Commit Graph

197 Commits (5c60478953b46e91074a54f5f6683b8afbef8968)

Author SHA1 Message Date
Dylan.Wen a6eb2939b1 raft: add test cases to improve test coverage 2016-09-28 10:19:30 +08:00
Xiang Li cfe717e926 Merge pull request #6275 from xiang90/raft_l
raft: support safe readonly request
2016-09-13 01:36:04 -05:00
Xiang Li 710b14ce56 raft: support safe readonly request
Implement raft readonly request described in raft thesis 6.4
along with the existing clock/lease based approach.
2016-09-12 15:13:52 +08:00
Dylan.Wen 68f2fdc1ff raft: add test case for leader transfer from follower 2016-09-08 17:22:52 +08:00
Ben Darnell a7a867c1e6 raft: Allow an election immediately after start with checkQuorum
Previously, the checkQuorum flag required an election timeout to
expire before a node could cast its first vote. This change permits
the node to cast a vote at any time when the leader is not known,
including immediately after startup.
2016-08-30 08:28:41 +08:00
swingbach@gmail.com 41dee84733 raft: fix #6096 2016-08-04 18:31:22 +08:00
swingbach@gmail.com c0a8da7fd0 raft: minor refactor 2016-08-02 08:46:43 +08:00
swingbach@gmail.com 992f628e6e raft: fix #6068 2016-07-30 03:27:29 +08:00
Xiang Li 7432e9fbe9 Merge pull request #5809 from swingbach/master
raft: make leader transferring workable when quorum check is on
2016-07-12 09:46:18 -07:00
Gyu-Ho Lee 6f3a40cb53 raft: set leader id in stepFollower
Follower has already set its leader ID from
previous append messages from the leader, but
to be consistent,  this adds a line to set its
leader id from leader snapshot message.
2016-07-11 16:37:31 -07:00
swingbach@gmail.com 0d9b6ba0ab raft: fix a few problems 2016-07-11 14:59:53 +08:00
swingbach@gmail.com e020b2a228 raft: make leader transferring workable when quorum check is on 2016-06-29 18:24:58 +08:00
swingbach@gmail.com 0faae33ace raft: implemented read-only query when quorum check is on 2016-06-28 10:52:53 +08:00
Xiang Li 9fee7732f6 Merge pull request #5468 from swingbach/master
implemented leader lease when quorum check is on.
2016-06-01 16:10:41 -07:00
swingbach@gmail.com 337ef64ed5 raft: implemented leader lease when quorum check is on 2016-06-02 06:17:27 +08:00
swingbach@gmail.com ff9d16a2e0 raft: fix tiny mistake of message type 2016-05-20 14:04:08 +08:00
swingbach@gmail.com 1e54117580 raft: add more comments for dueling candidates test case 2016-05-19 13:51:20 +08:00
swingbach@gmail.com c703ccab63 raft: add more assertions for dueling candidates test case 2016-05-19 13:50:14 +08:00
Xiang Li 910781ef5b raft: do not panic when removing all the nodes from cluster 2016-05-16 10:04:17 -07:00
Gyu-Ho Lee fe884f8209 raft: update LICENSE header 2016-05-12 20:49:15 -07:00
Gyu-Ho Lee 9108af9046 *: clean up from go vet, misspell 2016-04-10 23:16:56 -07:00
es-chow ac059eb8cb raft: transfer leader feature 2016-04-08 16:56:32 +08:00
Tamir Duberstein 68db18667a raft: correct doc comment 2016-04-06 08:43:42 -04:00
Gyu-Ho Lee c09f23c46d *: clean up bool comparison 2016-04-02 18:27:54 -07:00
Xiang Li 5d431b4782 raft: lower split vote rate 2016-04-01 12:11:03 -07:00
Anthony Romano 20461ab11a *: fix many typos 2016-01-31 21:42:39 -08:00
Gyu-Ho Lee 71c2a9bb3c *: fix minor typos, comments 2016-01-30 18:15:56 -08:00
Ben Darnell 22925a1d2f raft: Remove redundant `raft.Commit` field.
Keeping this field in sync with `raft.raftLog.committed` was
error-prone, so instead we synthesize the `HardState` on demand.

Fixes #4278.
2016-01-26 15:18:55 -05:00
Ben Darnell 46bb2582fe raft: Call maybeCommit after removing a node.
removeNode reduces the required quorum size, so some pending entries may
be able to commit after it is applied.

Discovered in cockroachdb/cockroach#3642
2016-01-20 11:05:48 -04:00
Hitoshi Mitake 9b2da76796 raft: remove go vet compliants 2015-12-16 13:29:23 +09:00
Xiang Li cc6d98bf89 etcdserver: only send snapshot when the member is active 2015-12-10 16:15:26 -08:00
Xiang Li a8cc1570d0 raft: support quorum check when raft is leader
If quorum check fails, the leader will step down to follower.
2015-11-24 09:36:37 -08:00
Ben Darnell fbeb58d265 raft: no-op instead of panic for Campaigning while leader
We need to be able to force an election (on one node) after creating a
new group (cockroachdb/cockroach#1384), but it is difficult to ensure
that our call to Campaign does not race with an election that may be
started by raft itself. A redundant call to Campaign should be a no-op
instead of a panic. (But the panic in becomeCandidate remains, because
we don't want to update the term or change the committed index in this
case)
2015-11-16 21:44:14 -05:00
Dmitry Smirnov b2f4a5f587 *: fix spelling issues (codespell).
Signed-off-by: Dmitry Smirnov <onlyjob@member.fsf.org>
2015-09-11 10:22:29 +10:00
Xiang Li 6b23a8131f *: test gofmt with -s and fix reported issues 2015-08-21 18:52:16 -07:00
es-chow cc362ccdad raft: set logger to raft so log context such as multinode groupID can be logged 2015-08-12 22:56:00 +08:00
Xiang Li b4022899eb raft: fix panic in send app
sendApp accesses the storage several times. Perviously, we
assume that the storage will not be modified during the read
opeartions. The assumption is not true since the storage can
be compacted between the read operations. If a compaction
causes a read entries error, we should not painc. Instead, we
can simply retry the sendApp logic until succeed.
2015-06-15 14:23:33 -07:00
Ben Darnell c9d507df11 raft: Use raft.Config in MultiNode. 2015-03-24 15:37:13 -04:00
Xiang Li b3fb052ad4 raft: make peers a prviate field in raft.Config 2015-03-24 11:10:07 -07:00
Xiang Li d9b5b56c82 raft: make raft configurable 2015-03-23 09:55:19 -07:00
Xiang Li 4a64373225 raft: add flow control for progress
Each progress has a inflighs sliding window. When the progress
is in replicate state, inflights will control the sending speed
of the leader.

The leader can have at most maxInflight number of inflight
messages for each replicate progress. Receving a appResp moves
forward the sliding window. Heartbeat response free one
slot if the window is full.
2015-03-20 20:04:33 -07:00
Xiang Li 7571b2cde2 raft: limit the size of msgApp
limit the max size of entries sent per message.
Lower the cost at probing state as we limit the size per message;
lower the penalty when aggressively decrease to a too low next.
2015-03-18 15:59:30 -07:00
Yicheng Qin 67194c0b22 raft: introduce progress states 2015-03-18 08:16:32 -07:00
Yicheng Qin be0bf2a2bd raft: fall back to bad path when unreachable 2015-03-11 13:21:23 -07:00
Yicheng Qin fbd5c81139 raft: remove shadowing of variables from test 2015-02-28 12:09:33 -08:00
Xiang Li 2af33fd494 raft: add reportUnreachable 2015-02-28 10:45:22 -08:00
Xiang Li 5ede18be74 raft: separate compact and createsnap in memory storage 2015-02-28 10:08:30 -08:00
Barak Michener 92dca0af0f *: remove shadowing of variables from etcd and add travis test
We've been bitten by this enough times that I wrote a tool so that
it never happens again.
2015-02-17 16:31:42 -05:00
Ben Darnell 33d2400063 raft: Send any waiting appends after receiving MsgAppResp.
This addresses a problem that comes up in the cockroach tests,
in which the order of messages may lead to deadlocks (due to
the fact that we don't have regular heartbeat timers in most
of our tests).
2015-01-27 17:43:29 -05:00
Jonathan Boulle f1ed69e883 *: switch to line comments for copyright
Build tags are not compatible with block comments.
Also adds copyright header to a few places it was missing.
2015-01-26 09:53:30 -08:00