It's split from appendConfigurationEntry so
// that it can be unit tested easily.
func nextConfiguration(current Configuration, currentIndex uint64, change configurationChangeRequest) (Configuration, error) {
if change.prevIndex > 0 && change.prevIndex != currentIndex {
return Configuration{}, fmt.Errorf("Configuration changed since %v (latest is %v)", change.prevIndex, currentIndex)
}
configuration := current.Clone()
switch change.command {
case AddStaging:
// TODO: barf on new address?
newServer := Server{
// TODO: This should add the server as Staging, to be automatically
// promoted to Voter later. However, the promoton to Voter is not yet
// implemented, and doing so is not trivial with the way the leader loop
// coordinates with the replication goroutines today. So, for now, the
// server will have a vote right away, and the Promote case below is
// unused.
Suffrage: Voter,
ID: change.serverID,
Address: change.serverAddress,
}
found := false
for i, server := range configuration.Servers {
if server.ID == change.serverID {
if server.Suffrage == Voter {
configuration.Servers[i].Address = change.serverAddress
} else {
configuration.Servers[i] = newServer
}
found = true
break
}
}
if !found {
configuration.Servers = append(configuration.Servers, newServer)
}
case AddNonvoter:
newServer := Server{
Suffrage: Nonvoter,
ID: change.serverID,
Address: change.serverAddress,
}
found := false
for i, server := range configuration.Servers {
if server.ID == change.serverID {
if server.Suffrage != Nonvoter {
configuration.Servers[i].Address = change.serverAddress
} else {
configuration.Servers[i] = newServer
}
found = true
break
}
}
if !found {
configuration.Servers = append(configuration.Servers, newServer)
}
case DemoteVoter:
for i, server := range configuration.Servers {
if server.ID == change.serverID {
configuration.Servers[i].Suffrage = Nonvoter
break
}
}
case RemoveServer:
for i, server := range configuration.Servers {
if server.ID == change.serverID {
configuration.Servers = append(configuration.Servers[:i], configuration.Servers[i+1:]...)
break
}
}
case Promote:
for i, server := range configuration.Servers {
if server.ID == change.serverID && server.Suffrage == Staging {
configuration.Servers[i].Suffrage = Voter
break
}
}
}
// Make sure we didn't do something bad like remove the last voter
if err := checkConfiguration(configuration); err != nil {
return Configuration{}, err
}
return configuration, nil
}
// encodePeers is used to serialize a Configuration into the old peers format.
// This is here for backwards compatibility when operating with a mix of old
// servers and should be removed once we deprecate support for protocol version 1.
func encodePeers(configuration Configuration, trans Transport) []byte {
// Gather up all the voters, other suffrage types are not supported by
// this data format.
var encPeers [][]byte
for _, server := range configuration.Servers {
if server.Suffrage == Voter {
encPeers = append(encPeers, trans.EncodePeer(server.ID, server.Address))
}
}
// Encode the entire array.
buf, err := encodeMsgPack(encPeers)
if err != nil {
panic(fmt.Errorf("failed to encode peers: %v", err))
}
return buf.Bytes()
}
// decodePeers is used to deserialize an old list of peers into a Configuration.
// This is here for backwards compatibility with old log entries and snapshots;
// it should be removed eventually.
func decodePeers(buf []byte, trans Transport) Configuration {
// Decode the buffer first.
var encPeers [][]byte
if err := decodeMsgPack(buf, &encPeers); err != nil {
panic(fmt.Errorf("failed to decode peers: %v", err))
}
// Deserialize each peer.
var servers []Server
for _, enc := range encPeers {
p := trans.DecodePeer(enc)
servers = append(servers, Server{
Suffrage: Voter,
ID: ServerID(p),
Address: ServerAddress(p),
})
}
return Configuration{
Servers: servers,
}
}
// encodeConfiguration serializes a Configuration using MsgPack, or panics on
// errors.
func encodeConfiguration(configuration Configuration) []byte {
buf, err := encodeMsgPack(configuration)
if err != nil {
panic(fmt.Errorf("failed to encode configuration: %v", err))
}
return buf.Bytes()
}
// decodeConfiguration deserializes a Configuration using MsgPack, or panics on
// errors.
func decodeConfiguration(buf []byte) Configuration {
var configuration Configuration
if err := decodeMsgPack(buf, &configuration); err != nil {
panic(fmt.Errorf("failed to decode configuration: %v", err))
}
return configuration
}

49
vendor/github.com/hashicorp/raft/discard_snapshot.go generated vendored Normal file
View File

@@ -0,0 +1,49 @@
package raft
import (
"fmt"
"io"
)
// DiscardSnapshotStore is used to successfully snapshot while
// always discarding the snapshot. This is useful for when the
// log should be truncated but no snapshot should be retained.
// This should never be used for production use, and is only
// suitable for testing.
type DiscardSnapshotStore struct{}
type DiscardSnapshotSink struct{}
// NewDiscardSnapshotStore is used to create a new DiscardSnapshotStore.
func NewDiscardSnapshotStore() *DiscardSnapshotStore {
return &DiscardSnapshotStore{}
}
func (d *DiscardSnapshotStore) Create(version SnapshotVersion, index, term uint64,
configuration Configuration, configurationIndex uint64, trans Transport) (SnapshotSink, error) {
return &DiscardSnapshotSink{}, nil
}
func (d *DiscardSnapshotStore) List() ([]*SnapshotMeta, error) {
return nil, nil
}
func (d *DiscardSnapshotStore) Open(id string) (*SnapshotMeta, io.ReadCloser, error) {
return nil, nil, fmt.Errorf("open is not supported")
}
func (d *DiscardSnapshotSink) Write(b []byte) (int, error) {
return len(b), nil
}
func (d *DiscardSnapshotSink) Close() error {
return nil
}
func (d *DiscardSnapshotSink) ID() string {
return "discard"
}
func (d *DiscardSnapshotSink) Cancel() error {
return nil
}

528
vendor/github.com/hashicorp/raft/file_snapshot.go generated vendored Normal file
View File

@@ -0,0 +1,528 @@
package raft
import (
"bufio"
"bytes"
"encoding/json"
"fmt"
"hash"
"hash/crc64"
"io"
"io/ioutil"
"log"
"os"
"path/filepath"
"runtime"
"sort"
"strings"
"time"
)
const (
testPath = "permTest"
snapPath = "snapshots"
metaFilePath = "meta.json"
stateFilePath = "state.bin"
tmpSuffix = ".tmp"
)
// FileSnapshotStore implements the SnapshotStore interface and allows
// snapshots to be made on the local disk.
type FileSnapshotStore struct {
path string
retain int
logger *log.Logger
}
type snapMetaSlice []*fileSnapshotMeta
// FileSnapshotSink implements SnapshotSink with a file.
type FileSnapshotSink struct {
store *FileSnapshotStore
logger *log.Logger
dir string
parentDir string
meta fileSnapshotMeta
stateFile *os.File
stateHash hash.Hash64
buffered *bufio.Writer
closed bool
}
// fileSnapshotMeta is stored on disk. We also put a CRC
// on disk so that we can verify the snapshot.
type fileSnapshotMeta struct {
SnapshotMeta
CRC []byte
}
// bufferedFile is returned when we open a snapshot. This way
// reads are buffered and the file still gets closed.
type bufferedFile struct {
bh *bufio.Reader
fh *os.File
}
func (b *bufferedFile) Read(p []byte) (n int, err error) {
return b.bh.Read(p)
}
func (b *bufferedFile) Close() error {
return b.fh.Close()
}
// NewFileSnapshotStoreWithLogger creates a new FileSnapshotStore based
// on a base directory. The `retain` parameter controls how many
// snapshots are retained. Must be at least 1.
func NewFileSnapshotStoreWithLogger(base string, retain int, logger *log.Logger) (*FileSnapshotStore, error) {
if retain < 1 {
return nil, fmt.Errorf("must retain at least one snapshot")
}
if logger == nil {
logger = log.New(os.Stderr, "", log.LstdFlags)
}
// Ensure our path exists
path := filepath.Join(base, snapPath)
if err := os.MkdirAll(path, 0755); err != nil && !os.IsExist(err) {
return nil, fmt.Errorf("snapshot path not accessible: %v", err)
}
// Setup the store
store := &FileSnapshotStore{
path: path,
retain: retain,
logger: logger,
}
// Do a permissions test
if err := store.testPermissions(); err != nil {
return nil, fmt.Errorf("permissions test failed: %v", err)
}
return store, nil
}
// NewFileSnapshotStore creates a new FileSnapshotStore based
// on a base directory. The `retain` parameter controls how many
// snapshots are retained. Must be at least 1.
func NewFileSnapshotStore(base string, retain int, logOutput io.Writer) (*FileSnapshotStore, error) {
if logOutput == nil {
logOutput = os.Stderr
}
return NewFileSnapshotStoreWithLogger(base, retain, log.New(logOutput, "", log.LstdFlags))
}
// testPermissions tries to touch a file in our path to see if it works.
func (f *FileSnapshotStore) testPermissions() error {
path := filepath.Join(f.path, testPath)
fh, err := os.Create(path)
if err != nil {
return err
}
if err = fh.Close(); err != nil {
return err
}
if err = os.Remove(path); err != nil {
return err
}
return nil
}
// snapshotName generates a name for the snapshot.
func snapshotName(term, index uint64) string {
now := time.Now()
msec := now.UnixNano() / int64(time.Millisecond)
return fmt.Sprintf("%d-%d-%d", term, index, msec)
}
// Create is used to start a new snapshot
func (f *FileSnapshotStore) Create(version SnapshotVersion, index, term uint64,
configuration Configuration, configurationIndex uint64, trans Transport) (SnapshotSink, error) {
// We only support version 1 snapshots at this time.
if version != 1 {
return nil, fmt.Errorf("unsupported snapshot version %d", version)
}
// Create a new path
name := snapshotName(term, index)
path := filepath.Join(f.path, name+tmpSuffix)
f.logger.Printf("[INFO] snapshot: Creating new snapshot at %s", path)
// Make the directory
if err := os.MkdirAll(path, 0755); err != nil {
f.logger.Printf("[ERR] snapshot: Failed to make snapshot directory: %v", err)
return nil, err
}
// Create the sink
sink := &FileSnapshotSink{
store: f,
logger: f.logger,
dir: path,
parentDir: f.path,
meta: fileSnapshotMeta{
SnapshotMeta: SnapshotMeta{
Version: version,
ID: name,
Index: index,
Term: term,
Peers: encodePeers(configuration, trans),
Configuration: configuration,
ConfigurationIndex: configurationIndex,
},
CRC: nil,
},
}
// Write out the meta data
if err := sink.writeMeta(); err != nil {
f.logger.Printf("[ERR] snapshot: Failed to write metadata: %v", err)
return nil, err
}
// Open the state file
statePath := filepath.Join(path, stateFilePath)
fh, err := os.Create(statePath)
if err != nil {
f.logger.Printf("[ERR] snapshot: Failed to create state file: %v", err)
return nil, err
}
sink.stateFile = fh
// Create a CRC64 hash
sink.stateHash = crc64.New(crc64.MakeTable(crc64.ECMA))
// Wrap both the hash and file in a MultiWriter with buffering
multi := io.MultiWriter(sink.stateFile, sink.stateHash)
sink.buffered = bufio.NewWriter(multi)
// Done
return sink, nil
}
// List returns available snapshots in the store.
func (f *FileSnapshotStore) List() ([]*SnapshotMeta, error) {
// Get the eligible snapshots
snapshots, err := f.getSnapshots()
if err != nil {
f.logger.Printf("[ERR] snapshot: Failed to get snapshots: %v", err)
return nil, err
}
var snapMeta []*SnapshotMeta
for _, meta := range snapshots {
snapMeta = append(snapMeta, &meta.SnapshotMeta)
if len(snapMeta) == f.retain {
break
}
}
return snapMeta, nil
}
// getSnapshots returns all the known snapshots.
func (f *FileSnapshotStore) getSnapshots() ([]*fileSnapshotMeta, error) {
// Get the eligible snapshots
snapshots, err := ioutil.ReadDir(f.path)
if err != nil {
f.logger.Printf("[ERR] snapshot: Failed to scan snapshot dir: %v", err)
return nil, err
}
// Populate the metadata
var snapMeta []*fileSnapshotMeta
for _, snap := range snapshots {
// Ignore any files
if !snap.IsDir() {
continue
}
// Ignore any temporary snapshots
dirName := snap.Name()
if strings.HasSuffix(dirName, tmpSuffix) {
f.logger.Printf("[WARN] snapshot: Found temporary snapshot: %v", dirName)
continue
}
// Try to read the meta data
meta, err := f.readMeta(dirName)
if err != nil {
f.logger.Printf("[WARN] snapshot: Failed to read metadata for %v: %v", dirName, err)
continue
}
// Make sure we can understand this version.
if meta.Version < SnapshotVersionMin || meta.Version > SnapshotVersionMax {
f.logger.Printf("[WARN] snapshot: Snapshot version for %v not supported: %d", dirName, meta.Version)
continue
}
// Append, but only return up to the retain count
snapMeta = append(snapMeta, meta)
}
// Sort the snapshot, reverse so we get new -> old
sort.Sort(sort.Reverse(snapMetaSlice(snapMeta)))
return snapMeta, nil
}
// readMeta is used to read the meta data for a given named backup
func (f *FileSnapshotStore) readMeta(name string) (*fileSnapshotMeta, error) {
// Open the meta file
metaPath := filepath.Join(f.path, name, metaFilePath)
fh, err := os.Open(metaPath)
if err != nil {
return nil, err
}
defer fh.Close()
// Buffer the file IO
buffered := bufio.NewReader(fh)
// Read in the JSON
meta := &fileSnapshotMeta{}
dec := json.NewDecoder(buffered)
if err := dec.Decode(meta); err != nil {
return nil, err
}
return meta, nil
}
// Open takes a snapshot ID and returns a ReadCloser for that snapshot.
func (f *FileSnapshotStore) Open(id string) (*SnapshotMeta, io.ReadCloser, error) {
// Get the metadata
meta, err := f.readMeta(id)
if err != nil {
f.logger.Printf("[ERR] snapshot: Failed to get meta data to open snapshot: %v", err)
return nil, nil, err
}
// Open the state file
statePath := filepath.Join(f.path, id, stateFilePath)
fh, err := os.Open(statePath)
if err != nil {
f.logger.Printf("[ERR] snapshot: Failed to open state file: %v", err)
return nil, nil, err
}
// Create a CRC64 hash
stateHash := crc64.New(crc64.MakeTable(crc64.ECMA))
// Compute the hash
_, err = io.Copy(stateHash, fh)
if err != nil {
f.logger.Printf("[ERR] snapshot: Failed to read state file: %v", err)
fh.Close()
return nil, nil, err
}
// Verify the hash
computed := stateHash.Sum(nil)
if bytes.Compare(meta.CRC, computed) != 0 {
f.logger.Printf("[ERR] snapshot: CRC checksum failed (stored: %v computed: %v)",
meta.CRC, computed)
fh.Close()
return nil, nil, fmt.Errorf("CRC mismatch")
}
// Seek to the start
if _, err := fh.Seek(0, 0); err != nil {
f.logger.Printf("[ERR] snapshot: State file seek failed: %v", err)
fh.Close()
return nil, nil, err
}
// Return a buffered file
buffered := &bufferedFile{
bh: bufio.NewReader(fh),
fh: fh,
}
return &meta.SnapshotMeta, buffered, nil
}
// ReapSnapshots reaps any snapshots beyond the retain count.
func (f *FileSnapshotStore) ReapSnapshots() error {
snapshots, err := f.getSnapshots()
if err != nil {
f.logger.Printf("[ERR] snapshot: Failed to get snapshots: %v", err)
return err
}
for i := f.retain; i < len(snapshots); i++ {
path := filepath.Join(f.path, snapshots[i].ID)
f.logger.Printf("[INFO] snapshot: reaping snapshot %v", path)
if err := os.RemoveAll(path); err != nil {
f.logger.Printf("[ERR] snapshot: Failed to reap snapshot %v: %v", path, err)
return err
}
}
return nil
}
// ID returns the ID of the snapshot, can be used with Open()
// after the snapshot is finalized.
func (s *FileSnapshotSink) ID() string {
return s.meta.ID
}
// Write is used to append to the state file. We write to the
// buffered IO object to reduce the amount of context switches.
func (s *FileSnapshotSink) Write(b []byte) (int, error) {
return s.buffered.Write(b)
}
// Close is used to indicate a successful end.
func (s *FileSnapshotSink) Close() error {
// Make sure close is idempotent
if s.closed {
return nil
}
s.closed = true
// Close the open handles
if err := s.finalize(); err != nil {
s.logger.Printf("[ERR] snapshot: Failed to finalize snapshot: %v", err)
if delErr := os.RemoveAll(s.dir); delErr != nil {
s.logger.Printf("[ERR] snapshot: Failed to delete temporary snapshot directory at path %v: %v", s.dir, delErr)
return delErr
}
return err
}
// Write out the meta data
if err := s.writeMeta(); err != nil {
s.logger.Printf("[ERR] snapshot: Failed to write metadata: %v", err)
return err
}
// Move the directory into place
newPath := strings.TrimSuffix(s.dir, tmpSuffix)
if err := os.Rename(s.dir, newPath); err != nil {
s.logger.Printf("[ERR] snapshot: Failed to move snapshot into place: %v", err)
return err
}
if runtime.GOOS != "windows" { //skipping fsync for directory entry edits on Windows, only needed for *nix style file systems
parentFH, err := os.Open(s.parentDir)
defer parentFH.Close()
if err != nil {
s.logger.Printf("[ERR] snapshot: Failed to open snapshot parent directory %v, error: %v", s.parentDir, err)
return err
}
if err = parentFH.Sync(); err != nil {
s.logger.Printf("[ERR] snapshot: Failed syncing parent directory %v, error: %v", s.parentDir, err)
return err
}
}
// Reap any old snapshots
if err := s.store.ReapSnapshots(); err != nil {
return err
}
return nil
}
// Cancel is used to indicate an unsuccessful end.
func (s *FileSnapshotSink) Cancel() error {
// Make sure close is idempotent
if s.closed {
return nil
}
s.closed = true
// Close the open handles
if err := s.finalize(); err != nil {
s.logger.Printf("[ERR] snapshot: Failed to finalize snapshot: %v", err)
return err
}
// Attempt to remove all artifacts
return os.RemoveAll(s.dir)
}
// finalize is used to close all of our resources.
func (s *FileSnapshotSink) finalize() error {
// Flush any remaining data
if err := s.buffered.Flush(); err != nil {
return err
}
// Sync to force fsync to disk
if err := s.stateFile.Sync(); err != nil {
return err
}
// Get the file size
stat, statErr := s.stateFile.Stat()
// Close the file
if err := s.stateFile.Close(); err != nil {
return err
}
// Set the file size, check after we close
if statErr != nil {
return statErr
}
s.meta.Size = stat.Size()
// Set the CRC
s.meta.CRC = s.stateHash.Sum(nil)
return nil
}
// writeMeta is used to write out the metadata we have.
func (s *FileSnapshotSink) writeMeta() error {
// Open the meta file
metaPath := filepath.Join(s.dir, metaFilePath)
fh, err := os.Create(metaPath)
if err != nil {
return err
}
defer fh.Close()
// Buffer the file IO
buffered := bufio.NewWriter(fh)
// Write out as JSON
enc := json.NewEncoder(buffered)
if err := enc.Encode(&s.meta); err != nil {
return err
}
if err = buffered.Flush(); err != nil {
return err
}
if err = fh.Sync(); err != nil {
return err
}
return nil
}
// Implement the sort interface for []*fileSnapshotMeta.
func (s snapMetaSlice) Len() int {
return len(s)
}
func (s snapMetaSlice) Less(i, j int) bool {
if s[i].Term != s[j].Term {
return s[i].Term < s[j].Term
}
if s[i].Index != s[j].Index {
return s[i].Index < s[j].Index
}
return s[i].ID < s[j].ID
}
func (s snapMetaSlice) Swap(i, j int) {
s[i], s[j] = s[j], s[i]
}

136
vendor/github.com/hashicorp/raft/fsm.go generated vendored Normal file
View File

@@ -0,0 +1,136 @@
package raft
import (
"fmt"
"io"
"time"
"github.com/armon/go-metrics"
)
// FSM provides an interface that can be implemented by
// clients to make use of the replicated log.
type FSM interface {
// Apply log is invoked once a log entry is committed.
// It returns a value which will be made available in the
// ApplyFuture returned by Raft.Apply method if that
// method was called on the same Raft node as the FSM.
Apply(*Log) interface{}
// Snapshot is used to support log compaction. This call should
// return an FSMSnapshot which can be used to save a point-in-time
// snapshot of the FSM. Apply and Snapshot are not called in multiple
// threads, but Apply will be called concurrently with Persist. This means
// the FSM should be implemented in a fashion that allows for concurrent
// updates while a snapshot is happening.
Snapshot() (FSMSnapshot, error)
// Restore is used to restore an FSM from a snapshot. It is not called
// concurrently with any other command. The FSM must discard all previous
// state.
Restore(io.ReadCloser) error
}
// FSMSnapshot is returned by an FSM in response to a Snapshot
// It must be safe to invoke FSMSnapshot methods with concurrent
// calls to Apply.
type FSMSnapshot interface {
// Persist should dump all necessary state to the WriteCloser 'sink',
// and call sink.Close() when finished or call sink.Cancel() on error.
Persist(sink SnapshotSink) error
// Release is invoked when we are finished with the snapshot.
Release()
}
// runFSM is a long running goroutine responsible for applying logs
// to the FSM. This is done async of other logs since we don't want
// the FSM to block our internal operations.
func (r *Raft) runFSM() {
var lastIndex, lastTerm uint64
commit := func(req *commitTuple) {
// Apply the log if a command
var resp interface{}
if req.log.Type == LogCommand {
start := time.Now()
resp = r.fsm.Apply(req.log)
metrics.MeasureSince([]string{"raft", "fsm", "apply"}, start)
}
// Update the indexes
lastIndex = req.log.Index
lastTerm = req.log.Term
// Invoke the future if given
if req.future != nil {
req.future.response = resp
req.future.respond(nil)
}
}
restore := func(req *restoreFuture) {
// Open the snapshot
meta, source, err := r.snapshots.Open(req.ID)
if err != nil {
req.respond(fmt.Errorf("failed to open snapshot %v: %v", req.ID, err))
return
}
// Attempt to restore
start := time.Now()
if err := r.fsm.Restore(source); err != nil {
req.respond(fmt.Errorf("failed to restore snapshot %v: %v", req.ID, err))
source.Close()
return
}
source.Close()
metrics.MeasureSince([]string{"raft", "fsm", "restore"}, start)
// Update the last index and term
lastIndex = meta.Index
lastTerm = meta.Term
req.respond(nil)
}
snapshot := func(req *reqSnapshotFuture) {
// Is there something to snapshot?
if lastIndex == 0 {
req.respond(ErrNothingNewToSnapshot)
return
}
// Start a snapshot
start := time.Now()
snap, err := r.fsm.Snapshot()
metrics.MeasureSince([]string{"raft", "fsm", "snapshot"}, start)
// Respond to the request
req.index = lastIndex
req.term = lastTerm
req.snapshot = snap
req.respond(err)
}
for {
select {
case ptr := <-r.fsmMutateCh:
switch req := ptr.(type) {
case *commitTuple:
commit(req)
case *restoreFuture:
restore(req)
default:
panic(fmt.Errorf("bad type passed to fsmMutateCh: %#v", ptr))
}
case req := <-r.fsmSnapshotCh:
snapshot(req)
case <-r.shutdownCh:
return
}
}
}

289
vendor/github.com/hashicorp/raft/future.go generated vendored Normal file
View File

@@ -0,0 +1,289 @@
package raft
import (
"fmt"
"io"
"sync"
"time"
)
// Future is used to represent an action that may occur in the future.
type Future interface {
// Error blocks until the future arrives and then
// returns the error status of the future.
// This may be called any number of times - all
// calls will return the same value.
// Note that it is not OK to call this method
// twice concurrently on the same Future instance.
Error() error
}
// IndexFuture is used for future actions that can result in a raft log entry
// being created.
type IndexFuture interface {
Future
// Index holds the index of the newly applied log entry.
// This must not be called until after the Error method has returned.
Index() uint64
}
// ApplyFuture is used for Apply and can return the FSM response.
type ApplyFuture interface {
IndexFuture
// Response returns the FSM response as returned
// by the FSM.Apply method. This must not be called
// until after the Error method has returned.
Response() interface{}
}
// ConfigurationFuture is used for GetConfiguration and can return the
// latest configuration in use by Raft.
type ConfigurationFuture interface {
IndexFuture
// Configuration contains the latest configuration. This must
// not be called until after the Error method has returned.
Configuration() Configuration
}
// SnapshotFuture is used for waiting on a user-triggered snapshot to complete.
type SnapshotFuture interface {
Future
// Open is a function you can call to access the underlying snapshot and
// its metadata. This must not be called until after the Error method
// has returned.
Open() (*SnapshotMeta, io.ReadCloser, error)
}
// errorFuture is used to return a static error.
type errorFuture struct {
err error
}
func (e errorFuture) Error() error {
return e.err
}
func (e errorFuture) Response() interface{} {
return nil
}
func (e errorFuture) Index() uint64 {
return 0
}
// deferError can be embedded to allow a future
// to provide an error in the future.
type deferError struct {
err error
errCh chan error
responded bool
}
func (d *deferError) init() {
d.errCh = make(chan error, 1)
}
func (d *deferError) Error() error {
if d.err != nil {
// Note that when we've received a nil error, this
// won't trigger, but the channel is closed after
// send so we'll still return nil below.
return d.err
}
if d.errCh == nil {
panic("waiting for response on nil channel")
}
d.err = <-d.errCh
return d.err
}
func (d *deferError) respond(err error) {
if d.errCh == nil {
return
}
if d.responded {
return
}
d.errCh <- err
close(d.errCh)
d.responded = true
}
// There are several types of requests that cause a configuration entry to
// be appended to the log. These are encoded here for leaderLoop() to process.
// This is internal to a single server.
type configurationChangeFuture struct {
logFuture
req configurationChangeRequest
}
// bootstrapFuture is used to attempt a live bootstrap of the cluster. See the
// Raft object's BootstrapCluster member function for more details.
type bootstrapFuture struct {
deferError
// configuration is the proposed bootstrap configuration to apply.
configuration Configuration
}
// logFuture is used to apply a log entry and waits until
// the log is considered committed.
type logFuture struct {
deferError
log Log
response interface{}
dispatch time.Time
}
func (l *logFuture) Response() interface{} {
return l.response
}
func (l *logFuture) Index() uint64 {
return l.log.Index
}
type shutdownFuture struct {
raft *Raft
}
func (s *shutdownFuture) Error() error {
if s.raft == nil {
return nil
}
s.raft.waitShutdown()
if closeable, ok := s.raft.trans.(WithClose); ok {
closeable.Close()
}
return nil
}
// userSnapshotFuture is used for waiting on a user-triggered snapshot to
// complete.
type userSnapshotFuture struct {
deferError
// opener is a function used to open the snapshot. This is filled in
// once the future returns with no error.
opener func() (*SnapshotMeta, io.ReadCloser, error)
}
// Open is a function you can call to access the underlying snapshot and its
// metadata.
func (u *userSnapshotFuture) Open() (*SnapshotMeta, io.ReadCloser, error) {
if u.opener == nil {
return nil, nil, fmt.Errorf("no snapshot available")
} else {
// Invalidate the opener so it can't get called multiple times,
// which isn't generally safe.
defer func() {
u.opener = nil
}()
return u.opener()
}
}
// userRestoreFuture is used for waiting on a user-triggered restore of an
// external snapshot to complete.
type userRestoreFuture struct {
deferError
// meta is the metadata that belongs with the snapshot.
meta *SnapshotMeta
// reader is the interface to read the snapshot contents from.
reader io.Reader
}
// reqSnapshotFuture is used for requesting a snapshot start.
// It is only used internally.
type reqSnapshotFuture struct {
deferError
// snapshot details provided by the FSM runner before responding
index uint64
term uint64
snapshot FSMSnapshot
}
// restoreFuture is used for requesting an FSM to perform a
// snapshot restore. Used internally only.
type restoreFuture struct {
deferError
ID string
}
// verifyFuture is used to verify the current node is still
// the leader. This is to prevent a stale read.
type verifyFuture struct {
deferError
notifyCh chan *verifyFuture
quorumSize int
votes int
voteLock sync.Mutex
}
// configurationsFuture is used to retrieve the current configurations. This is
// used to allow safe access to this information outside of the main thread.
type configurationsFuture struct {
deferError
configurations configurations
}
// Configuration returns the latest configuration in use by Raft.
func (c *configurationsFuture) Configuration() Configuration {
return c.configurations.latest
}
// Index returns the index of the latest configuration in use by Raft.
func (c *configurationsFuture) Index() uint64 {
return c.configurations.latestIndex
}
// vote is used to respond to a verifyFuture.
// This may block when responding on the notifyCh.
func (v *verifyFuture) vote(leader bool) {
v.voteLock.Lock()
defer v.voteLock.Unlock()
// Guard against having notified already
if v.notifyCh == nil {
return
}
if leader {
v.votes++
if v.votes >= v.quorumSize {
v.notifyCh <- v
v.notifyCh = nil
}
} else {
v.notifyCh <- v
v.notifyCh = nil
}
}
// appendFuture is used for waiting on a pipelined append
// entries RPC.
type appendFuture struct {
deferError
start time.Time
args *AppendEntriesRequest
resp *AppendEntriesResponse
}
func (a *appendFuture) Start() time.Time {
return a.start
}
func (a *appendFuture) Request() *AppendEntriesRequest {
return a.args
}
func (a *appendFuture) Response() *AppendEntriesResponse {
return a.resp
}

106
vendor/github.com/hashicorp/raft/inmem_snapshot.go generated vendored Normal file
View File

@@ -0,0 +1,106 @@
package raft
import (
"bytes"
"fmt"
"io"
"io/ioutil"
"sync"
)
// InmemSnapshotStore implements the SnapshotStore interface and
// retains only the most recent snapshot
type InmemSnapshotStore struct {
latest *InmemSnapshotSink
hasSnapshot bool
sync.RWMutex
}
// InmemSnapshotSink implements SnapshotSink in memory
type InmemSnapshotSink struct {
meta SnapshotMeta
contents *bytes.Buffer
}
// NewInmemSnapshotStore creates a blank new InmemSnapshotStore
func NewInmemSnapshotStore() *InmemSnapshotStore {
return &InmemSnapshotStore{
latest: &InmemSnapshotSink{
contents: &bytes.Buffer{},
},
}
}
// Create replaces the stored snapshot with a new one using the given args
func (m *InmemSnapshotStore) Create(version SnapshotVersion, index, term uint64,
configuration Configuration, configurationIndex uint64, trans Transport) (SnapshotSink, error) {
// We only support version 1 snapshots at this time.
if version != 1 {
return nil, fmt.Errorf("unsupported snapshot version %d", version)
}
name := snapshotName(term, index)
m.Lock()
defer m.Unlock()
sink := &InmemSnapshotSink{
meta: SnapshotMeta{
Version: version,
ID: name,
Index: index,
Term: term,
Peers: encodePeers(configuration, trans),
Configuration: configuration,
ConfigurationIndex: configurationIndex,
},
contents: &bytes.Buffer{},
}
m.hasSnapshot = true
m.latest = sink
return sink, nil
}
// List returns the latest snapshot taken
func (m *InmemSnapshotStore) List() ([]*SnapshotMeta, error) {
m.RLock()
defer m.RUnlock()
if !m.hasSnapshot {
return []*SnapshotMeta{}, nil
}
return []*SnapshotMeta{&m.latest.meta}, nil
}
// Open wraps an io.ReadCloser around the snapshot contents
func (m *InmemSnapshotStore) Open(id string) (*SnapshotMeta, io.ReadCloser, error) {
m.RLock()
defer m.RUnlock()
if m.latest.meta.ID != id {
return nil, nil, fmt.Errorf("[ERR] snapshot: failed to open snapshot id: %s", id)
}
return &m.latest.meta, ioutil.NopCloser(m.latest.contents), nil
}
// Write appends the given bytes to the snapshot contents
func (s *InmemSnapshotSink) Write(p []byte) (n int, err error) {
written, err := io.Copy(s.contents, bytes.NewReader(p))
s.meta.Size += written
return int(written), err
}
// Close updates the Size and is otherwise a no-op
func (s *InmemSnapshotSink) Close() error {
return nil
}
func (s *InmemSnapshotSink) ID() string {
return s.meta.ID
}
func (s *InmemSnapshotSink) Cancel() error {
return nil
}

125
vendor/github.com/hashicorp/raft/inmem_store.go generated vendored Normal file
View File

@@ -0,0 +1,125 @@
package raft
import (
"sync"
)
// InmemStore implements the LogStore and StableStore interface.
// It should NOT EVER be used for production. It is used only for
// unit tests. Use the MDBStore implementation instead.
type InmemStore struct {
l sync.RWMutex
lowIndex uint64
highIndex uint64
logs map[uint64]*Log
kv map[string][]byte
kvInt map[string]uint64
}
// NewInmemStore returns a new in-memory backend. Do not ever
// use for production. Only for testing.
func NewInmemStore() *InmemStore {
i := &InmemStore{
logs: make(map[uint64]*Log),
kv: make(map[string][]byte),
kvInt: make(map[string]uint64),
}
return i
}
// FirstIndex implements the LogStore interface.
func (i *InmemStore) FirstIndex() (uint64, error) {
i.l.RLock()
defer i.l.RUnlock()
return i.lowIndex, nil
}
// LastIndex implements the LogStore interface.
func (i *InmemStore) LastIndex() (uint64, error) {
i.l.RLock()
defer i.l.RUnlock()
return i.highIndex, nil
}
// GetLog implements the LogStore interface.
func (i *InmemStore) GetLog(index uint64, log *Log) error {
i.l.RLock()
defer i.l.RUnlock()
l, ok := i.logs[index]
if !ok {
return ErrLogNotFound
}
*log = *l
return nil
}
// StoreLog implements the LogStore interface.
func (i *InmemStore) StoreLog(log *Log) error {
return i.StoreLogs([]*Log{log})
}
// StoreLogs implements the LogStore interface.
func (i *InmemStore) StoreLogs(logs []*Log) error {
i.l.Lock()
defer i.l.Unlock()
for _, l := range logs {
i.logs[l.Index] = l
if i.lowIndex == 0 {
i.lowIndex = l.Index
}
if l.Index > i.highIndex {
i.highIndex = l.Index
}
}
return nil
}
// DeleteRange implements the LogStore interface.
func (i *InmemStore) DeleteRange(min, max uint64) error {
i.l.Lock()
defer i.l.Unlock()
for j := min; j <= max; j++ {
delete(i.logs, j)
}
if min <= i.lowIndex {
i.lowIndex = max + 1
}
if max >= i.highIndex {
i.highIndex = min - 1
}
if i.lowIndex > i.highIndex {
i.lowIndex = 0
i.highIndex = 0
}
return nil
}
// Set implements the StableStore interface.
func (i *InmemStore) Set(key []byte, val []byte) error {
i.l.Lock()
defer i.l.Unlock()
i.kv[string(key)] = val
return nil
}
// Get implements the StableStore interface.
func (i *InmemStore) Get(key []byte) ([]byte, error) {
i.l.RLock()
defer i.l.RUnlock()
return i.kv[string(key)], nil
}
// SetUint64 implements the StableStore interface.
func (i *InmemStore) SetUint64(key []byte, val uint64) error {
i.l.Lock()
defer i.l.Unlock()
i.kvInt[string(key)] = val
return nil
}
// GetUint64 implements the StableStore interface.
func (i *InmemStore) GetUint64(key []byte) (uint64, error) {
i.l.RLock()
defer i.l.RUnlock()
return i.kvInt[string(key)], nil
}

322
vendor/github.com/hashicorp/raft/inmem_transport.go generated vendored Normal file
View File

@@ -0,0 +1,322 @@
package raft
import (
"fmt"
"io"
"sync"
"time"
)
// NewInmemAddr returns a new in-memory addr with
// a randomly generate UUID as the ID.
func NewInmemAddr() ServerAddress {
return ServerAddress(generateUUID())
}
// inmemPipeline is used to pipeline requests for the in-mem transport.
type inmemPipeline struct {
trans *InmemTransport
peer *InmemTransport
peerAddr ServerAddress
doneCh chan AppendFuture
inprogressCh chan *inmemPipelineInflight
shutdown bool
shutdownCh chan struct{}
shutdownLock sync.Mutex
}
type inmemPipelineInflight struct {
future *appendFuture
respCh <-chan RPCResponse
}
// InmemTransport Implements the Transport interface, to allow Raft to be
// tested in-memory without going over a network.
type InmemTransport struct {
sync.RWMutex
consumerCh chan RPC
localAddr ServerAddress
peers map[ServerAddress]*InmemTransport
pipelines []*inmemPipeline
timeout time.Duration
}
// NewInmemTransport is used to initialize a new transport
// and generates a random local address if none is specified
func NewInmemTransport(addr ServerAddress) (ServerAddress, *InmemTransport) {
if string(addr) == "" {
addr = NewInmemAddr()
}
trans := &InmemTransport{
consumerCh: make(chan RPC, 16),
localAddr: addr,
peers: make(map[ServerAddress]*InmemTransport),
timeout: 50 * time.Millisecond,
}
return addr, trans
}
// SetHeartbeatHandler is used to set optional fast-path for
// heartbeats, not supported for this transport.
func (i *InmemTransport) SetHeartbeatHandler(cb func(RPC)) {
}
// Consumer implements the Transport interface.
func (i *InmemTransport) Consumer() <-chan RPC {
return i.consumerCh
}
// LocalAddr implements the Transport interface.
func (i *InmemTransport) LocalAddr() ServerAddress {
return i.localAddr
}
// AppendEntriesPipeline returns an interface that can be used to pipeline
// AppendEntries requests.
func (i *InmemTransport) AppendEntriesPipeline(id ServerID, target ServerAddress) (AppendPipeline, error) {
i.RLock()
peer, ok := i.peers[target]
i.RUnlock()
if !ok {
return nil, fmt.Errorf("failed to connect to peer: %v", target)
}
pipeline := newInmemPipeline(i, peer, target)
i.Lock()
i.pipelines = append(i.pipelines, pipeline)
i.Unlock()
return pipeline, nil
}
// AppendEntries implements the Transport interface.
func (i *InmemTransport) AppendEntries(id ServerID, target ServerAddress, args *AppendEntriesRequest, resp *AppendEntriesResponse) error {
rpcResp, err := i.makeRPC(target, args, nil, i.timeout)
if err != nil {
return err
}
// Copy the result back
out := rpcResp.Response.(*AppendEntriesResponse)
*resp = *out
return nil
}
// RequestVote implements the Transport interface.
func (i *InmemTransport) RequestVote(id ServerID, target ServerAddress, args *RequestVoteRequest, resp *RequestVoteResponse) error {
rpcResp, err := i.makeRPC(target, args, nil, i.timeout)
if err != nil {
return err
}
// Copy the result back
out := rpcResp.Response.(*RequestVoteResponse)
*resp = *out
return nil
}
// InstallSnapshot implements the Transport interface.
func (i *InmemTransport) InstallSnapshot(id ServerID, target ServerAddress, args *InstallSnapshotRequest, resp *InstallSnapshotResponse, data io.Reader) error {
rpcResp, err := i.makeRPC(target, args, data, 10*i.timeout)
if err != nil {
return err
}
// Copy the result back
out := rpcResp.Response.(*InstallSnapshotResponse)
*resp = *out
return nil
}
func (i *InmemTransport) makeRPC(target ServerAddress, args interface{}, r io.Reader, timeout time.Duration) (rpcResp RPCResponse, err error) {
i.RLock()
peer, ok := i.peers[target]
i.RUnlock()
if !ok {
err = fmt.Errorf("failed to connect to peer: %v", target)
return
}
// Send the RPC over
respCh := make(chan RPCResponse)
peer.consumerCh <- RPC{
Command: args,
Reader: r,
RespChan: respCh,
}
// Wait for a response
select {
case rpcResp = <-respCh:
if rpcResp.Error != nil {
err = rpcResp.Error
}
case <-time.After(timeout):
err = fmt.Errorf("command timed out")
}
return
}
// EncodePeer implements the Transport interface.
func (i *InmemTransport) EncodePeer(id ServerID, p ServerAddress) []byte {
return []byte(p)
}
// DecodePeer implements the Transport interface.
func (i *InmemTransport) DecodePeer(buf []byte) ServerAddress {
return ServerAddress(buf)
}
// Connect is used to connect this transport to another transport for
// a given peer name. This allows for local routing.
func (i *InmemTransport) Connect(peer ServerAddress, t Transport) {
trans := t.(*InmemTransport)
i.Lock()
defer i.Unlock()
i.peers[peer] = trans
}
// Disconnect is used to remove the ability to route to a given peer.
func (i *InmemTransport) Disconnect(peer ServerAddress) {
i.Lock()
defer i.Unlock()
delete(i.peers, peer)
// Disconnect any pipelines
n := len(i.pipelines)
for idx := 0; idx < n; idx++ {
if i.pipelines[idx].peerAddr == peer {
i.pipelines[idx].Close()
i.pipelines[idx], i.pipelines[n-1] = i.pipelines[n-1], nil
idx--
n--
}
}
i.pipelines = i.pipelines[:n]
}
// DisconnectAll is used to remove all routes to peers.
func (i *InmemTransport) DisconnectAll() {
i.Lock()
defer i.Unlock()
i.peers = make(map[ServerAddress]*InmemTransport)
// Handle pipelines
for _, pipeline := range i.pipelines {
pipeline.Close()
}
i.pipelines = nil
}
// Close is used to permanently disable the transport
func (i *InmemTransport) Close() error {
i.DisconnectAll()
return nil
}
func newInmemPipeline(trans *InmemTransport, peer *InmemTransport, addr ServerAddress) *inmemPipeline {
i := &inmemPipeline{
trans: trans,
peer: peer,
peerAddr: addr,
doneCh: make(chan AppendFuture, 16),
inprogressCh: make(chan *inmemPipelineInflight, 16),
shutdownCh: make(chan struct{}),
}
go i.decodeResponses()
return i
}
func (i *inmemPipeline) decodeResponses() {
timeout := i.trans.timeout
for {
select {
case inp := <-i.inprogressCh:
var timeoutCh <-chan time.Time
if timeout > 0 {
timeoutCh = time.After(timeout)
}
select {
case rpcResp := <-inp.respCh:
// Copy the result back
*inp.future.resp = *rpcResp.Response.(*AppendEntriesResponse)
inp.future.respond(rpcResp.Error)
select {
case i.doneCh <- inp.future:
case <-i.shutdownCh:
return
}
case <-timeoutCh:
inp.future.respond(fmt.Errorf("command timed out"))
select {
case i.doneCh <- inp.future:
case <-i.shutdownCh:
return
}
case <-i.shutdownCh:
return
}
case <-i.shutdownCh:
return
}
}
}
func (i *inmemPipeline) AppendEntries(args *AppendEntriesRequest, resp *AppendEntriesResponse) (AppendFuture, error) {
// Create a new future
future := &appendFuture{
start: time.Now(),
args: args,
resp: resp,
}
future.init()
// Handle a timeout
var timeout <-chan time.Time
if i.trans.timeout > 0 {
timeout = time.After(i.trans.timeout)
}
// Send the RPC over
respCh := make(chan RPCResponse, 1)
rpc := RPC{
Command: args,
RespChan: respCh,
}
select {
case i.peer.consumerCh <- rpc:
case <-timeout:
return nil, fmt.Errorf("command enqueue timeout")
case <-i.shutdownCh:
return nil, ErrPipelineShutdown
}
// Send to be decoded
select {
case i.inprogressCh <- &inmemPipelineInflight{future, respCh}:
return future, nil
case <-i.shutdownCh:
return nil, ErrPipelineShutdown
}
}
func (i *inmemPipeline) Consumer() <-chan AppendFuture {
return i.doneCh
}
func (i *inmemPipeline) Close() error {
i.shutdownLock.Lock()
defer i.shutdownLock.Unlock()
if i.shutdown {
return nil
}
i.shutdown = true
close(i.shutdownCh)
return nil
}

72
vendor/github.com/hashicorp/raft/log.go generated vendored Normal file
View File

@@ -0,0 +1,72 @@
package raft
// LogType describes various types of log entries.
type LogType uint8
const (
// LogCommand is applied to a user FSM.
LogCommand LogType = iota
// LogNoop is used to assert leadership.
LogNoop
// LogAddPeer is used to add a new peer. This should only be used with
// older protocol versions designed to be compatible with unversioned
// Raft servers. See comments in config.go for details.
LogAddPeerDeprecated
// LogRemovePeer is used to remove an existing peer. This should only be
// used with older protocol versions designed to be compatible with
// unversioned Raft servers. See comments in config.go for details.
LogRemovePeerDeprecated
// LogBarrier is used to ensure all preceding operations have been
// applied to the FSM. It is similar to LogNoop, but instead of returning
// once committed, it only returns once the FSM manager acks it. Otherwise
// it is possible there are operations committed but not yet applied to
// the FSM.
LogBarrier
// LogConfiguration establishes a membership change configuration. It is
// created when a server is added, removed, promoted, etc. Only used
// when protocol version 1 or greater is in use.
LogConfiguration
)
// Log entries are replicated to all members of the Raft cluster
// and form the heart of the replicated state machine.
type Log struct {
// Index holds the index of the log entry.
Index uint64
// Term holds the election term of the log entry.
Term uint64
// Type holds the type of the log entry.
Type LogType
// Data holds the log entry's type-specific data.
Data []byte
}
// LogStore is used to provide an interface for storing
// and retrieving logs in a durable fashion.
type LogStore interface {
// FirstIndex returns the first index written. 0 for no entries.
FirstIndex() (uint64, error)
// LastIndex returns the last index written. 0 for no entries.
LastIndex() (uint64, error)
// GetLog gets a log entry at a given index.
GetLog(index uint64, log *Log) error
// StoreLog stores a log entry.
StoreLog(log *Log) error
// StoreLogs stores multiple log entries.
StoreLogs(logs []*Log) error
// DeleteRange deletes a range of log entries. The range is inclusive.
DeleteRange(min, max uint64) error
}

79
vendor/github.com/hashicorp/raft/log_cache.go generated vendored Normal file
View File

@@ -0,0 +1,79 @@
package raft
import (
"fmt"
"sync"
)
// LogCache wraps any LogStore implementation to provide an
// in-memory ring buffer. This is used to cache access to
// the recently written entries. For implementations that do not
// cache themselves, this can provide a substantial boost by
// avoiding disk I/O on recent entries.
type LogCache struct {
store LogStore
cache []*Log
l sync.RWMutex
}
// NewLogCache is used to create a new LogCache with the
// given capacity and backend store.
func NewLogCache(capacity int, store LogStore) (*LogCache, error) {
if capacity <= 0 {
return nil, fmt.Errorf("capacity must be positive")
}
c := &LogCache{
store: store,
cache: make([]*Log, capacity),
}
return c, nil
}
func (c *LogCache) GetLog(idx uint64, log *Log) error {
// Check the buffer for an entry
c.l.RLock()
cached := c.cache[idx%uint64(len(c.cache))]
c.l.RUnlock()
// Check if entry is valid
if cached != nil && cached.Index == idx {
*log = *cached
return nil
}
// Forward request on cache miss
return c.store.GetLog(idx, log)
}
func (c *LogCache) StoreLog(log *Log) error {
return c.StoreLogs([]*Log{log})
}
func (c *LogCache) StoreLogs(logs []*Log) error {
// Insert the logs into the ring buffer
c.l.Lock()
for _, l := range logs {
c.cache[l.Index%uint64(len(c.cache))] = l
}
c.l.Unlock()
return c.store.StoreLogs(logs)
}
func (c *LogCache) FirstIndex() (uint64, error) {
return c.store.FirstIndex()
}
func (c *LogCache) LastIndex() (uint64, error) {
return c.store.LastIndex()
}
func (c *LogCache) DeleteRange(min, max uint64) error {
// Invalidate the cache on deletes
c.l.Lock()
c.cache = make([]*Log, len(c.cache))
c.l.Unlock()
return c.store.DeleteRange(min, max)
}

676
vendor/github.com/hashicorp/raft/net_transport.go generated vendored Normal file
View File

@@ -0,0 +1,676 @@
package raft
import (
"bufio"
"errors"
"fmt"
"io"
"log"
"net"
"os"
"sync"
"time"
"github.com/hashicorp/go-msgpack/codec"
)
const (
rpcAppendEntries uint8 = iota
rpcRequestVote
rpcInstallSnapshot
// DefaultTimeoutScale is the default TimeoutScale in a NetworkTransport.
DefaultTimeoutScale = 256 * 1024 // 256KB
// rpcMaxPipeline controls the maximum number of outstanding
// AppendEntries RPC calls.
rpcMaxPipeline = 128
)
var (
// ErrTransportShutdown is returned when operations on a transport are
// invoked after it's been terminated.
ErrTransportShutdown = errors.New("transport shutdown")
// ErrPipelineShutdown is returned when the pipeline is closed.
ErrPipelineShutdown = errors.New("append pipeline closed")
)
/*
NetworkTransport provides a network based transport that can be
used to communicate with Raft on remote machines. It requires
an underlying stream layer to provide a stream abstraction, which can
be simple TCP, TLS, etc.
This transport is very simple and lightweight. Each RPC request is
framed by sending a byte that indicates the message type, followed
by the MsgPack encoded request.
The response is an error string followed by the response object,
both are encoded using MsgPack.
InstallSnapshot is special, in that after the RPC request we stream
the entire state. That socket is not re-used as the connection state
is not known if there is an error.
*/
type NetworkTransport struct {
connPool map[ServerAddress][]*netConn
connPoolLock sync.Mutex
consumeCh chan RPC
heartbeatFn func(RPC)
heartbeatFnLock sync.Mutex
logger *log.Logger
maxPool int
serverAddressProvider ServerAddressProvider
shutdown bool
shutdownCh chan struct{}
shutdownLock sync.Mutex
stream StreamLayer
timeout time.Duration
TimeoutScale int
}
// NetworkTransportConfig encapsulates configuration for the network transport layer.
type NetworkTransportConfig struct {
// ServerAddressProvider is used to override the target address when establishing a connection to invoke an RPC
ServerAddressProvider ServerAddressProvider
Logger *log.Logger
// Dialer
Stream StreamLayer
// MaxPool controls how many connections we will pool
MaxPool int
// Timeout is used to apply I/O deadlines. For InstallSnapshot, we multiply
// the timeout by (SnapshotSize / TimeoutScale).
Timeout time.Duration
}
type ServerAddressProvider interface {
ServerAddr(id ServerID) (ServerAddress, error)
}
// StreamLayer is used with the NetworkTransport to provide
// the low level stream abstraction.
type StreamLayer interface {
net.Listener
// Dial is used to create a new outgoing connection
Dial(address ServerAddress, timeout time.Duration) (net.Conn, error)
}
type netConn struct {
target ServerAddress
conn net.Conn
r *bufio.Reader
w *bufio.Writer
dec *codec.Decoder
enc *codec.Encoder
}
func (n *netConn) Release() error {
return n.conn.Close()
}
type netPipeline struct {
conn *netConn
trans *NetworkTransport
doneCh chan AppendFuture
inprogressCh chan *appendFuture
shutdown bool
shutdownCh chan struct{}
shutdownLock sync.Mutex
}
// NewNetworkTransportWithConfig creates a new network transport with the given config struct
func NewNetworkTransportWithConfig(
config *NetworkTransportConfig,
) *NetworkTransport {
if config.Logger == nil {
config.Logger = log.New(os.Stderr, "", log.LstdFlags)
}
trans := &NetworkTransport{
connPool: make(map[ServerAddress][]*netConn),
consumeCh: make(chan RPC),
logger: config.Logger,
maxPool: config.MaxPool,
shutdownCh: make(chan struct{}),
stream: config.Stream,
timeout: config.Timeout,
TimeoutScale: DefaultTimeoutScale,
serverAddressProvider: config.ServerAddressProvider,
}
go trans.listen()
return trans
}
// NewNetworkTransport creates a new network transport with the given dialer
// and listener. The maxPool controls how many connections we will pool. The
// timeout is used to apply I/O deadlines. For InstallSnapshot, we multiply
// the timeout by (SnapshotSize / TimeoutScale).
func NewNetworkTransport(
stream StreamLayer,
maxPool int,
timeout time.Duration,
logOutput io.Writer,
) *NetworkTransport {
if logOutput == nil {
logOutput = os.Stderr
}
logger := log.New(logOutput, "", log.LstdFlags)
config := &NetworkTransportConfig{Stream: stream, MaxPool: maxPool, Timeout: timeout, Logger: logger}
return NewNetworkTransportWithConfig(config)
}
// NewNetworkTransportWithLogger creates a new network transport with the given logger, dialer
// and listener. The maxPool controls how many connections we will pool. The
// timeout is used to apply I/O deadlines. For InstallSnapshot, we multiply
// the timeout by (SnapshotSize / TimeoutScale).
func NewNetworkTransportWithLogger(
stream StreamLayer,
maxPool int,
timeout time.Duration,
logger *log.Logger,
) *NetworkTransport {
config := &NetworkTransportConfig{Stream: stream, MaxPool: maxPool, Timeout: timeout, Logger: logger}
return NewNetworkTransportWithConfig(config)
}
// SetHeartbeatHandler is used to setup a heartbeat handler
// as a fast-pass. This is to avoid head-of-line blocking from
// disk IO.
func (n *NetworkTransport) SetHeartbeatHandler(cb func(rpc RPC)) {
n.heartbeatFnLock.Lock()
defer n.heartbeatFnLock.Unlock()
n.heartbeatFn = cb
}
// Close is used to stop the network transport.
func (n *NetworkTransport) Close() error {
n.shutdownLock.Lock()
defer n.shutdownLock.Unlock()
if !n.shutdown {
close(n.shutdownCh)
n.stream.Close()
n.shutdown = true
}
return nil
}
// Consumer implements the Transport interface.
func (n *NetworkTransport) Consumer() <-chan RPC {
return n.consumeCh
}
// LocalAddr implements the Transport interface.
func (n *NetworkTransport) LocalAddr() ServerAddress {
return ServerAddress(n.stream.Addr().String())
}
// IsShutdown is used to check if the transport is shutdown.
func (n *NetworkTransport) IsShutdown() bool {
select {
case <-n.shutdownCh:
return true
default:
return false
}
}
// getExistingConn is used to grab a pooled connection.
func (n *NetworkTransport) getPooledConn(target ServerAddress) *netConn {
n.connPoolLock.Lock()
defer n.connPoolLock.Unlock()
conns, ok := n.connPool[target]
if !ok || len(conns) == 0 {
return nil
}
var conn *netConn
num := len(conns)
conn, conns[num-1] = conns[num-1], nil
n.connPool[target] = conns[:num-1]
return conn
}
// getConnFromAddressProvider returns a connection from the server address provider if available, or defaults to a connection using the target server address
func (n *NetworkTransport) getConnFromAddressProvider(id ServerID, target ServerAddress) (*netConn, error) {
address := n.getProviderAddressOrFallback(id, target)
return n.getConn(address)
}
func (n *NetworkTransport) getProviderAddressOrFallback(id ServerID, target ServerAddress) ServerAddress {
if n.serverAddressProvider != nil {
serverAddressOverride, err := n.serverAddressProvider.ServerAddr(id)
if err != nil {
n.logger.Printf("[WARN] Unable to get address for server id %v, using fallback address %v: %v", id, target, err)
} else {
return serverAddressOverride
}
}
return target
}
// getConn is used to get a connection from the pool.
func (n *NetworkTransport) getConn(target ServerAddress) (*netConn, error) {
// Check for a pooled conn
if conn := n.getPooledConn(target); conn != nil {
return conn, nil
}
// Dial a new connection
conn, err := n.stream.Dial(target, n.timeout)
if err != nil {
return nil, err
}
// Wrap the conn
netConn := &netConn{
target: target,
conn: conn,
r: bufio.NewReader(conn),
w: bufio.NewWriter(conn),
}
// Setup encoder/decoders
netConn.dec = codec.NewDecoder(netConn.r, &codec.MsgpackHandle{})
netConn.enc = codec.NewEncoder(netConn.w, &codec.MsgpackHandle{})
// Done
return netConn, nil
}
// returnConn returns a connection back to the pool.
func (n *NetworkTransport) returnConn(conn *netConn) {
n.connPoolLock.Lock()
defer n.connPoolLock.Unlock()
key := conn.target
conns, _ := n.connPool[key]
if !n.IsShutdown() && len(conns) < n.maxPool {
n.connPool[key] = append(conns, conn)
} else {
conn.Release()
}
}
// AppendEntriesPipeline returns an interface that can be used to pipeline
// AppendEntries requests.
func (n *NetworkTransport) AppendEntriesPipeline(id ServerID, target ServerAddress) (AppendPipeline, error) {
// Get a connection
conn, err := n.getConnFromAddressProvider(id, target)
if err != nil {
return nil, err
}
// Create the pipeline
return newNetPipeline(n, conn), nil
}
// AppendEntries implements the Transport interface.
func (n *NetworkTransport) AppendEntries(id ServerID, target ServerAddress, args *AppendEntriesRequest, resp *AppendEntriesResponse) error {
return n.genericRPC(id, target, rpcAppendEntries, args, resp)
}
// RequestVote implements the Transport interface.
func (n *NetworkTransport) RequestVote(id ServerID, target ServerAddress, args *RequestVoteRequest, resp *RequestVoteResponse) error {
return n.genericRPC(id, target, rpcRequestVote, args, resp)
}
// genericRPC handles a simple request/response RPC.
func (n *NetworkTransport) genericRPC(id ServerID, target ServerAddress, rpcType uint8, args interface{}, resp interface{}) error {
// Get a conn
conn, err := n.getConnFromAddressProvider(id, target)
if err != nil {
return err
}
// Set a deadline
if n.timeout > 0 {
conn.conn.SetDeadline(time.Now().Add(n.timeout))
}
// Send the RPC
if err = sendRPC(conn, rpcType, args); err != nil {
return err
}
// Decode the response
canReturn, err := decodeResponse(conn, resp)
if canReturn {
n.returnConn(conn)
}
return err
}
// InstallSnapshot implements the Transport interface.
func (n *NetworkTransport) InstallSnapshot(id ServerID, target ServerAddress, args *InstallSnapshotRequest, resp *InstallSnapshotResponse, data io.Reader) error {
// Get a conn, always close for InstallSnapshot
conn, err := n.getConnFromAddressProvider(id, target)
if err != nil {
return err
}
defer conn.Release()
// Set a deadline, scaled by request size
if n.timeout > 0 {
timeout := n.timeout * time.Duration(args.Size/int64(n.TimeoutScale))
if timeout < n.timeout {
timeout = n.timeout
}
conn.conn.SetDeadline(time.Now().Add(timeout))
}
// Send the RPC
if err = sendRPC(conn, rpcInstallSnapshot, args); err != nil {
return err
}
// Stream the state
if _, err = io.Copy(conn.w, data); err != nil {
return err
}
// Flush
if err = conn.w.Flush(); err != nil {
return err
}
// Decode the response, do not return conn
_, err = decodeResponse(conn, resp)
return err
}
// EncodePeer implements the Transport interface.
func (n *NetworkTransport) EncodePeer(id ServerID, p ServerAddress) []byte {
address := n.getProviderAddressOrFallback(id, p)
return []byte(address)
}
// DecodePeer implements the Transport interface.
func (n *NetworkTransport) DecodePeer(buf []byte) ServerAddress {
return ServerAddress(buf)
}
// listen is used to handling incoming connections.
func (n *NetworkTransport) listen() {
for {
// Accept incoming connections
conn, err := n.stream.Accept()
if err != nil {
if n.IsShutdown() {
return
}
n.logger.Printf("[ERR] raft-net: Failed to accept connection: %v", err)
continue
}
n.logger.Printf("[DEBUG] raft-net: %v accepted connection from: %v", n.LocalAddr(), conn.RemoteAddr())
// Handle the connection in dedicated routine
go n.handleConn(conn)
}
}
// handleConn is used to handle an inbound connection for its lifespan.
func (n *NetworkTransport) handleConn(conn net.Conn) {
defer conn.Close()
r := bufio.NewReader(conn)
w := bufio.NewWriter(conn)
dec := codec.NewDecoder(r, &codec.MsgpackHandle{})
enc := codec.NewEncoder(w, &codec.MsgpackHandle{})
for {
if err := n.handleCommand(r, dec, enc); err != nil {
if err != io.EOF {
n.logger.Printf("[ERR] raft-net: Failed to decode incoming command: %v", err)
}
return
}
if err := w.Flush(); err != nil {
n.logger.Printf("[ERR] raft-net: Failed to flush response: %v", err)
return
}
}
}
// handleCommand is used to decode and dispatch a single command.
func (n *NetworkTransport) handleCommand(r *bufio.Reader, dec *codec.Decoder, enc *codec.Encoder) error {
// Get the rpc type
rpcType, err := r.ReadByte()
if err != nil {
return err
}
// Create the RPC object
respCh := make(chan RPCResponse, 1)
rpc := RPC{
RespChan: respCh,
}
// Decode the command
isHeartbeat := false
switch rpcType {
case rpcAppendEntries:
var req AppendEntriesRequest
if err := dec.Decode(&req); err != nil {
return err
}
rpc.Command = &req
// Check if this is a heartbeat
if req.Term != 0 && req.Leader != nil &&
req.PrevLogEntry == 0 && req.PrevLogTerm == 0 &&
len(req.Entries) == 0 && req.LeaderCommitIndex == 0 {
isHeartbeat = true
}
case rpcRequestVote:
var req RequestVoteRequest
if err := dec.Decode(&req); err != nil {
return err
}
rpc.Command = &req
case rpcInstallSnapshot:
var req InstallSnapshotRequest
if err := dec.Decode(&req); err != nil {
return err
}
rpc.Command = &req
rpc.Reader = io.LimitReader(r, req.Size)
default:
return fmt.Errorf("unknown rpc type %d", rpcType)
}
// Check for heartbeat fast-path
if isHeartbeat {
n.heartbeatFnLock.Lock()
fn := n.heartbeatFn
n.heartbeatFnLock.Unlock()
if fn != nil {
fn(rpc)
goto RESP
}
}
// Dispatch the RPC
select {
case n.consumeCh <- rpc:
case <-n.shutdownCh:
return ErrTransportShutdown
}
// Wait for response
RESP:
select {
case resp := <-respCh:
// Send the error first
respErr := ""
if resp.Error != nil {
respErr = resp.Error.Error()
}
if err := enc.Encode(respErr); err != nil {
return err
}
// Send the response
if err := enc.Encode(resp.Response); err != nil {
return err
}
case <-n.shutdownCh:
return ErrTransportShutdown
}
return nil
}
// decodeResponse is used to decode an RPC response and reports whether
// the connection can be reused.
func decodeResponse(conn *netConn, resp interface{}) (bool, error) {
// Decode the error if any
var rpcError string
if err := conn.dec.Decode(&rpcError); err != nil {
conn.Release()
return false, err
}
// Decode the response
if err := conn.dec.Decode(resp); err != nil {
conn.Release()
return false, err
}
// Format an error if any
if rpcError != "" {
return true, fmt.Errorf(rpcError)
}
return true, nil
}
// sendRPC is used to encode and send the RPC.
func sendRPC(conn *netConn, rpcType uint8, args interface{}) error {
// Write the request type
if err := conn.w.WriteByte(rpcType); err != nil {
conn.Release()
return err
}
// Send the request
if err := conn.enc.Encode(args); err != nil {
conn.Release()
return err
}
// Flush
if err := conn.w.Flush(); err != nil {
conn.Release()
return err
}
return nil
}
// newNetPipeline is used to construct a netPipeline from a given
// transport and connection.
func newNetPipeline(trans *NetworkTransport, conn *netConn) *netPipeline {
n := &netPipeline{
conn: conn,
trans: trans,
doneCh: make(chan AppendFuture, rpcMaxPipeline),
inprogressCh: make(chan *appendFuture, rpcMaxPipeline),
shutdownCh: make(chan struct{}),
}
go n.decodeResponses()
return n
}
// decodeResponses is a long running routine that decodes the responses
// sent on the connection.
func (n *netPipeline) decodeResponses() {
timeout := n.trans.timeout
for {
select {
case future := <-n.inprogressCh:
if timeout > 0 {
n.conn.conn.SetReadDeadline(time.Now().Add(timeout))
}
_, err := decodeResponse(n.conn, future.resp)
future.respond(err)
select {
case n.doneCh <- future:
case <-n.shutdownCh:
return
}
case <-n.shutdownCh:
return
}
}
}
// AppendEntries is used to pipeline a new append entries request.
func (n *netPipeline) AppendEntries(args *AppendEntriesRequest, resp *AppendEntriesResponse) (AppendFuture, error) {
// Create a new future
future := &appendFuture{
start: time.Now(),
args: args,
resp: resp,
}
future.init()
// Add a send timeout
if timeout := n.trans.timeout; timeout > 0 {
n.conn.conn.SetWriteDeadline(time.Now().Add(timeout))
}
// Send the RPC
if err := sendRPC(n.conn, rpcAppendEntries, future.args); err != nil {
return nil, err
}
// Hand-off for decoding, this can also cause back-pressure
// to prevent too many inflight requests
select {
case n.inprogressCh <- future:
return future, nil
case <-n.shutdownCh:
return nil, ErrPipelineShutdown
}
}
// Consumer returns a channel that can be used to consume complete futures.
func (n *netPipeline) Consumer() <-chan AppendFuture {
return n.doneCh
}
// Closed is used to shutdown the pipeline connection.
func (n *netPipeline) Close() error {
n.shutdownLock.Lock()
defer n.shutdownLock.Unlock()
if n.shutdown {
return nil
}
// Release the connection
n.conn.Release()
n.shutdown = true
close(n.shutdownCh)
return nil
}

117
vendor/github.com/hashicorp/raft/observer.go generated vendored Normal file
View File

@@ -0,0 +1,117 @@
package raft
import (
"sync/atomic"
)
// Observation is sent along the given channel to observers when an event occurs.
type Observation struct {
// Raft holds the Raft instance generating the observation.
Raft *Raft
// Data holds observation-specific data. Possible types are
// *RequestVoteRequest and RaftState.
Data interface{}
}
// nextObserverId is used to provide a unique ID for each observer to aid in
// deregistration.
var nextObserverID uint64
// FilterFn is a function that can be registered in order to filter observations.
// The function reports whether the observation should be included - if
// it returns false, the observation will be filtered out.
type FilterFn func(o *Observation) bool
// Observer describes what to do with a given observation.
type Observer struct {
// numObserved and numDropped are performance counters for this observer.
// 64 bit types must be 64 bit aligned to use with atomic operations on
// 32 bit platforms, so keep them at the top of the struct.
numObserved uint64
numDropped uint64
// channel receives observations.
channel chan Observation
// blocking, if true, will cause Raft to block when sending an observation
// to this observer. This should generally be set to false.
blocking bool
// filter will be called to determine if an observation should be sent to
// the channel.
filter FilterFn
// id is the ID of this observer in the Raft map.
id uint64
}
// NewObserver creates a new observer that can be registered
// to make observations on a Raft instance. Observations
// will be sent on the given channel if they satisfy the
// given filter.
//
// If blocking is true, the observer will block when it can't
// send on the channel, otherwise it may discard events.
func NewObserver(channel chan Observation, blocking bool, filter FilterFn) *Observer {
return &Observer{
channel: channel,
blocking: blocking,
filter: filter,
id: atomic.AddUint64(&nextObserverID, 1),
}
}
// GetNumObserved returns the number of observations.
func (or *Observer) GetNumObserved() uint64 {
return atomic.LoadUint64(&or.numObserved)
}
// GetNumDropped returns the number of dropped observations due to blocking.
func (or *Observer) GetNumDropped() uint64 {
return atomic.LoadUint64(&or.numDropped)
}
// RegisterObserver registers a new observer.
func (r *Raft) RegisterObserver(or *Observer) {
r.observersLock.Lock()
defer r.observersLock.Unlock()
r.observers[or.id] = or
}
// DeregisterObserver deregisters an observer.
func (r *Raft) DeregisterObserver(or *Observer) {
r.observersLock.Lock()
defer r.observersLock.Unlock()
delete(r.observers, or.id)
}
// observe sends an observation to every observer.
func (r *Raft) observe(o interface{}) {
// In general observers should not block. But in any case this isn't
// disastrous as we only hold a read lock, which merely prevents
// registration / deregistration of observers.
r.observersLock.RLock()
defer r.observersLock.RUnlock()
for _, or := range r.observers {
// It's wasteful to do this in the loop, but for the common case
// where there are no observers we won't create any objects.
ob := Observation{Raft: r, Data: o}
if or.filter != nil && !or.filter(&ob) {
continue
}
if or.channel == nil {
continue
}
if or.blocking {
or.channel <- ob
atomic.AddUint64(&or.numObserved, 1)
} else {
select {
case or.channel <- ob:
atomic.AddUint64(&or.numObserved, 1)
default:
atomic.AddUint64(&or.numDropped, 1)
}
}
}
}

98
vendor/github.com/hashicorp/raft/peersjson.go generated vendored Normal file
View File

@@ -0,0 +1,98 @@
package raft
import (
"bytes"
"encoding/json"
"io/ioutil"
)
// ReadPeersJSON consumes a legacy peers.json file in the format of the old JSON
// peer store and creates a new-style configuration structure. This can be used
// to migrate this data or perform manual recovery when running protocol versions
// that can interoperate with older, unversioned Raft servers. This should not be
// used once server IDs are in use, because the old peers.json file didn't have
// support for these, nor non-voter suffrage types.
func ReadPeersJSON(path string) (Configuration, error) {
// Read in the file.
buf, err := ioutil.ReadFile(path)
if err != nil {
return Configuration{}, err
}
// Parse it as JSON.
var peers []string
dec := json.NewDecoder(bytes.NewReader(buf))
if err := dec.Decode(&peers); err != nil {
return Configuration{}, err
}
// Map it into the new-style configuration structure. We can only specify
// voter roles here, and the ID has to be the same as the address.
var configuration Configuration
for _, peer := range peers {
server := Server{
Suffrage: Voter,
ID: ServerID(peer),
Address: ServerAddress(peer),
}
configuration.Servers = append(configuration.Servers, server)
}
// We should only ingest valid configurations.
if err := checkConfiguration(configuration); err != nil {
return Configuration{}, err
}
return configuration, nil
}
// configEntry is used when decoding a new-style peers.json.
type configEntry struct {
// ID is the ID of the server (a UUID, usually).
ID ServerID `json:"id"`
// Address is the host:port of the server.
Address ServerAddress `json:"address"`
// NonVoter controls the suffrage. We choose this sense so people
// can leave this out and get a Voter by default.
NonVoter bool `json:"non_voter"`
}
// ReadConfigJSON reads a new-style peers.json and returns a configuration
// structure. This can be used to perform manual recovery when running protocol
// versions that use server IDs.
func ReadConfigJSON(path string) (Configuration, error) {
// Read in the file.
buf, err := ioutil.ReadFile(path)
if err != nil {
return Configuration{}, err
}
// Parse it as JSON.
var peers []configEntry
dec := json.NewDecoder(bytes.NewReader(buf))
if err := dec.Decode(&peers); err != nil {
return Configuration{}, err
}
// Map it into the new-style configuration structure.
var configuration Configuration
for _, peer := range peers {
suffrage := Voter
if peer.NonVoter {
suffrage = Nonvoter
}
server := Server{
Suffrage: suffrage,
ID: peer.ID,
Address: peer.Address,
}
configuration.Servers = append(configuration.Servers, server)
}
// We should only ingest valid configurations.
if err := checkConfiguration(configuration); err != nil {
return Configuration{}, err
}
return configuration, nil
}

1459
vendor/github.com/hashicorp/raft/raft.go generated vendored Normal file
View File

File diff suppressed because it is too large Load Diff

561
vendor/github.com/hashicorp/raft/replication.go generated vendored Normal file
View File

@@ -0,0 +1,561 @@
package raft
import (
"errors"
"fmt"
"sync"
"time"
"github.com/armon/go-metrics"
)
const (
maxFailureScale = 12
failureWait = 10 * time.Millisecond
)
var (
// ErrLogNotFound indicates a given log entry is not available.
ErrLogNotFound = errors.New("log not found")
// ErrPipelineReplicationNotSupported can be returned by the transport to
// signal that pipeline replication is not supported in general, and that
// no error message should be produced.
ErrPipelineReplicationNotSupported = errors.New("pipeline replication not supported")
)
// followerReplication is in charge of sending snapshots and log entries from
// this leader during this particular term to a remote follower.
type followerReplication struct {
// peer contains the network address and ID of the remote follower.
peer Server
// commitment tracks the entries acknowledged by followers so that the
// leader's commit index can advance. It is updated on successsful
// AppendEntries responses.
commitment *commitment
// stopCh is notified/closed when this leader steps down or the follower is
// removed from the cluster. In the follower removed case, it carries a log
// index; replication should be attempted with a best effort up through that
// index, before exiting.
stopCh chan uint64
// triggerCh is notified every time new entries are appended to the log.
triggerCh chan struct{}
// currentTerm is the term of this leader, to be included in AppendEntries
// requests.
currentTerm uint64
// nextIndex is the index of the next log entry to send to the follower,
// which may fall past the end of the log.
nextIndex uint64
// lastContact is updated to the current time whenever any response is
// received from the follower (successful or not). This is used to check
// whether the leader should step down (Raft.checkLeaderLease()).
lastContact time.Time
// lastContactLock protects 'lastContact'.
lastContactLock sync.RWMutex
// failures counts the number of failed RPCs since the last success, which is
// used to apply backoff.
failures uint64
// notifyCh is notified to send out a heartbeat, which is used to check that
// this server is still leader.
notifyCh chan struct{}
// notify is a list of futures to be resolved upon receipt of an
// acknowledgement, then cleared from this list.
notify []*verifyFuture
// notifyLock protects 'notify'.
notifyLock sync.Mutex
// stepDown is used to indicate to the leader that we
// should step down based on information from a follower.
stepDown chan struct{}
// allowPipeline is used to determine when to pipeline the AppendEntries RPCs.
// It is private to this replication goroutine.
allowPipeline bool
}
// notifyAll is used to notify all the waiting verify futures
// if the follower believes we are still the leader.
func (s *followerReplication) notifyAll(leader bool) {
// Clear the waiting notifies minimizing lock time
s.notifyLock.Lock()
n := s.notify
s.notify = nil
s.notifyLock.Unlock()
// Submit our votes
for _, v := range n {
v.vote(leader)
}
}
// LastContact returns the time of last contact.
func (s *followerReplication) LastContact() time.Time {
s.lastContactLock.RLock()
last := s.lastContact
s.lastContactLock.RUnlock()
return last
}
// setLastContact sets the last contact to the current time.
func (s *followerReplication) setLastContact() {
s.lastContactLock.Lock()
s.lastContact = time.Now()
s.lastContactLock.Unlock()
}
// replicate is a long running routine that replicates log entries to a single
// follower.
func (r *Raft) replicate(s *followerReplication) {
// Start an async heartbeating routing
stopHeartbeat := make(chan struct{})
defer close(stopHeartbeat)
r.goFunc(func() { r.heartbeat(s, stopHeartbeat) })
RPC:
shouldStop := false
for !shouldStop {
select {
case maxIndex := <-s.stopCh:
// Make a best effort to replicate up to this index
if maxIndex > 0 {
r.replicateTo(s, maxIndex)
}
return
case <-s.triggerCh:
lastLogIdx, _ := r.getLastLog()
shouldStop = r.replicateTo(s, lastLogIdx)
case <-randomTimeout(r.conf.CommitTimeout): // TODO: what is this?
lastLogIdx, _ := r.getLastLog()
shouldStop = r.replicateTo(s, lastLogIdx)
}
// If things looks healthy, switch to pipeline mode
if !shouldStop && s.allowPipeline {
goto PIPELINE
}
}
return
PIPELINE:
// Disable until re-enabled
s.allowPipeline = false
// Replicates using a pipeline for high performance. This method
// is not able to gracefully recover from errors, and so we fall back
// to standard mode on failure.
if err := r.pipelineReplicate(s); err != nil {
if err != ErrPipelineReplicationNotSupported {
r.logger.Printf("[ERR] raft: Failed to start pipeline replication to %s: %s", s.peer, err)
}
}
goto RPC
}
// replicateTo is a helper to replicate(), used to replicate the logs up to a
// given last index.
// If the follower log is behind, we take care to bring them up to date.
func (r *Raft) replicateTo(s *followerReplication, lastIndex uint64) (shouldStop bool) {
// Create the base request
var req AppendEntriesRequest
var resp AppendEntriesResponse
var start time.Time
START:
// Prevent an excessive retry rate on errors
if s.failures > 0 {
select {
case <-time.After(backoff(failureWait, s.failures, maxFailureScale)):
case <-r.shutdownCh:
}
}
// Setup the request
if err := r.setupAppendEntries(s, &req, s.nextIndex, lastIndex); err == ErrLogNotFound {
goto SEND_SNAP
} else if err != nil {
return
}
// Make the RPC call
start = time.Now()
if err := r.trans.AppendEntries(s.peer.ID, s.peer.Address, &req, &resp); err != nil {
r.logger.Printf("[ERR] raft: Failed to AppendEntries to %v: %v", s.peer, err)
s.failures++
return
}
appendStats(string(s.peer.ID), start, float32(len(req.Entries)))
// Check for a newer term, stop running
if resp.Term > req.Term {
r.handleStaleTerm(s)
return true
}
// Update the last contact
s.setLastContact()
// Update s based on success
if resp.Success {
// Update our replication state
updateLastAppended(s, &req)
// Clear any failures, allow pipelining
s.failures = 0
s.allowPipeline = true
} else {
s.nextIndex = max(min(s.nextIndex-1, resp.LastLog+1), 1)
if resp.NoRetryBackoff {
s.failures = 0
} else {
s.failures++
}
r.logger.Printf("[WARN] raft: AppendEntries to %v rejected, sending older logs (next: %d)", s.peer, s.nextIndex)
}
CHECK_MORE:
// Poll the stop channel here in case we are looping and have been asked
// to stop, or have stepped down as leader. Even for the best effort case
// where we are asked to replicate to a given index and then shutdown,
// it's better to not loop in here to send lots of entries to a straggler
// that's leaving the cluster anyways.
select {
case <-s.stopCh:
return true
default:
}
// Check if there are more logs to replicate
if s.nextIndex <= lastIndex {
goto START
}
return
// SEND_SNAP is used when we fail to get a log, usually because the follower
// is too far behind, and we must ship a snapshot down instead
SEND_SNAP:
if stop, err := r.sendLatestSnapshot(s); stop {
return true
} else if err != nil {
r.logger.Printf("[ERR] raft: Failed to send snapshot to %v: %v", s.peer, err)
return
}
// Check if there is more to replicate
goto CHECK_MORE
}
// sendLatestSnapshot is used to send the latest snapshot we have
// down to our follower.
func (r *Raft) sendLatestSnapshot(s *followerReplication) (bool, error) {
// Get the snapshots
snapshots, err := r.snapshots.List()
if err != nil {
r.logger.Printf("[ERR] raft: Failed to list snapshots: %v", err)
return false, err
}
// Check we have at least a single snapshot
if len(snapshots) == 0 {
return false, fmt.Errorf("no snapshots found")
}
// Open the most recent snapshot
snapID := snapshots[0].ID
meta, snapshot, err := r.snapshots.Open(snapID)
if err != nil {
r.logger.Printf("[ERR] raft: Failed to open snapshot %v: %v", snapID, err)
return false, err
}
defer snapshot.Close()
// Setup the request
req := InstallSnapshotRequest{
RPCHeader: r.getRPCHeader(),
SnapshotVersion: meta.Version,
Term: s.currentTerm,
Leader: r.trans.EncodePeer(r.localID, r.localAddr),
LastLogIndex: meta.Index,
LastLogTerm: meta.Term,
Peers: meta.Peers,
Size: meta.Size,
Configuration: encodeConfiguration(meta.Configuration),
ConfigurationIndex: meta.ConfigurationIndex,
}
// Make the call
start := time.Now()
var resp InstallSnapshotResponse
if err := r.trans.InstallSnapshot(s.peer.ID, s.peer.Address, &req, &resp, snapshot); err != nil {
r.logger.Printf("[ERR] raft: Failed to install snapshot %v: %v", snapID, err)
s.failures++
return false, err
}
metrics.MeasureSince([]string{"raft", "replication", "installSnapshot", string(s.peer.ID)}, start)
// Check for a newer term, stop running
if resp.Term > req.Term {
r.handleStaleTerm(s)
return true, nil
}
// Update the last contact
s.setLastContact()
// Check for success
if resp.Success {
// Update the indexes
s.nextIndex = meta.Index + 1
s.commitment.match(s.peer.ID, meta.Index)
// Clear any failures
s.failures = 0
// Notify we are still leader
s.notifyAll(true)
} else {
s.failures++
r.logger.Printf("[WARN] raft: InstallSnapshot to %v rejected", s.peer)
}
return false, nil
}
// heartbeat is used to periodically invoke AppendEntries on a peer
// to ensure they don't time out. This is done async of replicate(),
// since that routine could potentially be blocked on disk IO.
func (r *Raft) heartbeat(s *followerReplication, stopCh chan struct{}) {
var failures uint64
req := AppendEntriesRequest{
RPCHeader: r.getRPCHeader(),
Term: s.currentTerm,
Leader: r.trans.EncodePeer(r.localID, r.localAddr),
}
var resp AppendEntriesResponse
for {
// Wait for the next heartbeat interval or forced notify
select {
case <-s.notifyCh:
case <-randomTimeout(r.conf.HeartbeatTimeout / 10):
case <-stopCh:
return
}
start := time.Now()
if err := r.trans.AppendEntries(s.peer.ID, s.peer.Address, &req, &resp); err != nil {
r.logger.Printf("[ERR] raft: Failed to heartbeat to %v: %v", s.peer.Address, err)
failures++
select {
case <-time.After(backoff(failureWait, failures, maxFailureScale)):
case <-stopCh:
}
} else {
s.setLastContact()
failures = 0
metrics.MeasureSince([]string{"raft", "replication", "heartbeat", string(s.peer.ID)}, start)
s.notifyAll(resp.Success)
}
}
}
// pipelineReplicate is used when we have synchronized our state with the follower,
// and want to switch to a higher performance pipeline mode of replication.
// We only pipeline AppendEntries commands, and if we ever hit an error, we fall
// back to the standard replication which can handle more complex situations.
func (r *Raft) pipelineReplicate(s *followerReplication) error {
// Create a new pipeline
pipeline, err := r.trans.AppendEntriesPipeline(s.peer.ID, s.peer.Address)
if err != nil {
return err
}
defer pipeline.Close()
// Log start and stop of pipeline
r.logger.Printf("[INFO] raft: pipelining replication to peer %v", s.peer)
defer r.logger.Printf("[INFO] raft: aborting pipeline replication to peer %v", s.peer)
// Create a shutdown and finish channel
stopCh := make(chan struct{})
finishCh := make(chan struct{})
// Start a dedicated decoder
r.goFunc(func() { r.pipelineDecode(s, pipeline, stopCh, finishCh) })
// Start pipeline sends at the last good nextIndex
nextIndex := s.nextIndex
shouldStop := false
SEND:
for !shouldStop {
select {
case <-finishCh:
break SEND
case maxIndex := <-s.stopCh:
// Make a best effort to replicate up to this index
if maxIndex > 0 {
r.pipelineSend(s, pipeline, &nextIndex, maxIndex)
}
break SEND
case <-s.triggerCh:
lastLogIdx, _ := r.getLastLog()
shouldStop = r.pipelineSend(s, pipeline, &nextIndex, lastLogIdx)
case <-randomTimeout(r.conf.CommitTimeout):
lastLogIdx, _ := r.getLastLog()
shouldStop = r.pipelineSend(s, pipeline, &nextIndex, lastLogIdx)
}
}
// Stop our decoder, and wait for it to finish
close(stopCh)
select {
case <-finishCh:
case <-r.shutdownCh:
}
return nil
}
// pipelineSend is used to send data over a pipeline. It is a helper to
// pipelineReplicate.
func (r *Raft) pipelineSend(s *followerReplication, p AppendPipeline, nextIdx *uint64, lastIndex uint64) (shouldStop bool) {
// Create a new append request
req := new(AppendEntriesRequest)
if err := r.setupAppendEntries(s, req, *nextIdx, lastIndex); err != nil {
return true
}
// Pipeline the append entries
if _, err := p.AppendEntries(req, new(AppendEntriesResponse)); err != nil {
r.logger.Printf("[ERR] raft: Failed to pipeline AppendEntries to %v: %v", s.peer, err)
return true
}
// Increase the next send log to avoid re-sending old logs
if n := len(req.Entries); n > 0 {
last := req.Entries[n-1]
*nextIdx = last.Index + 1
}
return false
}
// pipelineDecode is used to decode the responses of pipelined requests.
func (r *Raft) pipelineDecode(s *followerReplication, p AppendPipeline, stopCh, finishCh chan struct{}) {
defer close(finishCh)
respCh := p.Consumer()
for {
select {
case ready := <-respCh:
req, resp := ready.Request(), ready.Response()
appendStats(string(s.peer.ID), ready.Start(), float32(len(req.Entries)))
// Check for a newer term, stop running
if resp.Term > req.Term {
r.handleStaleTerm(s)
return
}
// Update the last contact
s.setLastContact()
// Abort pipeline if not successful
if !resp.Success {
return
}
// Update our replication state
updateLastAppended(s, req)
case <-stopCh:
return
}
}
}
// setupAppendEntries is used to setup an append entries request.
func (r *Raft) setupAppendEntries(s *followerReplication, req *AppendEntriesRequest, nextIndex, lastIndex uint64) error {
req.RPCHeader = r.getRPCHeader()
req.Term = s.currentTerm
req.Leader = r.trans.EncodePeer(r.localID, r.localAddr)
req.LeaderCommitIndex = r.getCommitIndex()
if err := r.setPreviousLog(req, nextIndex); err != nil {
return err
}
if err := r.setNewLogs(req, nextIndex, lastIndex); err != nil {
return err
}
return nil
}
// setPreviousLog is used to setup the PrevLogEntry and PrevLogTerm for an
// AppendEntriesRequest given the next index to replicate.
func (r *Raft) setPreviousLog(req *AppendEntriesRequest, nextIndex uint64) error {
// Guard for the first index, since there is no 0 log entry
// Guard against the previous index being a snapshot as well
lastSnapIdx, lastSnapTerm := r.getLastSnapshot()
if nextIndex == 1 {
req.PrevLogEntry = 0
req.PrevLogTerm = 0
} else if (nextIndex - 1) == lastSnapIdx {
req.PrevLogEntry = lastSnapIdx
req.PrevLogTerm = lastSnapTerm
} else {
var l Log
if err := r.logs.GetLog(nextIndex-1, &l); err != nil {
r.logger.Printf("[ERR] raft: Failed to get log at index %d: %v",
nextIndex-1, err)
return err
}
// Set the previous index and term (0 if nextIndex is 1)
req.PrevLogEntry = l.Index
req.PrevLogTerm = l.Term
}
return nil
}
// setNewLogs is used to setup the logs which should be appended for a request.
func (r *Raft) setNewLogs(req *AppendEntriesRequest, nextIndex, lastIndex uint64) error {
// Append up to MaxAppendEntries or up to the lastIndex
req.Entries = make([]*Log, 0, r.conf.MaxAppendEntries)
maxIndex := min(nextIndex+uint64(r.conf.MaxAppendEntries)-1, lastIndex)
for i := nextIndex; i <= maxIndex; i++ {
oldLog := new(Log)
if err := r.logs.GetLog(i, oldLog); err != nil {
r.logger.Printf("[ERR] raft: Failed to get log at index %d: %v", i, err)
return err
}
req.Entries = append(req.Entries, oldLog)
}
return nil
}
// appendStats is used to emit stats about an AppendEntries invocation.
func appendStats(peer string, start time.Time, logs float32) {
metrics.MeasureSince([]string{"raft", "replication", "appendEntries", "rpc", peer}, start)
metrics.IncrCounter([]string{"raft", "replication", "appendEntries", "logs", peer}, logs)
}
// handleStaleTerm is used when a follower indicates that we have a stale term.
func (r *Raft) handleStaleTerm(s *followerReplication) {
r.logger.Printf("[ERR] raft: peer %v has newer term, stopping replication", s.peer)
s.notifyAll(false) // No longer leader
asyncNotifyCh(s.stepDown)
}
// updateLastAppended is used to update follower replication state after a
// successful AppendEntries RPC.
// TODO: This isn't used during InstallSnapshot, but the code there is similar.
func updateLastAppended(s *followerReplication, req *AppendEntriesRequest) {
// Mark any inflight logs as committed
if logs := req.Entries; len(logs) > 0 {
last := logs[len(logs)-1]
s.nextIndex = last.Index + 1
s.commitment.match(s.peer.ID, last.Index)
}
// Notify still leader
s.notifyAll(true)
}

239
vendor/github.com/hashicorp/raft/snapshot.go generated vendored Normal file
View File

@@ -0,0 +1,239 @@
package raft
import (
"fmt"
"io"
"time"
"github.com/armon/go-metrics"
)
// SnapshotMeta is for metadata of a snapshot.
type SnapshotMeta struct {
// Version is the version number of the snapshot metadata. This does not cover
// the application's data in the snapshot, that should be versioned
// separately.
Version SnapshotVersion
// ID is opaque to the store, and is used for opening.
ID string
// Index and Term store when the snapshot was taken.
Index uint64
Term uint64
// Peers is deprecated and used to support version 0 snapshots, but will
// be populated in version 1 snapshots as well to help with upgrades.
Peers []byte
// Configuration and ConfigurationIndex are present in version 1
// snapshots and later.
Configuration Configuration
ConfigurationIndex uint64
// Size is the size of the snapshot in bytes.
Size int64
}
// SnapshotStore interface is used to allow for flexible implementations
// of snapshot storage and retrieval. For example, a client could implement
// a shared state store such as S3, allowing new nodes to restore snapshots
// without streaming from the leader.
type SnapshotStore interface {
// Create is used to begin a snapshot at a given index and term, and with
// the given committed configuration. The version parameter controls
// which snapshot version to create.
Create(version SnapshotVersion, index, term uint64, configuration Configuration,
configurationIndex uint64, trans Transport) (SnapshotSink, error)
// List is used to list the available snapshots in the store.
// It should return then in descending order, with the highest index first.
List() ([]*SnapshotMeta, error)
// Open takes a snapshot ID and provides a ReadCloser. Once close is
// called it is assumed the snapshot is no longer needed.
Open(id string) (*SnapshotMeta, io.ReadCloser, error)
}
// SnapshotSink is returned by StartSnapshot. The FSM will Write state
// to the sink and call Close on completion. On error, Cancel will be invoked.
type SnapshotSink interface {
io.WriteCloser
ID() string
Cancel() error
}
// runSnapshots is a long running goroutine used to manage taking
// new snapshots of the FSM. It runs in parallel to the FSM and
// main goroutines, so that snapshots do not block normal operation.
func (r *Raft) runSnapshots() {
for {
select {
case <-randomTimeout(r.conf.SnapshotInterval):
// Check if we should snapshot
if !r.shouldSnapshot() {
continue
}
// Trigger a snapshot
if _, err := r.takeSnapshot(); err != nil {
r.logger.Printf("[ERR] raft: Failed to take snapshot: %v", err)
}
case future := <-r.userSnapshotCh:
// User-triggered, run immediately
id, err := r.takeSnapshot()
if err != nil {
r.logger.Printf("[ERR] raft: Failed to take snapshot: %v", err)
} else {
future.opener = func() (*SnapshotMeta, io.ReadCloser, error) {
return r.snapshots.Open(id)
}
}
future.respond(err)
case <-r.shutdownCh:
return
}
}
}
// shouldSnapshot checks if we meet the conditions to take
// a new snapshot.
func (r *Raft) shouldSnapshot() bool {
// Check the last snapshot index
lastSnap, _ := r.getLastSnapshot()
// Check the last log index
lastIdx, err := r.logs.LastIndex()
if err != nil {
r.logger.Printf("[ERR] raft: Failed to get last log index: %v", err)
return false
}
// Compare the delta to the threshold
delta := lastIdx - lastSnap
return delta >= r.conf.SnapshotThreshold
}
// takeSnapshot is used to take a new snapshot. This must only be called from
// the snapshot thread, never the main thread. This returns the ID of the new
// snapshot, along with an error.
func (r *Raft) takeSnapshot() (string, error) {
defer metrics.MeasureSince([]string{"raft", "snapshot", "takeSnapshot"}, time.Now())
// Create a request for the FSM to perform a snapshot.
snapReq := &reqSnapshotFuture{}
snapReq.init()
// Wait for dispatch or shutdown.
select {
case r.fsmSnapshotCh <- snapReq:
case <-r.shutdownCh:
return "", ErrRaftShutdown
}
// Wait until we get a response
if err := snapReq.Error(); err != nil {
if err != ErrNothingNewToSnapshot {
err = fmt.Errorf("failed to start snapshot: %v", err)
}
return "", err
}
defer snapReq.snapshot.Release()
// Make a request for the configurations and extract the committed info.
// We have to use the future here to safely get this information since
// it is owned by the main thread.
configReq := &configurationsFuture{}
configReq.init()
select {
case r.configurationsCh <- configReq:
case <-r.shutdownCh:
return "", ErrRaftShutdown
}
if err := configReq.Error(); err != nil {
return "", err
}
committed := configReq.configurations.committed
committedIndex := configReq.configurations.committedIndex
// We don't support snapshots while there's a config change outstanding
// since the snapshot doesn't have a means to represent this state. This
// is a little weird because we need the FSM to apply an index that's
// past the configuration change, even though the FSM itself doesn't see
// the configuration changes. It should be ok in practice with normal
// application traffic flowing through the FSM. If there's none of that
// then it's not crucial that we snapshot, since there's not much going
// on Raft-wise.
if snapReq.index < committedIndex {
return "", fmt.Errorf("cannot take snapshot now, wait until the configuration entry at %v has been applied (have applied %v)",
committedIndex, snapReq.index)
}
// Create a new snapshot.
r.logger.Printf("[INFO] raft: Starting snapshot up to %d", snapReq.index)
start := time.Now()
version := getSnapshotVersion(r.protocolVersion)
sink, err := r.snapshots.Create(version, snapReq.index, snapReq.term, committed, committedIndex, r.trans)
if err != nil {
return "", fmt.Errorf("failed to create snapshot: %v", err)
}
metrics.MeasureSince([]string{"raft", "snapshot", "create"}, start)
// Try to persist the snapshot.
start = time.Now()
if err := snapReq.snapshot.Persist(sink); err != nil {
sink.Cancel()
return "", fmt.Errorf("failed to persist snapshot: %v", err)
}
metrics.MeasureSince([]string{"raft", "snapshot", "persist"}, start)
// Close and check for error.
if err := sink.Close(); err != nil {
return "", fmt.Errorf("failed to close snapshot: %v", err)
}
// Update the last stable snapshot info.
r.setLastSnapshot(snapReq.index, snapReq.term)
// Compact the logs.
if err := r.compactLogs(snapReq.index); err != nil {
return "", err
}
r.logger.Printf("[INFO] raft: Snapshot to %d complete", snapReq.index)
return sink.ID(), nil
}
// compactLogs takes the last inclusive index of a snapshot
// and trims the logs that are no longer needed.
func (r *Raft) compactLogs(snapIdx uint64) error {
defer metrics.MeasureSince([]string{"raft", "compactLogs"}, time.Now())
// Determine log ranges to compact
minLog, err := r.logs.FirstIndex()
if err != nil {
return fmt.Errorf("failed to get first log index: %v", err)
}
// Check if we have enough logs to truncate
lastLogIdx, _ := r.getLastLog()
if lastLogIdx <= r.conf.TrailingLogs {
return nil
}
// Truncate up to the end of the snapshot, or `TrailingLogs`
// back from the head, which ever is further back. This ensures
// at least `TrailingLogs` entries, but does not allow logs
// after the snapshot to be removed.
maxLog := min(snapIdx, lastLogIdx-r.conf.TrailingLogs)
// Log this
r.logger.Printf("[INFO] raft: Compacting logs from %d to %d", minLog, maxLog)
// Compact the logs
if err := r.logs.DeleteRange(minLog, maxLog); err != nil {
return fmt.Errorf("log compaction failed: %v", err)
}
return nil
}

15
vendor/github.com/hashicorp/raft/stable.go generated vendored Normal file
View File

@@ -0,0 +1,15 @@
package raft
// StableStore is used to provide stable storage
// of key configurations to ensure safety.
type StableStore interface {
Set(key []byte, val []byte) error
// Get returns the value for key, or an empty byte slice if key was not found.
Get(key []byte) ([]byte, error)
SetUint64(key []byte, val uint64) error
// GetUint64 returns the uint64 value for key, or 0 if key was not found.
GetUint64(key []byte) (uint64, error)
}

171
vendor/github.com/hashicorp/raft/state.go generated vendored Normal file
View File

@@ -0,0 +1,171 @@
package raft
import (
"sync"
"sync/atomic"
)
// RaftState captures the state of a Raft node: Follower, Candidate, Leader,
// or Shutdown.
type RaftState uint32
const (
// Follower is the initial state of a Raft node.
Follower RaftState = iota
// Candidate is one of the valid states of a Raft node.
Candidate
// Leader is one of the valid states of a Raft node.
Leader
// Shutdown is the terminal state of a Raft node.
Shutdown
)
func (s RaftState) String() string {
switch s {
case Follower:
return "Follower"
case Candidate:
return "Candidate"
case Leader:
return "Leader"
case Shutdown:
return "Shutdown"
default:
return "Unknown"
}
}
// raftState is used to maintain various state variables
// and provides an interface to set/get the variables in a
// thread safe manner.
type raftState struct {
// currentTerm commitIndex, lastApplied, must be kept at the top of
// the struct so they're 64 bit aligned which is a requirement for
// atomic ops on 32 bit platforms.
// The current term, cache of StableStore
currentTerm uint64
// Highest committed log entry
commitIndex uint64
// Last applied log to the FSM
lastApplied uint64
// protects 4 next fields
lastLock sync.Mutex
// Cache the latest snapshot index/term
lastSnapshotIndex uint64
lastSnapshotTerm uint64
// Cache the latest log from LogStore
lastLogIndex uint64
lastLogTerm uint64
// Tracks running goroutines
routinesGroup sync.WaitGroup
// The current state
state RaftState
}
func (r *raftState) getState() RaftState {
stateAddr := (*uint32)(&r.state)
return RaftState(atomic.LoadUint32(stateAddr))
}
func (r *raftState) setState(s RaftState) {
stateAddr := (*uint32)(&r.state)
atomic.StoreUint32(stateAddr, uint32(s))
}
func (r *raftState) getCurrentTerm() uint64 {
return atomic.LoadUint64(&r.currentTerm)
}
func (r *raftState) setCurrentTerm(term uint64) {
atomic.StoreUint64(&r.currentTerm, term)
}
func (r *raftState) getLastLog() (index, term uint64) {
r.lastLock.Lock()
index = r.lastLogIndex
term = r.lastLogTerm
r.lastLock.Unlock()
return
}
func (r *raftState) setLastLog(index, term uint64) {
r.lastLock.Lock()
r.lastLogIndex = index
r.lastLogTerm = term
r.lastLock.Unlock()
}
func (r *raftState) getLastSnapshot() (index, term uint64) {
r.lastLock.Lock()
index = r.lastSnapshotIndex
term = r.lastSnapshotTerm
r.lastLock.Unlock()
return
}
func (r *raftState) setLastSnapshot(index, term uint64) {
r.lastLock.Lock()
r.lastSnapshotIndex = index
r.lastSnapshotTerm = term
r.lastLock.Unlock()
}
func (r *raftState) getCommitIndex() uint64 {
return atomic.LoadUint64(&r.commitIndex)
}
func (r *raftState) setCommitIndex(index uint64) {
atomic.StoreUint64(&r.commitIndex, index)
}
func (r *raftState) getLastApplied() uint64 {
return atomic.LoadUint64(&r.lastApplied)
}
func (r *raftState) setLastApplied(index uint64) {
atomic.StoreUint64(&r.lastApplied, index)
}
// Start a goroutine and properly handle the race between a routine
// starting and incrementing, and exiting and decrementing.
func (r *raftState) goFunc(f func()) {
r.routinesGroup.Add(1)
go func() {
defer r.routinesGroup.Done()
f()
}()
}
func (r *raftState) waitShutdown() {
r.routinesGroup.Wait()
}
// getLastIndex returns the last index in stable storage.
// Either from the last log or from the last snapshot.
func (r *raftState) getLastIndex() uint64 {
r.lastLock.Lock()
defer r.lastLock.Unlock()
return max(r.lastLogIndex, r.lastSnapshotIndex)
}
// getLastEntry returns the last index and term in stable storage.
// Either from the last log or from the last snapshot.
func (r *raftState) getLastEntry() (uint64, uint64) {
r.lastLock.Lock()
defer r.lastLock.Unlock()
if r.lastLogIndex >= r.lastSnapshotIndex {
return r.lastLogIndex, r.lastLogTerm
}
return r.lastSnapshotIndex, r.lastSnapshotTerm
}

116
vendor/github.com/hashicorp/raft/tcp_transport.go generated vendored Normal file
View File

@@ -0,0 +1,116 @@
package raft
import (
"errors"
"io"
"log"
"net"
"time"
)
var (
errNotAdvertisable = errors.New("local bind address is not advertisable")
errNotTCP = errors.New("local address is not a TCP address")
)
// TCPStreamLayer implements StreamLayer interface for plain TCP.
type TCPStreamLayer struct {
advertise net.Addr
listener *net.TCPListener
}
// NewTCPTransport returns a NetworkTransport that is built on top of
// a TCP streaming transport layer.
func NewTCPTransport(
bindAddr string,
advertise net.Addr,
maxPool int,
timeout time.Duration,
logOutput io.Writer,
) (*NetworkTransport, error) {
return newTCPTransport(bindAddr, advertise, func(stream StreamLayer) *NetworkTransport {
return NewNetworkTransport(stream, maxPool, timeout, logOutput)
})
}
// NewTCPTransportWithLogger returns a NetworkTransport that is built on top of
// a TCP streaming transport layer, with log output going to the supplied Logger
func NewTCPTransportWithLogger(
bindAddr string,
advertise net.Addr,
maxPool int,
timeout time.Duration,
logger *log.Logger,
) (*NetworkTransport, error) {
return newTCPTransport(bindAddr, advertise, func(stream StreamLayer) *NetworkTransport {
return NewNetworkTransportWithLogger(stream, maxPool, timeout, logger)
})
}
// NewTCPTransportWithLogger returns a NetworkTransport that is built on top of
// a TCP streaming transport layer, using a default logger and the address provider
func NewTCPTransportWithConfig(
bindAddr string,
advertise net.Addr,
config *NetworkTransportConfig,
) (*NetworkTransport, error) {
return newTCPTransport(bindAddr, advertise, func(stream StreamLayer) *NetworkTransport {
config.Stream = stream
return NewNetworkTransportWithConfig(config)
})
}
func newTCPTransport(bindAddr string,
advertise net.Addr,
transportCreator func(stream StreamLayer) *NetworkTransport) (*NetworkTransport, error) {
// Try to bind
list, err := net.Listen("tcp", bindAddr)
if err != nil {
return nil, err
}
// Create stream
stream := &TCPStreamLayer{
advertise: advertise,
listener: list.(*net.TCPListener),
}
// Verify that we have a usable advertise address
addr, ok := stream.Addr().(*net.TCPAddr)
if !ok {
list.Close()
return nil, errNotTCP
}
if addr.IP.IsUnspecified() {
list.Close()
return nil, errNotAdvertisable
}
// Create the network transport
trans := transportCreator(stream)
return trans, nil
}
// Dial implements the StreamLayer interface.
func (t *TCPStreamLayer) Dial(address ServerAddress, timeout time.Duration) (net.Conn, error) {
return net.DialTimeout("tcp", string(address), timeout)
}
// Accept implements the net.Listener interface.
func (t *TCPStreamLayer) Accept() (c net.Conn, err error) {
return t.listener.Accept()
}
// Close implements the net.Listener interface.
func (t *TCPStreamLayer) Close() (err error) {
return t.listener.Close()
}
// Addr implements the net.Listener interface.
func (t *TCPStreamLayer) Addr() net.Addr {
// Use an advertise addr if provided
if t.advertise != nil {
return t.advertise
}
return t.listener.Addr()
}

124
vendor/github.com/hashicorp/raft/transport.go generated vendored Normal file
View File

@@ -0,0 +1,124 @@
package raft
import (
"io"
"time"
)
// RPCResponse captures both a response and a potential error.
type RPCResponse struct {
Response interface{}
Error error
}
// RPC has a command, and provides a response mechanism.
type RPC struct {
Command interface{}
Reader io.Reader // Set only for InstallSnapshot
RespChan chan<- RPCResponse
}
// Respond is used to respond with a response, error or both
func (r *RPC) Respond(resp interface{}, err error) {
r.RespChan <- RPCResponse{resp, err}
}
// Transport provides an interface for network transports
// to allow Raft to communicate with other nodes.
type Transport interface {
// Consumer returns a channel that can be used to
// consume and respond to RPC requests.
Consumer() <-chan RPC
// LocalAddr is used to return our local address to distinguish from our peers.
LocalAddr() ServerAddress
// AppendEntriesPipeline returns an interface that can be used to pipeline
// AppendEntries requests.
AppendEntriesPipeline(id ServerID, target ServerAddress) (AppendPipeline, error)
// AppendEntries sends the appropriate RPC to the target node.
AppendEntries(id ServerID, target ServerAddress, args *AppendEntriesRequest, resp *AppendEntriesResponse) error
// RequestVote sends the appropriate RPC to the target node.
RequestVote(id ServerID, target ServerAddress, args *RequestVoteRequest, resp *RequestVoteResponse) error
// InstallSnapshot is used to push a snapshot down to a follower. The data is read from
// the ReadCloser and streamed to the client.
InstallSnapshot(id ServerID, target ServerAddress, args *InstallSnapshotRequest, resp *InstallSnapshotResponse, data io.Reader) error
// EncodePeer is used to serialize a peer's address.
EncodePeer(id ServerID, addr ServerAddress) []byte
// DecodePeer is used to deserialize a peer's address.
DecodePeer([]byte) ServerAddress
// SetHeartbeatHandler is used to setup a heartbeat handler
// as a fast-pass. This is to avoid head-of-line blocking from
// disk IO. If a Transport does not support this, it can simply
// ignore the call, and push the heartbeat onto the Consumer channel.
SetHeartbeatHandler(cb func(rpc RPC))
}
// WithClose is an interface that a transport may provide which
// allows a transport to be shut down cleanly when a Raft instance
// shuts down.
//
// It is defined separately from Transport as unfortunately it wasn't in the
// original interface specification.
type WithClose interface {
// Close permanently closes a transport, stopping
// any associated goroutines and freeing other resources.
Close() error
}
// LoopbackTransport is an interface that provides a loopback transport suitable for testing
// e.g. InmemTransport. It's there so we don't have to rewrite tests.
type LoopbackTransport interface {
Transport // Embedded transport reference
WithPeers // Embedded peer management
WithClose // with a close routine
}
// WithPeers is an interface that a transport may provide which allows for connection and
// disconnection. Unless the transport is a loopback transport, the transport specified to
// "Connect" is likely to be nil.
type WithPeers interface {
Connect(peer ServerAddress, t Transport) // Connect a peer
Disconnect(peer ServerAddress) // Disconnect a given peer
DisconnectAll() // Disconnect all peers, possibly to reconnect them later
}
// AppendPipeline is used for pipelining AppendEntries requests. It is used
// to increase the replication throughput by masking latency and better
// utilizing bandwidth.
type AppendPipeline interface {
// AppendEntries is used to add another request to the pipeline.
// The send may block which is an effective form of back-pressure.
AppendEntries(args *AppendEntriesRequest, resp *AppendEntriesResponse) (AppendFuture, error)
// Consumer returns a channel that can be used to consume
// response futures when they are ready.
Consumer() <-chan AppendFuture
// Close closes the pipeline and cancels all inflight RPCs
Close() error
}
// AppendFuture is used to return information about a pipelined AppendEntries request.
type AppendFuture interface {
Future
// Start returns the time that the append request was started.
// It is always OK to call this method.
Start() time.Time
// Request holds the parameters of the AppendEntries call.
// It is always OK to call this method.
Request() *AppendEntriesRequest
// Response holds the results of the AppendEntries call.
// This method must only be called after the Error
// method returns, and will only be valid on success.
Response() *AppendEntriesResponse
}

133
vendor/github.com/hashicorp/raft/util.go generated vendored Normal file
View File

@@ -0,0 +1,133 @@
package raft
import (
"bytes"
crand "crypto/rand"
"fmt"
"math"
"math/big"
"math/rand"
"time"
"github.com/hashicorp/go-msgpack/codec"
)
func init() {
// Ensure we use a high-entropy seed for the psuedo-random generator
rand.Seed(newSeed())
}
// returns an int64 from a crypto random source
// can be used to seed a source for a math/rand.
func newSeed() int64 {
r, err := crand.Int(crand.Reader, big.NewInt(math.MaxInt64))
if err != nil {
panic(fmt.Errorf("failed to read random bytes: %v", err))
}
return r.Int64()
}
// randomTimeout returns a value that is between the minVal and 2x minVal.
func randomTimeout(minVal time.Duration) <-chan time.Time {
if minVal == 0 {
return nil
}
extra := (time.Duration(rand.Int63()) % minVal)
return time.After(minVal + extra)
}
// min returns the minimum.
func min(a, b uint64) uint64 {
if a <= b {
return a
}
return b
}
// max returns the maximum.
func max(a, b uint64) uint64 {
if a >= b {
return a
}
return b
}
// generateUUID is used to generate a random UUID.
func generateUUID() string {
buf := make([]byte, 16)
if _, err := crand.Read(buf); err != nil {
panic(fmt.Errorf("failed to read random bytes: %v", err))
}
return fmt.Sprintf("%08x-%04x-%04x-%04x-%12x",
buf[0:4],
buf[4:6],
buf[6:8],
buf[8:10],
buf[10:16])
}
// asyncNotifyCh is used to do an async channel send
// to a single channel without blocking.
func asyncNotifyCh(ch chan struct{}) {
select {
case ch <- struct{}{}:
default:
}
}
// drainNotifyCh empties out a single-item notification channel without
// blocking, and returns whether it received anything.
func drainNotifyCh(ch chan struct{}) bool {
select {
case <-ch:
return true
default:
return false
}
}
// asyncNotifyBool is used to do an async notification
// on a bool channel.
func asyncNotifyBool(ch chan bool, v bool) {
select {
case ch <- v:
default:
}
}
// Decode reverses the encode operation on a byte slice input.
func decodeMsgPack(buf []byte, out interface{}) error {
r := bytes.NewBuffer(buf)
hd := codec.MsgpackHandle{}
dec := codec.NewDecoder(r, &hd)
return dec.Decode(out)
}
// Encode writes an encoded object to a new bytes buffer.
func encodeMsgPack(in interface{}) (*bytes.Buffer, error) {
buf := bytes.NewBuffer(nil)
hd := codec.MsgpackHandle{}
enc := codec.NewEncoder(buf, &hd)
err := enc.Encode(in)
return buf, err
}
// backoff is used to compute an exponential backoff
// duration. Base time is scaled by the current round,
// up to some maximum scale factor.
func backoff(base time.Duration, round, limit uint64) time.Duration {
power := min(round, limit)
for power > 2 {
base *= 2
power--
}
return base
}
// Needed for sorting []uint64, used to determine commitment
type uint64Slice []uint64
func (p uint64Slice) Len() int { return len(p) }
func (p uint64Slice) Less(i, j int) bool { return p[i] < p[j] }
func (p uint64Slice) Swap(i, j int) { p[i], p[j] = p[j], p[i] }