From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp33.i.mail.ru (smtp33.i.mail.ru [94.100.177.93]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dev.tarantool.org (Postfix) with ESMTPS id A81A0469719 for ; Tue, 17 Nov 2020 11:59:04 +0300 (MSK) References: From: Serge Petrenko Message-ID: <84b85640-b75a-c206-ca84-e4cae2bced64@tarantool.org> Date: Tue, 17 Nov 2020 11:59:03 +0300 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset="utf-8"; format="flowed" Content-Transfer-Encoding: 8bit Content-Language: en-GB Subject: Re: [Tarantool-patches] [PATCH 05/12] raft: stop using instance_id List-Id: Tarantool development patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Vladislav Shpilevoy , tarantool-patches@dev.tarantool.org, gorcunov@gmail.com 17.11.2020 03:02, Vladislav Shpilevoy пишет: > Raft is being moved to a separate library in src/lib. It means, > it can't depend on anything from box/. > > The patch makes raft stop using instance_id. > > Instead, it has a new option 'instance_id'. It is stored inside > struct raft as 'self', and should be configured using > raft_cfg_instance_id(). > > The configuration is done when bootstrap ends and the instance_id > is either recovered successfully, or the instance is anonymous. > > While working on this, I also considered introducing a new > function raft_boot() instead of raft_cfg_instance_id(). Which I > would also use to configure vclock later. Raft_boot() would be > meant to be called only one time with non-dynamic parameters > instance_id and vclock. > > But then I decided to keep adding new raft_cfg_*() functions. > Because: > > - It is more consistent with the existing options; > > - Does not require to think about too many different functions > like raft_create(), raft_boot(), raft_cfg_*() and in which order > to call them; > > Also I was thinking to introduce a single raft_cfg() like I did > in swim with swim_cfg(), to reduce number of raft_cfg_*() > functions, but decided it would be even worse with so many > options. > > Part of #5303 > --- LGTM. > src/box/box.cc | 10 +++++----- > src/box/raftlib.c | 32 ++++++++++++++++++++------------ > src/box/raftlib.h | 9 +++++++++ > 3 files changed, 34 insertions(+), 17 deletions(-) > > diff --git a/src/box/box.cc b/src/box/box.cc > index cc0d7b81d..8f5f3558e 100644 > --- a/src/box/box.cc > +++ b/src/box/box.cc > @@ -2763,12 +2763,12 @@ box_cfg_xc(void) > * Fill in leader election parameters after bootstrap. Before it is not > * possible - there may be relevant data to recover from WAL and > * snapshot. Also until recovery is done, it is not possible to write > - * new records into WAL. It is also totally safe, because relaying is > - * not started until the box is configured. So it can't happen, that > - * this election-enabled node will try to relay to another > - * election-enabled node without election actually enabled leading to > - * disconnect. > + * new records into WAL. Another reason - before recovery is done, > + * instance_id is not known, so Raft simply can't work. > */ > + if (!replication_anon) > + raft_cfg_instance_id(box_raft(), instance_id); > + > if (box_set_election_timeout() != 0) > diag_raise(); > /* > diff --git a/src/box/raftlib.c b/src/box/raftlib.c > index 0657fa85a..ca1940ba6 100644 > --- a/src/box/raftlib.c > +++ b/src/box/raftlib.c > @@ -296,7 +296,7 @@ raft_process_msg(struct raft *raft, const struct raft_request *req, > say_info("RAFT: message %s from %u", raft_request_to_string(req), > source); > assert(source > 0); > - assert(source != instance_id); > + assert(source != raft->self); > if (req->term == 0 || req->state == 0) { > diag_set(ClientError, ER_PROTOCOL, "Raft term and state can't " > "be zero"); > @@ -337,7 +337,7 @@ raft_process_msg(struct raft *raft, const struct raft_request *req, > raft->leader); > break; > } > - if (req->vote == instance_id) { > + if (req->vote == raft->self) { > /* > * This is entirely valid. This instance could > * request a vote, then become a follower or > @@ -373,7 +373,7 @@ raft_process_msg(struct raft *raft, const struct raft_request *req, > break; > case RAFT_STATE_CANDIDATE: > /* Check if this is a vote for a competing candidate. */ > - if (req->vote != instance_id) { > + if (req->vote != raft->self) { > say_info("RAFT: vote request is skipped - " > "competing candidate"); > break; > @@ -382,7 +382,7 @@ raft_process_msg(struct raft *raft, const struct raft_request *req, > * Vote for self was requested earlier in this round, > * and now was answered by some other instance. > */ > - assert(raft->volatile_vote == instance_id); > + assert(raft->volatile_vote == raft->self); > bool was_set = bit_set(&raft->vote_mask, source); > raft->vote_count += !was_set; > if (raft->vote_count < raft->election_quorum) { > @@ -547,7 +547,7 @@ end_dump: > } else if (raft->leader != 0) { > /* There is a known leader. Wait until it is dead. */ > raft_sm_wait_leader_dead(raft); > - } else if (raft->vote == instance_id) { > + } else if (raft->vote == raft->self) { > /* Just wrote own vote. */ > if (raft->election_quorum == 1) > raft_sm_become_leader(raft); > @@ -561,7 +561,7 @@ end_dump: > raft_sm_wait_election_end(raft); > } else { > /* No leaders, no votes. */ > - raft_sm_schedule_new_vote(raft, instance_id); > + raft_sm_schedule_new_vote(raft, raft->self); > } > } else { > memset(&req, 0, sizeof(req)); > @@ -596,7 +596,7 @@ raft_worker_handle_broadcast(struct raft *raft) > req.vote = raft->vote; > req.state = raft->state; > if (req.state == RAFT_STATE_CANDIDATE) { > - assert(raft->vote == instance_id); > + assert(raft->vote == raft->self); > req.vclock = &replicaset.vclock; > } > replicaset_foreach(replica) > @@ -652,7 +652,7 @@ raft_sm_become_leader(struct raft *raft) > assert(raft->is_candidate); > assert(!raft->is_write_in_progress); > raft->state = RAFT_STATE_LEADER; > - raft->leader = instance_id; > + raft->leader = raft->self; > ev_timer_stop(loop(), &raft->timer); > /* Make read-write (if other subsystems allow that. */ > box_update_ro_summary(); > @@ -682,14 +682,14 @@ raft_sm_become_candidate(struct raft *raft) > say_info("RAFT: enter candidate state with 1 self vote"); > assert(raft->state == RAFT_STATE_FOLLOWER); > assert(raft->leader == 0); > - assert(raft->vote == instance_id); > + assert(raft->vote == raft->self); > assert(raft->is_candidate); > assert(!raft->is_write_in_progress); > assert(raft->election_quorum > 1); > raft->state = RAFT_STATE_CANDIDATE; > raft->vote_count = 1; > raft->vote_mask = 0; > - bit_set(&raft->vote_mask, instance_id); > + bit_set(&raft->vote_mask, raft->self); > raft_sm_wait_election_end(raft); > /* State is visible and it is changed - broadcast. */ > raft_schedule_broadcast(raft); > @@ -736,7 +736,7 @@ raft_sm_schedule_new_election(struct raft *raft) > assert(raft->is_candidate); > /* Everyone is a follower until its vote for self is persisted. */ > raft_sm_schedule_new_term(raft, raft->term + 1); > - raft_sm_schedule_new_vote(raft, instance_id); > + raft_sm_schedule_new_vote(raft, raft->self); > box_update_ro_summary(); > } > > @@ -783,7 +783,7 @@ raft_sm_wait_election_end(struct raft *raft) > assert(raft->is_candidate); > assert(raft->state == RAFT_STATE_FOLLOWER || > (raft->state == RAFT_STATE_CANDIDATE && > - raft->volatile_vote == instance_id)); > + raft->volatile_vote == raft->self)); > assert(raft->leader == 0); > double election_timeout = raft->election_timeout + > raft_new_random_election_shift(raft); > @@ -979,6 +979,14 @@ raft_cfg_death_timeout(struct raft *raft, double death_timeout) > } > } > > +void > +raft_cfg_instance_id(struct raft *raft, uint32_t instance_id) > +{ > + assert(raft->self == 0); > + assert(instance_id != 0); > + raft->self = instance_id; > +} > + > void > raft_new_term(struct raft *raft) > { > diff --git a/src/box/raftlib.h b/src/box/raftlib.h > index c9c13136e..f75ed2567 100644 > --- a/src/box/raftlib.h > +++ b/src/box/raftlib.h > @@ -95,6 +95,8 @@ const char * > raft_state_str(uint32_t state); > > struct raft { > + /** Instance ID of this node. */ > + uint32_t self; > /** Instance ID of leader of the current term. */ > uint32_t leader; > /** State of the instance. */ > @@ -241,6 +243,13 @@ raft_cfg_election_quorum(struct raft *raft, int election_quorum); > void > raft_cfg_death_timeout(struct raft *raft, double death_timeout); > > +/** > + * Configure ID of the given Raft instance. The ID can't be changed after it is > + * assigned first time. > + */ > +void > +raft_cfg_instance_id(struct raft *raft, uint32_t instance_id); > + > /** > * Bump the term. When it is persisted, the node checks if there is a leader, > * and if there is not, a new election is started. That said, this function can -- Serge Petrenko