76a3de9a33
Require a non-nil Storage parameter in newLog.
...
Callers must in general have a reference to their Storage objects to
transfer entries from Ready to Storage, so it doesn't make sense to
create a hidden Storage for them.
By explicitly creating Storage objects in tests we can remove a
few casts of raftLog's storage field.
2014-11-12 16:38:50 -05:00
78cbb1512c
raft: nodes return sorted ids
...
This makes raft.softState return the same result when its soft state is
not changed.
2014-11-11 22:58:15 -08:00
25b6590547
raft: introduce log storage interface.
...
This change splits the raftLog.entries array into an in-memory
"unstable" list and a pluggable interface for retrieving entries that
have been persisted to disk. An in-memory implementation of this
interface is provided which behaves the same as the old version;
in a future commit etcdserver could replace the MemoryStorage with
one backed by the WAL.
2014-11-10 17:40:39 -05:00
21987c8701
raft: remove raftLog.resetUnstable and resetNextEnts
...
These methods are no longer used outside of tests and are redundant with
the new stableTo and appliedTo methods.
2014-11-06 17:18:00 -05:00
b99633207c
raft: minor cleanup in comments
2014-10-30 10:01:44 -07:00
0cf0cb3d02
raft: add tests for progress.maybeDecr
2014-10-29 22:57:25 -07:00
81304b2b7e
raft: move test helper function into tests
2014-10-29 14:04:45 -07:00
bffe611fe6
raft: add tests based on section 5.2 in raft paper
2014-10-29 10:17:09 -07:00
507300130b
raft: add tests for ignoring heartbeat reply
2014-10-24 11:50:21 -07:00
e200d2a8e2
etcdserver/raft: remove msgDenied, removedNodes, shouldStop
...
The future plan is to do all these in etcdserver level.
2014-10-20 15:13:18 -07:00
7a4d42166b
*: add license header to all source files
2014-10-17 15:41:22 -07:00
fc42bdb904
raft: remove unused compactThreshold
2014-10-16 17:11:10 -07:00
32c38820c1
raft: protobuf messageType
2014-10-13 11:13:43 -07:00
1eb1020717
raft: fix raft test
2014-10-09 14:42:29 +08:00
8bbbaa88b2
*: raft related int64 -> uint64
2014-10-09 14:29:21 +08:00
af5b8c6c44
raft: int64 -> uint64
2014-10-09 14:26:43 +08:00
38af14b0f4
Merge pull request #1260 from coreos/snap_rm
...
raft: save removed nodes in snapshot
2014-10-09 08:00:33 +08:00
73f2aaf98f
raft: removedSlice -> removedNodes
2014-10-09 06:55:25 +08:00
c67fd14fe8
Merge pull request #1257 from bdarnell/cleanups
...
Raft: assorted cleanups (golint and go vet)
2014-10-09 05:55:21 +08:00
d2e858587f
Raft: a few more improvements to test messages.
2014-10-08 15:07:11 -04:00
7b61565c0a
raft: save removed nodes in snapshot
2014-10-08 15:33:55 +08:00
1cd3345e00
raft: address issues with election timeout
2014-10-08 07:41:17 +08:00
36558b1924
Raft: fix printf strings found by go vet.
2014-10-07 18:44:06 -04:00
3ad0df3722
Raft: fix problems reported by golint.
2014-10-07 18:44:06 -04:00
f65d117462
raft: add a test for randElectionTimeout
2014-10-07 20:34:15 +08:00
d7d6f84f64
raft: rand election timeout
2014-10-07 20:12:49 +08:00
45e4a8643a
raft: add tests for raft.compact
2014-10-07 16:03:11 +08:00
5587e0d73f
raft: compact takes index and nodes parameters
...
Before this commit, compact always compact log at current appliedindex of raft.
This prevents us from doing non-blocking snapshot since we have to make snapshot
and compact atomically. To prepare for non-blocking snapshot, this commit make
compact supports index and nodes parameters. After completing snapshot, the applier
should call compact with the snapshot index and the nodes at snapshot index to do
a compaction at snapsohot index.
2014-10-07 16:03:11 +08:00
01ecc60a88
Merge pull request #1203 from coreos/fix_raft
...
raft: commitIndex=min(leaderCommit, index of last new entry)
2014-10-03 22:24:07 +08:00
172bd7d096
raft: add test for maybeappend change
2014-10-03 22:21:35 +08:00
16ba77767e
raft: do not decrease nextIndex and send entries for stale reply
2014-10-03 13:41:27 +08:00
8490904f20
Merge pull request #1224 from unihorn/149
...
raft: msg.Denied -> msg.Reject
2014-10-02 12:24:09 -07:00
fff918c672
raft: msg.Denied -> msg.Reject
...
Change the field name because it has msgDenied already.
2014-10-02 12:22:12 -07:00
182c8316e1
raft: refine comment for doc and removed list tests
2014-10-01 14:57:39 -07:00
e4a6c9651a
raft: add removed
...
The usage of removed:
1. tell removed node about its removal explicitly using msgDenied
2. prevent removed node disrupt cluster progress by launching leader election
It is set when apply node removal, or receive msgDenied.
2014-10-01 14:57:38 -07:00
d7b4e44a66
raft: heartbeat is only response for maintaining leader dominance
2014-09-29 16:57:43 -07:00
e26ff32fd8
raft: fix error msg
2014-09-28 21:17:51 -07:00
51529cc3f2
raft: remove index field in msg AppResp
2014-09-28 21:13:53 -07:00
adefd83855
raft: remove index field in msg voteResp
2014-09-28 21:13:43 -07:00
86473d8a27
raft: add msg denied field
2014-09-28 21:13:33 -07:00
1ca03d8e9d
raft: move logic to separate func
2014-09-24 10:23:44 -07:00
b07be74a82
raft: stop tickElection when the node is not in peer list
...
This prevents the bug like this:
1. a node sends join to a cluster and succeeds
2. it starts with empty peers and waits for sync, but it have not
received anything
3. election timeout passes, and it promotes itself to leader
4. it commits some log entry
5. its log conflicts with the cluster's
2014-09-23 23:15:02 -07:00
bc7b0108dc
raft: ConfigChange -> ConfChange
2014-09-23 12:02:44 -07:00
d92931853e
raft: Config -> ConfigChange
...
Configure -> ProposeConfigChange
AddNode, RemoveNode -> ApplyConfigChange
2014-09-22 23:39:53 -07:00
b82d70871f
raft: use EntryType in protobuf
2014-09-22 15:44:46 -07:00
ff6705b94b
raft: add Configure, AddNode, RemoveNode
...
Configure is used to propose config change. AddNode and RemoveNode is
used to apply cluster change to raft state machine. They are the
basics for dynamic configuration.
2014-09-22 15:43:13 -07:00
023dc7cba2
etcdserver: add SYNC request
2014-09-16 13:42:03 -07:00
f9c12e2053
Merge pull request #1075 from coreos/fix_heartbeat
...
raft: fix heartbeat
2014-09-15 10:04:12 -07:00
21d116d3e1
raft: fix heartbeat
2014-09-15 09:58:22 -07:00
e7ea6a374a
main: check node id is not noneid
2014-09-14 23:28:11 -07:00
2927cc6e3b
raft: refine plus assignment
2014-09-14 22:07:32 -07:00
30289dad5c
raft: more descriptive panic info
2014-09-12 11:39:53 -07:00
dc1357afa9
raft: use raft helper function in testing
2014-09-11 19:26:29 -07:00
0e0fc2bd24
raft: use [1,n] as address list for raft state machines
2014-09-11 19:23:05 -07:00
84c2bd0b7d
raft: set From of local message to be local id
2014-09-11 19:18:47 -07:00
4e31bb308d
raft: use const firstId to track ids
2014-09-09 19:07:20 -07:00
b50f96e2e1
raft: set none to be 0
2014-09-09 16:47:21 -07:00
9997c9488a
*: fix a few small issues identified by go vet
2014-09-08 23:52:36 -07:00
8dfa490e49
raft: fix ignore term
2014-09-04 14:33:51 -07:00
8d9b7b1680
raft: remove entry type
2014-09-03 15:24:47 -07:00
8463421448
raft: remove configuration
2014-09-03 15:23:05 -07:00
3c77693881
raft: make tick send out messages. all state machine actions should be taken inside step function.
2014-09-03 10:17:09 -07:00
d218034630
boom
2014-09-03 09:20:20 -07:00
e8e588c67b
raft: move protobufs into raftpb
2014-09-03 09:20:17 -07:00
ddd219f297
many: marshal message
2014-09-03 09:20:16 -07:00
ad307c6965
raft: go fmt
2014-09-03 09:20:14 -07:00
4aa15294a8
raft: re-remove clusterId from raft
2014-09-03 09:20:14 -07:00
134a962222
raft: move raft2 to raft
2014-09-03 09:20:14 -07:00
0453d09af6
raft: moved into new raft
2014-09-03 09:20:11 -07:00
a5df254e53
raft: add clusterId to snapshot
2014-09-03 09:20:08 -07:00
01c40fcf50
raft: not recover from outdated snapshot
2014-09-03 09:20:02 -07:00
ba63cf666d
raft: add recover
2014-09-03 09:20:02 -07:00
6030261363
etcd/raft: add snap
2014-09-03 09:20:02 -07:00
54b4f52e48
raft: add index to entry
2014-09-03 09:20:01 -07:00
311db876b0
raft: add unstableState
2014-09-03 09:19:58 -07:00
1288e1f39d
raft: log->raftlog
2014-09-03 09:19:58 -07:00
2665cc1cc8
raft: heartbeat should not contain entries
2014-09-03 09:19:57 -07:00
79689872af
raft: use From field to distinguish local message
2014-09-03 09:19:56 -07:00
3921295b21
raft: update lead to none when receives vaild msgVote
2014-09-03 09:19:50 -07:00
447d7dc51b
raft: fix log append; add tests
2014-09-03 09:19:49 -07:00
0886e0ddf4
raft: change term to atomicInt
2014-09-03 09:05:14 -07:00
30f4d9faea
raft: change index and term to int64
2014-09-03 09:05:14 -07:00
e11c7f35b4
raft: add atomicInt
2014-09-03 09:05:14 -07:00
2af0ad505a
raft: make id int64
2014-09-03 09:05:12 -07:00
5651272ec8
raft: handle snapshot message
2014-09-03 09:05:12 -07:00
2a11c1487c
raft: sm.compact and sm.restore
2014-09-03 09:05:12 -07:00
62a90e77b3
raft: send Normal with nil Data when leader is elected out
2014-09-03 09:05:12 -07:00
c86f484712
raft: make judge clear
2014-09-03 09:05:12 -07:00
7dc02b947d
raft: refine TestVote
2014-09-03 09:05:12 -07:00
095251f1fa
raft: Network supports discontinuous ids
2014-09-03 09:05:12 -07:00
5778c49689
raft: set term in reset
2014-09-03 09:05:11 -07:00
3817661f82
raft: rename ConfigAdd/ConfigRemove -> AddNode/RemoveNode
2014-09-03 09:05:11 -07:00
9f315ffe10
raft: make entry type public
2014-09-03 09:05:11 -07:00
e5b9e22518
raft: Step returns ok
2014-09-03 09:05:11 -07:00
cc3a8e26c8
raft: add leader change test for configuration
2014-09-03 09:05:10 -07:00
abd2448931
raft: addr -> id
2014-09-03 09:05:10 -07:00
1a75beb57c
raft: add confAdd and confRemove entry type
2014-09-03 09:05:09 -07:00
c03fbf68d6
raft: add conf safety
...
To make configuration change safe without adding configuration protocol:
1. We only allow to add/remove one node at a time.
2. We only allow one uncommitted configuration entry in the log.
These two rules can make sure there is no disjoint quorums in both current cluster and the
future(after applied any number of committed entries or uncommitted entries in log) clusters.
We add a type field in Entry structure for two reasons:
1. Statemachine needs to know if there is a pending configuration change.
2. Configuration entry should be executed by raft package rather application who is using raft.
2014-09-03 09:05:09 -07:00
853a458a0d
raft: change ins from array to map
2014-09-03 09:05:09 -07:00
5d710c0f7a
raft: add recvMsgBeat test
2014-09-03 09:05:08 -07:00
12572e5412
raft: add state transition test
2014-09-03 09:05:08 -07:00
920b80c41f
raft: add single node commit test
2014-09-03 09:05:08 -07:00
8111d4fbb8
raft: introduce network.cut
2014-09-03 09:05:08 -07:00
a6a63d116f
raft: add a test for commit rule
2014-09-03 09:05:08 -07:00
4d22ff90d5
raft: refactor network simulator
...
Signed-off-by: Blake Mizerany <blake.mizerany@gmail.com>
2014-09-03 09:05:08 -07:00
c32d34166e
raft: leader updates its own match; tries to commit after a prop
2014-09-03 09:05:08 -07:00
167ef7e8b0
raft: test leader replies to appResp
2014-09-03 09:05:08 -07:00
b3de2b3450
raft: testVote - breaks state machine init to multiple lines
2014-09-03 09:05:07 -07:00
59a720d8be
raft: vote test for candidate and leader
2014-09-03 09:05:07 -07:00
93b08502e4
raft: check voteFor
2014-09-03 09:05:07 -07:00
c1c45575be
raft: ins: []*index -> []index
...
It could make raft faster, use less storage.
2014-09-03 09:05:07 -07:00
1170c21f89
raft: add a single candidate test
2014-09-03 09:05:07 -07:00
0cdd1b58a4
raft: rename log.commit to log.committed
2014-09-03 09:05:07 -07:00
4c609ec59c
raft: new log struct
2014-09-03 09:05:06 -07:00
74737b76cc
raft: leader commit and test
...
Signed-off-by: Blake Mizerany <blake.mizerany@gmail.com>
2014-09-03 09:05:06 -07:00
c223eca938
raft: clean stateMachine
2014-09-03 09:05:06 -07:00
73e3394d2d
raft: reverse sort to figure out the ci
2014-09-03 09:05:05 -07:00
50e0db4038
raft: attempt first version of Interface
2014-09-03 09:05:05 -07:00
8d7be33dd8
raft: move to raft pkg
2014-09-03 09:05:05 -07:00
Ben Darnell