810 Commits

Author SHA1 Message Date
Gyu-Ho Lee
0199bdc266 *: fix 'ineffassign' issues
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-10-03 10:14:33 -07:00
Alex Robinson
b9c4f5b22a raft: fix bullet point indentation in README 2017-09-18 16:07:51 -04:00
Xiang
9801fd7297 raft: ensure CheckQuorum is enabled when readonlyoption is lease based 2017-09-17 10:46:12 -07:00
gladiator
58b98c6a14 raft: check leader request when becomeFollower 2017-09-15 08:23:18 +08:00
Gyu-Ho Lee
f65aee0759 *: replace 'golang.org/x/net/context' with 'context'
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-09-07 13:39:42 -07:00
irfan sharif
248384a468 raft: (re-)introduce TestRecvMsgPreVote
TestRecvMsgPreVote was intended to be introduced in
github.com/coreos/etcd/pull/6624 but was uncapitalized (search for
testRecvMsgPreVote instead) and then subsequently removed due to it
being unused.
2017-09-01 10:45:47 -04:00
Janardhan Reddy
b847cde981 raft: update doc for persisting entries before hardstate 2017-08-31 16:24:28 +10:00
gladiator
42cc64a9e5 raft: add TestPreVoteWithSplitVote 2017-08-02 17:59:28 +08:00
gladiator
3740793b42 raft: reset votes when becomePreCandidate 2017-08-01 22:42:09 +08:00
irfan sharif
a92ceeec25 raft: introduce/fix TestNodeWithSmallerTermCanCompleteElection
TestNodeWithSmallerTermCanCompleteElection tests the scenario where a
node that has been partitioned away (and fallen behind) rejoins the
cluster at about the same time the leader node gets partitioned away.
Previously the cluster would come to a standstill when run with PreVote
enabled.

When responding to Msg{Pre,}Vote messages we now include the term from
the message, not the local term. To see why consider the case where a
single node was previously partitioned away and it's local term is now
of date. If we include the local term (recall that for pre-votes we
don't update the local term), the (pre-)campaigning node on the other
end will proceed to ignore the message (it ignores all out of date
messages).
The term in the original message and current local term are the same in
the case of regular votes, but different for pre-votes.

NB: Had to change TestRecvMsgVote to include pb.Message.Term when
sending MsgVote messages. The new sanity checks on MsgVoteResp
(m.Term != 0) would panic with the old test as raft.Term would be equal
to 0 when responding with MsgVoteResp messages.
2017-07-21 02:26:02 -04:00
smetro
e461017ac5 raft: add DisableProposalForwarding option
this allows users to disable followers from forwarding proposals to the
leader.
2017-06-21 14:58:28 -07:00
Nikola Kovacs
66687da3ba *: grammar fixes, it's -> its 2017-06-15 18:23:16 +02:00
Anthony Romano
4ebeba0e18 *: regen protofiles with latest protobuf tools 2017-06-12 15:14:43 -07:00
yuleixiao
44a49ff45a raft: return leaderTransferee at raft status 2017-05-11 12:45:56 +08:00
Xiang Li
db6f45e939 Merge pull request #7830 from aaronlehmann/new-nodes-start-active
raft: Set the RecentActive flag for newly added nodes
2017-05-05 08:59:25 -07:00
Aaron Lehmann
9451fa1f9c raft: Add unit test TestAddNodeCheckQuorum
This test verifies that adding a node does not cause the leader to step
down until at least one full ElectionTick cycle elapses.

Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
2017-05-04 15:04:30 -07:00
Tess Rinearson
3d504737e4 add chain core to raft users list 2017-05-02 11:23:25 -07:00
Aaron Lehmann
52613b262b raft: Set the RecentActive flag for newly added nodes
I found that enabling the CheckQuorum flag led to spurious leader
elections when new nodes joined. It looks like in the time between a new
node joining the cluster, and that node first communicating with the
leader, the quorum check could fail because the new node looks inactive.
To solve this, set the RecentActive flag when nodes are first added.
This gives a grace period for the node to communicate before it causes
the quorum check to fail.

Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
2017-04-27 11:19:29 -07:00
Alex Robinson
45406d8486 raft: Avoid holding unneeded memory in unstable log's entries array
Accumulation of old entries in the underlying array backing the
entries slice has been found to cause massive memory growth in
CockroachDB for workloads that do large (1MB) writes
(https://github.com/cockroachdb/cockroach/issues/14776)

This doesn't appear to have much consistent effect on the raft
benchmarks, although it's worth noting that they vary quite a bit
between runs so it's kind of tough to draw strong conclusions from them.
Let me know if there are any different benchmarks you'd like me to run!

Fixes #7746

benchmark              old ns/op     new ns/op     delta
BenchmarkOneNode-8     3283          3125          -4.81%

benchmark              old allocs     new allocs     delta
BenchmarkOneNode-8     6              6              +0.00%

benchmark              old bytes     new bytes     delta
BenchmarkOneNode-8     796           727           -8.67%

benchmark                     old ns/op     new ns/op     delta
BenchmarkProposal3Nodes-8     4269          4337          +1.59%

benchmark                     old allocs     new allocs     delta
BenchmarkProposal3Nodes-8     15             13             -13.33%

benchmark                     old bytes     new bytes     delta
BenchmarkProposal3Nodes-8     5839          4544          -22.18%
2017-04-18 10:55:16 -04:00
Anthony Romano
78a5eb79b5 *: add swagger and grpc-gateway assets for v3lock and v3election 2017-04-10 15:21:07 -07:00
fanmin shi
f18ae033a7 raft: use rs.req.Entries[0].Data as the key for deletion in advance()
advance() should use rs.req.Entries[0].Data as the context instead of
req.Context for deletion. Since req.Context is never set, there won't be
any context being deleted from pendingReadIndex; results mem leak.

FIXES #7571
2017-03-24 12:31:21 -07:00
Gyu-Ho Lee
3d75395875 *: remove never-unused vars, minor lint fix
Signed-off-by: Gyu-Ho Lee <gyuhox@gmail.com>
2017-03-06 14:59:12 -08:00
Anthony Romano
c8a2c7f64f *: eschew you from documentation
Removed line wrapping in affected files as well.
2017-03-06 11:40:46 -08:00
Xiang
931cf3454a raft: make TestNodeTick reliable
TestNodeTick relies on a unreliable func `waitForSchedule` when running
with GOMAXPROCS > 1. This commit changes the test to make sure we stop
the node afte it drains the tick chan. The test should be reliable now.
2017-03-01 17:35:58 -08:00
Dylan.Wen
bc6bebe7b0 raft: revise test case and fix typo 2017-02-21 15:23:42 +08:00
Dylan.Wen
9342647e0c raft: fix read index request for #7331 2017-02-17 09:45:41 +08:00
Peter Mattis
ab03a42f06 raft: add Ready.MustSync
Add Ready.MustSync which indicates that the hard state and raft log
entries in a Ready message must be synchronously written to persistent
storage.
2017-02-13 15:13:21 -05:00
Manjunath A Kumatagi
0914b8b707 test: Fix gosimple errors
Getting gosimple suggestion while running test script, so this PR is for fixing gosimple S1019 check.
raft/node_test.go:456:40: should use make([]raftpb.Entry, 1) instead (S1019)
raft/node_test.go:457:49: should use make([]raftpb.Entry, 1) instead (S1019)
raft/node_test.go:458:43: should use make([]raftpb.Message, 1) instead (S1019)

Refer https://github.com/dominikh/go-tools/blob/master/cmd/gosimple/README.md#checks for more information.
2017-02-09 08:01:28 -05:00
Xiang Li
b32a8010a7 Merge pull request #7121 from hhkbp2/add-test-case
raft: add RawNode test case for #6866
2017-01-09 23:37:23 -08:00
Dylan.Wen
16135165c2 raft: add RawNode test case for #6866 2017-01-10 10:55:57 +08:00
Dylan.Wen
7bb768ba34 raft: fix test case for #7042 2017-01-09 16:52:02 +08:00
Xiang Li
fc8cd44c72 raft: use status to test node stop
n.Tick() is async. It can be racy when running with n.Stop().

n.Status() is sync and  has a feedback mechnism internally. So there wont be
any race between n.Status() and n.Stop() call.
2017-01-03 15:18:48 -08:00
Xiang Li
b940e0d514 Merge pull request #7042 from petermattis/pmattis/resume-after-heartbeat-resp
raft: resume paused followers on receipt of MsgHeartbeatResp
2016-12-27 21:15:53 -08:00
Dylan.Wen
920b155f17 raft: fix pre-vote tests 2016-12-26 14:31:59 +08:00
Peter Mattis
e625400f1d raft: resume paused followers on receipt of MsgHeartbeatResp
Previously, paused followers were resumed upon sending a MsgHearbeat.

Fixes #7037
2016-12-20 08:22:09 -05:00
杜军
8151d4d0bc raft: make memory storage set method thread safe 2016-12-20 18:48:52 +08:00
Gyu-Ho Lee
0c5d1d5641 raft: simplify boolean comparison, remove unused 2016-12-12 10:07:14 -08:00
Xiang Li
529806dba1 Merge pull request #6935 from bdarnell/election-test
raft: Fix election "logs converge" test
2016-12-06 16:45:39 -08:00
Ben Darnell
f60a5d6025 raft: Export Progress.IsPaused
CockroachDB would like to use this method for monitoring.
2016-12-04 13:14:08 +08:00
Ben Darnell
340ba8353c raft: Fix election "logs converge" test
The "logs converge" case in TestLeaderElectionPreVote was incorrectly
passing because some nodes were not actually using the preVoteConfig.
This test case was more complex than its siblings and it was not
verifying what it wanted to verify, so pull it out into a separate test
where everything can be tested more explicitly.

Fixes #6895
2016-12-03 17:29:15 +08:00
Xiang Li
f2eb8560ed raft: fix TestNodeProposeAddDuplicateNode
Only send signal after applying conf change.
Or deadlock might happen if raft node receives
ready without conf change when the test server
is slow.
2016-11-20 21:59:31 -08:00
Vincent Lee
e6d1ebcc1d raft: use the channel instead of sleep to make test case reliable 2016-11-21 13:30:15 +08:00
Vincent Lee
bc6f5ad53e raft: fix test case for data race 2016-11-21 10:30:36 +08:00
Vincent Lee
62bd5477b9 raft: fix test case, should wait config propose applied 2016-11-21 10:10:34 +08:00
Vincent Lee
16e3ab0f11 raft: test case to check the duplicate add node propose 2016-11-20 16:58:11 +08:00
Vincent Lee
4401d88546 raft: add node should reset the pendingConf state
After add node conf proposed twice with the same node id, the pending state is not reset because
the addNode returned without setting the pending state at the second
time and the pending state will always be true unless other conf changed. During this we
can not add any new node because the propose will be ignored since the
pending state is true.
2016-11-17 15:50:13 +08:00
Alexander Morozov
7afc490c95 raft: return empty status if node is stopped
If the node is stopped, then Status can hang forever because there is no
event loop to answer. So, just return empty status to avoid deadlocks.

Fix #6855

Signed-off-by: Alexander Morozov <lk4d4math@gmail.com>
2016-11-15 15:45:23 -08:00
Gyu-Ho Lee
b8b72f80f9 *: revendor, update proto files 2016-11-10 12:02:00 -08:00
fanmin shi
c2fd42b556 etcdserver, clientv3: add "!=" to txn
adding != to compare is a requested functionality from a etcd user

FIX #6719
2016-11-09 14:28:36 -08:00
Ben Darnell
2f34547d39 raft: Check promotable() in MsgTimeoutNow handling
If MsgTimeoutNow arrived after a node was removed, the node could start
and win an election, then panic in becomeLeader (see
cockroachdb/cockroach#8535)
2016-11-07 20:02:21 +08:00