Marked all 'integrational, e2e' as skipped in the --short mode.
Thanks to this we will be able to significantly simplify ./test script.
The run currently takes ~23s.
With (follow up) move of ~clientv3/snapshot to integration tests (as
part of modularization), we can expect this to fall to 5-10s.
```
% time go test --short ./... --count=1
ok go.etcd.io/etcd/v3 0.098s
? go.etcd.io/etcd/v3/Documentation/learning/lock/client [no test files]
? go.etcd.io/etcd/v3/Documentation/learning/lock/storage [no test files]
ok go.etcd.io/etcd/v3/auth 0.724s
? go.etcd.io/etcd/v3/auth/authpb [no test files]
ok go.etcd.io/etcd/v3/client 0.166s
ok go.etcd.io/etcd/v3/client/integration 0.166s
ok go.etcd.io/etcd/v3/clientv3 3.219s
ok go.etcd.io/etcd/v3/clientv3/balancer 1.102s
? go.etcd.io/etcd/v3/clientv3/balancer/connectivity [no test files]
? go.etcd.io/etcd/v3/clientv3/balancer/picker [no test files]
? go.etcd.io/etcd/v3/clientv3/balancer/resolver/endpoint [no test files]
ok go.etcd.io/etcd/v3/clientv3/clientv3util 0.096s [no tests to run]
ok go.etcd.io/etcd/v3/clientv3/concurrency 3.323s
? go.etcd.io/etcd/v3/clientv3/credentials [no test files]
ok go.etcd.io/etcd/v3/clientv3/integration 0.131s
? go.etcd.io/etcd/v3/clientv3/leasing [no test files]
? go.etcd.io/etcd/v3/clientv3/mirror [no test files]
ok go.etcd.io/etcd/v3/clientv3/namespace 0.041s
ok go.etcd.io/etcd/v3/clientv3/naming 0.115s
ok go.etcd.io/etcd/v3/clientv3/ordering 0.121s
ok go.etcd.io/etcd/v3/clientv3/snapshot 19.325s
ok go.etcd.io/etcd/v3/clientv3/yaml 0.090s
ok go.etcd.io/etcd/v3/contrib/raftexample 7.572s
? go.etcd.io/etcd/v3/contrib/recipes [no test files]
ok go.etcd.io/etcd/v3/embed 0.282s
ok go.etcd.io/etcd/v3/etcdctl 0.054s
? go.etcd.io/etcd/v3/etcdctl/ctlv2 [no test files]
ok go.etcd.io/etcd/v3/etcdctl/ctlv2/command 0.117s
? go.etcd.io/etcd/v3/etcdctl/ctlv3 [no test files]
ok go.etcd.io/etcd/v3/etcdctl/ctlv3/command 0.070s
ok go.etcd.io/etcd/v3/etcdmain 0.172s
ok go.etcd.io/etcd/v3/etcdserver 1.698s
? go.etcd.io/etcd/v3/etcdserver/api [no test files]
ok go.etcd.io/etcd/v3/etcdserver/api/etcdhttp 0.075s
ok go.etcd.io/etcd/v3/etcdserver/api/membership 0.104s
? go.etcd.io/etcd/v3/etcdserver/api/membership/membershippb [no test files]
ok go.etcd.io/etcd/v3/etcdserver/api/rafthttp 0.181s
ok go.etcd.io/etcd/v3/etcdserver/api/snap 0.078s
? go.etcd.io/etcd/v3/etcdserver/api/snap/snappb [no test files]
ok go.etcd.io/etcd/v3/etcdserver/api/v2auth 0.142s
ok go.etcd.io/etcd/v3/etcdserver/api/v2discovery 0.035s
ok go.etcd.io/etcd/v3/etcdserver/api/v2error 0.043s
ok go.etcd.io/etcd/v3/etcdserver/api/v2http 0.070s
ok go.etcd.io/etcd/v3/etcdserver/api/v2http/httptypes 0.031s
? go.etcd.io/etcd/v3/etcdserver/api/v2stats [no test files]
ok go.etcd.io/etcd/v3/etcdserver/api/v2store 0.645s
ok go.etcd.io/etcd/v3/etcdserver/api/v2v3 0.218s
? go.etcd.io/etcd/v3/etcdserver/api/v3alarm [no test files]
? go.etcd.io/etcd/v3/etcdserver/api/v3client [no test files]
ok go.etcd.io/etcd/v3/etcdserver/api/v3compactor 1.765s
? go.etcd.io/etcd/v3/etcdserver/api/v3election [no test files]
? go.etcd.io/etcd/v3/etcdserver/api/v3election/v3electionpb [no test files]
? go.etcd.io/etcd/v3/etcdserver/api/v3election/v3electionpb/gw [no test files]
? go.etcd.io/etcd/v3/etcdserver/api/v3lock [no test files]
? go.etcd.io/etcd/v3/etcdserver/api/v3lock/v3lockpb [no test files]
? go.etcd.io/etcd/v3/etcdserver/api/v3lock/v3lockpb/gw [no test files]
ok go.etcd.io/etcd/v3/etcdserver/api/v3rpc 0.091s
ok go.etcd.io/etcd/v3/etcdserver/api/v3rpc/rpctypes 0.012s
ok go.etcd.io/etcd/v3/etcdserver/cindex 0.054s
ok go.etcd.io/etcd/v3/etcdserver/etcdserverpb 0.039s
? go.etcd.io/etcd/v3/etcdserver/etcdserverpb/gw [no test files]
ok go.etcd.io/etcd/v3/functional/agent 0.094s
? go.etcd.io/etcd/v3/functional/cmd/etcd-agent [no test files]
? go.etcd.io/etcd/v3/functional/cmd/etcd-proxy [no test files]
? go.etcd.io/etcd/v3/functional/cmd/etcd-runner [no test files]
? go.etcd.io/etcd/v3/functional/cmd/etcd-tester [no test files]
ok go.etcd.io/etcd/v3/functional/rpcpb 0.060s
? go.etcd.io/etcd/v3/functional/runner [no test files]
ok go.etcd.io/etcd/v3/functional/tester 0.079s
ok go.etcd.io/etcd/v3/integration 0.684s
ok go.etcd.io/etcd/v3/integration/embed 0.101s
ok go.etcd.io/etcd/v3/lease 3.455s
ok go.etcd.io/etcd/v3/lease/leasehttp 2.185s
? go.etcd.io/etcd/v3/lease/leasepb [no test files]
ok go.etcd.io/etcd/v3/mvcc 7.246s
ok go.etcd.io/etcd/v3/mvcc/backend 0.354s
? go.etcd.io/etcd/v3/mvcc/mvccpb [no test files]
ok go.etcd.io/etcd/v3/pkg/adt 0.025s
? go.etcd.io/etcd/v3/pkg/contention [no test files]
? go.etcd.io/etcd/v3/pkg/cpuutil [no test files]
ok go.etcd.io/etcd/v3/pkg/crc 0.008s
? go.etcd.io/etcd/v3/pkg/debugutil [no test files]
ok go.etcd.io/etcd/v3/pkg/expect 0.015s
ok go.etcd.io/etcd/v3/pkg/fileutil 0.268s
ok go.etcd.io/etcd/v3/pkg/flags 0.021s
ok go.etcd.io/etcd/v3/pkg/httputil 0.020s
ok go.etcd.io/etcd/v3/pkg/idutil 0.008s
ok go.etcd.io/etcd/v3/pkg/ioutil 0.025s
ok go.etcd.io/etcd/v3/pkg/logutil 0.047s
? go.etcd.io/etcd/v3/pkg/mock/mockserver [no test files]
? go.etcd.io/etcd/v3/pkg/mock/mockstorage [no test files]
? go.etcd.io/etcd/v3/pkg/mock/mockstore [no test files]
? go.etcd.io/etcd/v3/pkg/mock/mockwait [no test files]
ok go.etcd.io/etcd/v3/pkg/netutil 1.024s
ok go.etcd.io/etcd/v3/pkg/osutil 0.021s
ok go.etcd.io/etcd/v3/pkg/pathutil 0.008s
ok go.etcd.io/etcd/v3/pkg/pbutil 0.008s
ok go.etcd.io/etcd/v3/pkg/proxy 4.081s
ok go.etcd.io/etcd/v3/pkg/report 0.008s
? go.etcd.io/etcd/v3/pkg/runtime [no test files]
ok go.etcd.io/etcd/v3/pkg/schedule 0.009s
ok go.etcd.io/etcd/v3/pkg/srv 0.019s
ok go.etcd.io/etcd/v3/pkg/stringutil 0.008s
? go.etcd.io/etcd/v3/pkg/systemd [no test files]
ok go.etcd.io/etcd/v3/pkg/testutil 0.023s
ok go.etcd.io/etcd/v3/pkg/tlsutil 3.965s
ok go.etcd.io/etcd/v3/pkg/traceutil 0.034s
ok go.etcd.io/etcd/v3/pkg/transport 0.532s
ok go.etcd.io/etcd/v3/pkg/types 0.028s
ok go.etcd.io/etcd/v3/pkg/wait 0.023s
ok go.etcd.io/etcd/v3/proxy/grpcproxy 0.101s
? go.etcd.io/etcd/v3/proxy/grpcproxy/adapter [no test files]
? go.etcd.io/etcd/v3/proxy/grpcproxy/cache [no test files]
ok go.etcd.io/etcd/v3/proxy/httpproxy 0.044s
ok go.etcd.io/etcd/v3/proxy/tcpproxy 0.047s
ok go.etcd.io/etcd/v3/raft 0.312s
ok go.etcd.io/etcd/v3/raft/confchange 0.183s
ok go.etcd.io/etcd/v3/raft/quorum 0.316s
ok go.etcd.io/etcd/v3/raft/raftpb 0.024s
ok go.etcd.io/etcd/v3/raft/rafttest 0.640s
ok go.etcd.io/etcd/v3/raft/tracker 0.026s
ok go.etcd.io/etcd/v3/tests/e2e 0.077s
? go.etcd.io/etcd/v3/tools/benchmark [no test files]
? go.etcd.io/etcd/v3/tools/benchmark/cmd [no test files]
? go.etcd.io/etcd/v3/tools/etcd-dump-db [no test files]
ok go.etcd.io/etcd/v3/tools/etcd-dump-logs 0.088s
? go.etcd.io/etcd/v3/tools/etcd-dump-metrics [no test files]
? go.etcd.io/etcd/v3/tools/local-tester/bridge [no test files]
? go.etcd.io/etcd/v3/version [no test files]
ok go.etcd.io/etcd/v3/wal 1.517s
? go.etcd.io/etcd/v3/wal/walpb [no test files]
go test --short ./... --count=1 76.12s user 12.57s system 375% cpu 23.635 total
```
* raft: check conf change before campaign
Signed-off-by: Jay Lee <BusyJayLee@gmail.com>
* raft: extract hup function
Signed-off-by: Jay Lee <BusyJayLee@gmail.com>
* raft: check pending conf change for transferleader
Signed-off-by: Jay Lee <BusyJayLee@gmail.com>
This change makes the etcd package compatible with the existing Go
ecosystem for module versioning.
Used this tool to update package imports:
https://github.com/KSubedi/gomove
Apps typically maintain the raft configuration as part of the state
machine. As a result, they want to be able to reject configuration change
entries at apply time based on the state on which the entry is supposed
to be applied. When this happens, the app should not call
ApplyConfChange, but the comments did not make this clear.
As a result, it was tempting to pass an empty pb.ConfChange or it's V2
version instead of not calling ApplyConfChange.
However, an empty V1 or V2 proto aren't noops when the configuration is
joint: an empty V1 change is treated internally as a single
configuration change for NodeID zero and will cause a panic when applied
in a joint state. An empty V2 proto is treated as a signal to leave a
joint state, which means that the app's config and raft's would diverge.
The comments updated in this commit now ask users to not call
ApplyConfState when they reject a conf change. Apps that never use joint
consensus can keep their old behavior since the distinction only matters
when in a joint state, but we don't want to encourage that.
The code doing so was undertested and buggy: it would launch multiple
attempts to transition out when the conf change was not the last element
in the log.
This commit fixes the problem and adds a regression test. It also
reworks the code to handle a former untested edge case, in which the
auto-transition append is refused. This can't happen any more with the
current version of the code because this proposal has size zero and is
special cased in increaseUncommittedSize. Last but not least, the
auto-leave proposal now also bumps pendingConfIndex, which was not done
previously due to an oversight.
When a leader removes itself, it will retain its leadership but not
accept new proposals, making the range effectively stuck until manual
intervention triggers a campaign event.
This commit documents the behavior. It does not correct it yet.
Verifiy the behavior in various v1 and v2 conf change operations.
This also includes various fixups, notably it adds protection
against transitioning in and out of new configs when this is not
permissible.
There are more threads to pull, but those are left for future commits.
When the leader applied a new configuration that added voters, it would
not immediately probe these voters, delaying when they would be caught
up.
I noticed this while writing an interaction-driven test, which has now
been cleaned up and completed.
It is a helper case to attach a debugger to when a problem needs
to be investigated in a longer test file. In such a case, add the
following stanza immediately before the interesting behavior starts:
_breakpoint:
----
ok
and set a breakpoint on the _breakpoint case.
Initializing at LastIndex+1 meant that new peers would not be probed
immediately when they appeared in the leader's config, which delays
their getting caught up.
It has often been tedious to test the interactions between multi-member
Raft groups, especially when many steps were required to reach a certain
scenario. Often, this boilerplate was as boring as it is hard to write
and hard to maintain, making it attractive to resort to shortcuts
whenever possible, which in turn tended to undercut how meaningful and
maintainable the tests ended up being - that is, if the tests were even
written, which sometimes they weren't.
This change introduces a datadriven framework specifically for testing
deterministically the interaction between multiple members of a raft group
with the goal of reducing the friction for writing these tests to near
zero.
In the near term, this will be used to add thorough testing for joint
consensus (which is already available today, but wildly undertested),
but just converting an existing test into this framework has shown that
the concise representation and built-in inspection of log messages
highlights unexpected behavior much more readily than the previous unit
tests did (the test in question is `snapshot_succeed_via_app_resp`; the
reader is invited to compare the old and new version of it).
The main building block is `InteractionEnv`, which holds on to the state
of the whole system and exposes various relevant methods for
manipulating it, including but not limited to adding nodes, delivering
and dropping messages, and proposing configuration changes. All of this
is extensible so that in the future I hope to use it to explore the
phenomena discussed in
https://github.com/etcd-io/etcd/issues/7625#issuecomment-488798263
which requires injecting appropriate "crash points" in the Ready
handling loop. Discussions of the "what if X happened in state Y"
can quickly be made concrete by "scripting up an interaction test".
Additionally, this framework is intentionally not kept internal to the
raft package.. Though this is in its infancy, a goal is that it should
be possible for a suite of interaction tests to allow applications to
validate that their Storage implementation behaves accordingly, simply
by running a raft-provided interaction suite against their Storage.
While writing interaction tests for joint configuration changes, I
realized that this wasn't working yet - restoring had no notion of
the joint configuration and was simply dropping it on the floor.
This commit introduces a helper `confchange.Restore` which takes
a `ConfState` and initializes a `Tracker` from it.
This is then used both in `(*raft).restore` as well as in `newRaft`.
This is helpful for upcoming testing work which allows datadriven
testing of the interaction of multiple nodes. This testing requires
determinism to work correctly.
It turns out that that learners must be allowed to cast votes.
This seems counter- intuitive but is necessary in the situation in which
a learner has been promoted (i.e. is now a voter) but has not learned
about this yet.
For example, consider a group in which id=1 is a learner and id=2 and
id=3 are voters. A configuration change promoting 1 can be committed on
the quorum `{2,3}` without the config change being appended to the
learner's log. If the leader (say 2) fails, there are de facto two
voters remaining. Only 3 can win an election (due to its log containing
all committed entries), but to do so it will need 1 to vote. But 1
considers itself a learner and will continue to do so until 3 has
stepped up as leader, replicates the conf change to 1, and 1 applies it.
Ultimately, by receiving a request to vote, the learner realizes that
the candidate believes it to be a voter, and that it should act
accordingly. The candidate's config may be stale, too; but in that case
it won't win the election, at least in the absence of the bug discussed
in:
https://github.com/etcd-io/etcd/issues/7625#issuecomment-488798263.