[Tarantool-patches] [PATCH 2/2] raft: fix crash on candidate cfg during WAL write

[Serge Petrenko]

07.11.2020 19:45, Vladislav Shpilevoy пишет:
> Raft state machine crashed if it was configured to be a candidate
> during a WAL write with a known leader.
>
> It tried to start waiting for the leader death, but should have
> waited for the WAL write end first.
>
> The code tried to handle it, but the order of 'if' conditions was
> wrong. WAL write being in progress was checked last, but should
> have been checked first.
>
> Closes #5506

Hi! Thanks  for the patch!

Please see some minor comments below.

> ---
>   src/box/raft.c                                | 10 +--
>   .../gh-5506-election-on-off.result            | 85 +++++++++++++++++++
>   .../gh-5506-election-on-off.test.lua          | 37 ++++++++
>   3 files changed, 127 insertions(+), 5 deletions(-)
>
> diff --git a/src/box/raft.c b/src/box/raft.c
> index 3a99a0f26..8f8b59ba6 100644
> --- a/src/box/raft.c
> +++ b/src/box/raft.c
> @@ -960,17 +960,17 @@ raft_cfg_is_candidate(bool is_candidate)
>   
>   	if (raft.is_candidate) {
>   		assert(raft.state == RAFT_STATE_FOLLOWER);
> -		if (raft.leader != 0) {
> -			raft_sm_wait_leader_dead();
> -		} else if (raft_is_fully_on_disk()) {
> -			raft_sm_wait_leader_found();
> -		} else {
> +		if (raft.is_write_in_progress) {
>   			/*
>   			 * If there is an on-going WAL write, it means there was
>   			 * some node who sent newer data to this node. So it is
>   			 * probably a better candidate. Anyway can't do anything
>   			 * until the new state is fully persisted.
>   			 */
> +		} else if (raft.leader != 0) {
> +			raft_sm_wait_leader_dead();
> +		} else {
> +			raft_sm_wait_leader_found();
>   		}
>   	} else {
>   		if (raft.state != RAFT_STATE_LEADER) {
> diff --git a/test/replication/gh-5506-election-on-off.result b/test/replication/gh-5506-election-on-off.result
> index 1a718396f..4fbc31986 100644
> --- a/test/replication/gh-5506-election-on-off.result
> +++ b/test/replication/gh-5506-election-on-off.result
> @@ -47,6 +47,91 @@ box.error.injection.set("ERRINJ_WAL_DELAY", false)
>    | - ok
>    | ...
>   
> +box.cfg{                                                                        \
> +    election_mode = old_election_mode,                                          \
> +}
> + | ---
> + | ...
> +
> +--
> +-- Another crash could happen when election mode was configured to be
> +-- 'candidate' with a known leader, but there was a not finished WAL write.
> +-- The node tried to start waiting for the leader death, even though with an
> +-- active WAL write it should wait for its end first.
> +--
> +box.schema.user.grant('guest', 'super')


box.schema.user.grant('guest', 'replication') is enough.


> + | ---
> + | ...
> +test_run:cmd('create server replica with rpl_master=default,\
> +              script="replication/replica.lua"')
> + | ---
> + | - true
> + | ...
> +test_run:cmd('start server replica with wait=True, wait_load=True')
> + | ---
> + | - true
> + | ...
> +
> +test_run:switch('replica')
> + | ---
> + | - true
> + | ...
> +box.cfg{election_mode = 'voter'}
> + | ---
> + | ...
> +box.error.injection.set("ERRINJ_WAL_DELAY_COUNTDOWN", 0)
> + | ---
> + | - ok
> + | ...
> +
> +test_run:switch('default')
> + | ---
> + | - true
> + | ...
> +box.cfg{election_mode = 'candidate'}
> + | ---
> + | ...
> +test_run:wait_cond(function()                                                   \
> +    return box.info.election.leader ~= 0                                        \
> +end)


Wouldn't it be simpler to wait for `box.info.election.state == 'leader'`?


> + | ---
> + | - true
> + | ...
> +
> +test_run:switch('replica')
> + | ---
> + | - true
> + | ...
> +test_run:wait_cond(function()                                                   \
> +    return box.error.injection.get("ERRINJ_WAL_DELAY")                          \
> +end)
> + | ---
> + | - true
> + | ...
> +box.cfg{election_mode = 'candidate'}
> + | ---
> + | ...
> +box.error.injection.set("ERRINJ_WAL_DELAY", false)
> + | ---
> + | - ok
> + | ...
> +
> +test_run:switch('default')
> + | ---
> + | - true
> + | ...
> +test_run:cmd('stop server replica')
> + | ---
> + | - true
> + | ...
> +test_run:cmd('delete server replica')
> + | ---
> + | - true
> + | ...
> +
> +box.schema.user.revoke('guest', 'super')
> + | ---
> + | ...
>   box.cfg{                                                                        \
>       election_mode = old_election_mode,                                          \
>       replication_timeout = old_replication_timeout,                              \
> diff --git a/test/replication/gh-5506-election-on-off.test.lua b/test/replication/gh-5506-election-on-off.test.lua
> index 290408f06..bb89477d1 100644
> --- a/test/replication/gh-5506-election-on-off.test.lua
> +++ b/test/replication/gh-5506-election-on-off.test.lua
> @@ -25,6 +25,43 @@ box.cfg{election_mode = 'off'}
>   box.cfg{election_mode = 'candidate'}
>   box.error.injection.set("ERRINJ_WAL_DELAY", false)
>   
> +box.cfg{                                                                        \
> +    election_mode = old_election_mode,                                          \
> +}
> +
> +--
> +-- Another crash could happen when election mode was configured to be
> +-- 'candidate' with a known leader, but there was a not finished WAL write.
> +-- The node tried to start waiting for the leader death, even though with an
> +-- active WAL write it should wait for its end first.
> +--
> +box.schema.user.grant('guest', 'super')
> +test_run:cmd('create server replica with rpl_master=default,\
> +              script="replication/replica.lua"')
> +test_run:cmd('start server replica with wait=True, wait_load=True')
> +
> +test_run:switch('replica')
> +box.cfg{election_mode = 'voter'}
> +box.error.injection.set("ERRINJ_WAL_DELAY_COUNTDOWN", 0)
> +
> +test_run:switch('default')
> +box.cfg{election_mode = 'candidate'}
> +test_run:wait_cond(function()                                                   \
> +    return box.info.election.leader ~= 0                                        \
> +end)
> +
> +test_run:switch('replica')
> +test_run:wait_cond(function()                                                   \
> +    return box.error.injection.get("ERRINJ_WAL_DELAY")                          \
> +end)
> +box.cfg{election_mode = 'candidate'}
> +box.error.injection.set("ERRINJ_WAL_DELAY", false)
> +
> +test_run:switch('default')
> +test_run:cmd('stop server replica')
> +test_run:cmd('delete server replica')
> +
> +box.schema.user.revoke('guest', 'super')
>   box.cfg{                                                                        \
>       election_mode = old_election_mode,                                          \
>       replication_timeout = old_replication_timeout,                              \

-- 
Serge Petrenko