path: root/src/rabbit_variable_queue.erl

%% The contents of this file are subject to the Mozilla Public License
%% Version 1.1 (the "License"); you may not use this file except in
%% compliance with the License. You may obtain a copy of the License
%% at http://www.mozilla.org/MPL/
%%
%% Software distributed under the License is distributed on an "AS IS"
%% basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
%% the License for the specific language governing rights and
%% limitations under the License.
%%
%% The Original Code is RabbitMQ.
%%
%% The Initial Developer of the Original Code is GoPivotal, Inc.
%% Copyright (c) 2007-2014 GoPivotal, Inc.  All rights reserved.
%%

-module(rabbit_variable_queue).

-export([init/3, terminate/2, delete_and_terminate/2, purge/1, purge_acks/1,
         publish/5, publish_delivered/4, discard/3, drain_confirmed/1,
         dropwhile/2, fetchwhile/4, fetch/2, drop/2, ack/2, requeue/2,
         ackfold/4, fold/3, len/1, is_empty/1, depth/1,
         set_ram_duration_target/2, ram_duration/1, needs_timeout/1, timeout/1,
         handle_pre_hibernate/1, resume/1, msg_rates/1,
         status/1, invoke/3, is_duplicate/2, multiple_routing_keys/0]).

-export([start/1, stop/0]).

%% exported for testing only
-export([start_msg_store/2, stop_msg_store/0, init/5]).

%%----------------------------------------------------------------------------
%% Definitions:

%% alpha: this is a message where both the message itself, and its
%%        position within the queue are held in RAM
%%
%% beta: this is a message where the message itself is only held on
%%        disk, but its position within the queue is held in RAM.
%%
%% gamma: this is a message where the message itself is only held on
%%        disk, but its position is both in RAM and on disk.
%%
%% delta: this is a collection of messages, represented by a single
%%        term, where the messages and their position are only held on
%%        disk.
%%
%% Note that for persistent messages, the message and its position
%% within the queue are always held on disk, *in addition* to being in
%% one of the above classifications.
%%
%% Also note that within this code, the term gamma seldom
%% appears. It's frequently the case that gammas are defined by betas
%% who have had their queue position recorded on disk.
%%
%% In general, messages move q1 -> q2 -> delta -> q3 -> q4, though
%% many of these steps are frequently skipped. q1 and q4 only hold
%% alphas, q2 and q3 hold both betas and gammas. When a message
%% arrives, its classification is determined. It is then added to the
%% rightmost appropriate queue.
%%
%% If a new message is determined to be a beta or gamma, q1 is
%% empty. If a new message is determined to be a delta, q1 and q2 are
%% empty (and actually q4 too).
%%
%% When removing messages from a queue, if q4 is empty then q3 is read
%% directly. If q3 becomes empty then the next segment's worth of
%% messages from delta are read into q3, reducing the size of
%% delta. If the queue is non empty, either q4 or q3 contain
%% entries. It is never permitted for delta to hold all the messages
%% in the queue.
%%
%% The duration indicated to us by the memory_monitor is used to
%% calculate, given our current ingress and egress rates, how many
%% messages we should hold in RAM (i.e. as alphas). We track the
%% ingress and egress rates for both messages and pending acks and
%% rates for both are considered when calculating the number of
%% messages to hold in RAM. When we need to push alphas to betas or
%% betas to gammas, we favour writing out messages that are further
%% from the head of the queue. This minimises writes to disk, as the
%% messages closer to the tail of the queue stay in the queue for
%% longer, thus do not need to be replaced as quickly by sending other
%% messages to disk.
%%
%% Whilst messages are pushed to disk and forgotten from RAM as soon
%% as requested by a new setting of the queue RAM duration, the
%% inverse is not true: we only load messages back into RAM as
%% demanded as the queue is read from. Thus only publishes to the
%% queue will take up available spare capacity.
%%
%% When we report our duration to the memory monitor, we calculate
%% average ingress and egress rates over the last two samples, and
%% then calculate our duration based on the sum of the ingress and
%% egress rates. More than two samples could be used, but it's a
%% balance between responding quickly enough to changes in
%% producers/consumers versus ignoring temporary blips. The problem
%% with temporary blips is that with just a few queues, they can have
%% substantial impact on the calculation of the average duration and
%% hence cause unnecessary I/O. Another alternative is to increase the
%% amqqueue_process:RAM_DURATION_UPDATE_PERIOD to beyond 5
%% seconds. However, that then runs the risk of being too slow to
%% inform the memory monitor of changes. Thus a 5 second interval,
%% plus a rolling average over the last two samples seems to work
%% well in practice.
%%
%% The sum of the ingress and egress rates is used because the egress
%% rate alone is not sufficient. Adding in the ingress rate means that
%% queues which are being flooded by messages are given more memory,
%% resulting in them being able to process the messages faster (by
%% doing less I/O, or at least deferring it) and thus helping keep
%% their mailboxes empty and thus the queue as a whole is more
%% responsive. If such a queue also has fast but previously idle
%% consumers, the consumer can then start to be driven as fast as it
%% can go, whereas if only egress rate was being used, the incoming
%% messages may have to be written to disk and then read back in,
%% resulting in the hard disk being a bottleneck in driving the
%% consumers. Generally, we want to give Rabbit every chance of
%% getting rid of messages as fast as possible and remaining
%% responsive, and using only the egress rate impacts that goal.
%%
%% Once the queue has more alphas than the target_ram_count, the
%% surplus must be converted to betas, if not gammas, if not rolled
%% into delta. The conditions under which these transitions occur
%% reflect the conflicting goals of minimising RAM cost per msg, and
%% minimising CPU cost per msg. Once the msg has become a beta, its
%% payload is no longer in RAM, thus a read from the msg_store must
%% occur before the msg can be delivered, but the RAM cost of a beta
%% is the same as a gamma, so converting a beta to gamma will not free
%% up any further RAM. To reduce the RAM cost further, the gamma must
%% be rolled into delta. Whilst recovering a beta or a gamma to an
%% alpha requires only one disk read (from the msg_store), recovering
%% a msg from within delta will require two reads (queue_index and
%% then msg_store). But delta has a near-0 per-msg RAM cost. So the
%% conflict is between using delta more, which will free up more
%% memory, but require additional CPU and disk ops, versus using delta
%% less and gammas and betas more, which will cost more memory, but
%% require fewer disk ops and less CPU overhead.
%%
%% In the case of a persistent msg published to a durable queue, the
%% msg is immediately written to the msg_store and queue_index. If
%% then additionally converted from an alpha, it'll immediately go to
%% a gamma (as it's already in queue_index), and cannot exist as a
%% beta. Thus a durable queue with a mixture of persistent and
%% transient msgs in it which has more messages than permitted by the
%% target_ram_count may contain an interspersed mixture of betas and
%% gammas in q2 and q3.
%%
%% There is then a ratio that controls how many betas and gammas there
%% can be. This is based on the target_ram_count and thus expresses
%% the fact that as the number of permitted alphas in the queue falls,
%% so should the number of betas and gammas fall (i.e. delta
%% grows). If q2 and q3 contain more than the permitted number of
%% betas and gammas, then the surplus are forcibly converted to gammas
%% (as necessary) and then rolled into delta. The ratio is that
%% delta/(betas+gammas+delta) equals
%% (betas+gammas+delta)/(target_ram_count+betas+gammas+delta). I.e. as
%% the target_ram_count shrinks to 0, so must betas and gammas.
%%
%% The conversion of betas to gammas is done in batches of at least
%% ?IO_BATCH_SIZE. This value should not be too small, otherwise the
%% frequent operations on the queues of q2 and q3 will not be
%% effectively amortised (switching the direction of queue access
%% defeats amortisation). Note that there is a natural upper bound due
%% to credit_flow limits on the alpha to beta conversion.
%%
%% The conversion from alphas to betas is chunked due to the
%% credit_flow limits of the msg_store. This further smooths the
%% effects of changes to the target_ram_count and ensures the queue
%% remains responsive even when there is a large amount of IO work to
%% do. The 'resume' callback is utilised to ensure that conversions
%% are done as promptly as possible whilst ensuring the queue remains
%% responsive.
%%
%% In the queue we keep track of both messages that are pending
%% delivery and messages that are pending acks. In the event of a
%% queue purge, we only need to load qi segments if the queue has
%% elements in deltas (i.e. it came under significant memory
%% pressure). In the event of a queue deletion, in addition to the
%% preceding, by keeping track of pending acks in RAM, we do not need
%% to search through qi segments looking for messages that are yet to
%% be acknowledged.
%%
%% Pending acks are recorded in memory by storing the message itself.
%% If the message has been sent to disk, we do not store the message
%% content. During memory reduction, pending acks containing message
%% content have that content removed and the corresponding messages
%% are pushed out to disk.
%%
%% Messages from pending acks are returned to q4, q3 and delta during
%% requeue, based on the limits of seq_id contained in each. Requeued
%% messages retain their original seq_id, maintaining order
%% when requeued.
%%
%% The order in which alphas are pushed to betas and pending acks
%% are pushed to disk is determined dynamically. We always prefer to
%% push messages for the source (alphas or acks) that is growing the
%% fastest (with growth measured as avg. ingress - avg. egress).
%%
%% Notes on Clean Shutdown
%% (This documents behaviour in variable_queue, queue_index and
%% msg_store.)
%%
%% In order to try to achieve as fast a start-up as possible, if a
%% clean shutdown occurs, we try to save out state to disk to reduce
%% work on startup. In the msg_store this takes the form of the
%% index_module's state, plus the file_summary ets table, and client
%% refs. In the VQ, this takes the form of the count of persistent
%% messages in the queue and references into the msg_stores. The
%% queue_index adds to these terms the details of its segments and
%% stores the terms in the queue directory.
%%
%% Two message stores are used. One is created for persistent messages
%% to durable queues that must survive restarts, and the other is used
%% for all other messages that just happen to need to be written to
%% disk. On start up we can therefore nuke the transient message
%% store, and be sure that the messages in the persistent store are
%% all that we need.
%%
%% The references to the msg_stores are there so that the msg_store
%% knows to only trust its saved state if all of the queues it was
%% previously talking to come up cleanly. Likewise, the queues
%% themselves (esp queue_index) skips work in init if all the queues
%% and msg_store were shutdown cleanly. This gives both good speed
%% improvements and also robustness so that if anything possibly went
%% wrong in shutdown (or there was subsequent manual tampering), all
%% messages and queues that can be recovered are recovered, safely.
%%
%% To delete transient messages lazily, the variable_queue, on
%% startup, stores the next_seq_id reported by the queue_index as the
%% transient_threshold. From that point on, whenever it's reading a
%% message off disk via the queue_index, if the seq_id is below this
%% threshold and the message is transient then it drops the message
%% (the message itself won't exist on disk because it would have been
%% stored in the transient msg_store which would have had its saved
%% state nuked on startup). This avoids the expensive operation of
%% scanning the entire queue on startup in order to delete transient
%% messages that were only pushed to disk to save memory.
%%
%%----------------------------------------------------------------------------

-behaviour(rabbit_backing_queue).

-record(vqstate,
        { q1,
          q2,
          delta,
          q3,
          q4,
          next_seq_id,
          ram_pending_ack,
          disk_pending_ack,
          index_state,
          msg_store_clients,
          durable,
          transient_threshold,

          len,
          persistent_count,

          target_ram_count,
          ram_msg_count,
          ram_msg_count_prev,
          ram_ack_count_prev,
          out_counter,
          in_counter,
          rates,
          msgs_on_disk,
          msg_indices_on_disk,
          unconfirmed,
          confirmed,
          ack_out_counter,
          ack_in_counter
        }).

-record(rates, { in, out, ack_in, ack_out, timestamp }).

-record(msg_status,
        { seq_id,
          msg_id,
          msg,
          is_persistent,
          is_delivered,
          msg_on_disk,
          index_on_disk,
          msg_props
        }).

-record(delta,
        { start_seq_id, %% start_seq_id is inclusive
          count,
          end_seq_id    %% end_seq_id is exclusive
        }).

%% When we discover that we should write some indices to disk for some
%% betas, the IO_BATCH_SIZE sets the number of betas that we must be
%% due to write indices for before we do any work at all.
-define(IO_BATCH_SIZE, 2048). %% next power-of-2 after ?CREDIT_DISC_BOUND
-define(PERSISTENT_MSG_STORE, msg_store_persistent).
-define(TRANSIENT_MSG_STORE,  msg_store_transient).
-define(QUEUE, lqueue).

-include("rabbit.hrl").

%%----------------------------------------------------------------------------

-rabbit_upgrade({multiple_routing_keys, local, []}).

-ifdef(use_specs).

-type(timestamp() :: {non_neg_integer(), non_neg_integer(), non_neg_integer()}).
-type(seq_id()  :: non_neg_integer()).

-type(rates() :: #rates { in        :: float(),
                          out       :: float(),
                          ack_in    :: float(),
                          ack_out   :: float(),
                          timestamp :: timestamp()}).

-type(delta() :: #delta { start_seq_id :: non_neg_integer(),
                          count        :: non_neg_integer(),
                          end_seq_id   :: non_neg_integer() }).

%% The compiler (rightfully) complains that ack() and state() are
%% unused. For this reason we duplicate a -spec from
%% rabbit_backing_queue with the only intent being to remove
%% warnings. The problem here is that we can't parameterise the BQ
%% behaviour by these two types as we would like to. We still leave
%% these here for documentation purposes.
-type(ack() :: seq_id()).
-type(state() :: #vqstate {
             q1                    :: ?QUEUE:?QUEUE(),
             q2                    :: ?QUEUE:?QUEUE(),
             delta                 :: delta(),
             q3                    :: ?QUEUE:?QUEUE(),
             q4                    :: ?QUEUE:?QUEUE(),
             next_seq_id           :: seq_id(),
             ram_pending_ack       :: gb_tree(),
             disk_pending_ack      :: gb_tree(),
             index_state           :: any(),
             msg_store_clients     :: 'undefined' | {{any(), binary()},
                                                    {any(), binary()}},
             durable               :: boolean(),
             transient_threshold   :: non_neg_integer(),

             len                   :: non_neg_integer(),
             persistent_count      :: non_neg_integer(),

             target_ram_count      :: non_neg_integer() | 'infinity',
             ram_msg_count         :: non_neg_integer(),
             ram_msg_count_prev    :: non_neg_integer(),
             out_counter           :: non_neg_integer(),
             in_counter            :: non_neg_integer(),
             rates                 :: rates(),
             msgs_on_disk          :: gb_set(),
             msg_indices_on_disk   :: gb_set(),
             unconfirmed           :: gb_set(),
             confirmed             :: gb_set(),
             ack_out_counter       :: non_neg_integer(),
             ack_in_counter        :: non_neg_integer() }).
%% Duplicated from rabbit_backing_queue
-spec(ack/2 :: ([ack()], state()) -> {[rabbit_guid:guid()], state()}).

-spec(multiple_routing_keys/0 :: () -> 'ok').

-endif.

-define(BLANK_DELTA, #delta { start_seq_id = undefined,
                              count        = 0,
                              end_seq_id   = undefined }).
-define(BLANK_DELTA_PATTERN(Z), #delta { start_seq_id = Z,
                                         count        = 0,
                                         end_seq_id   = Z }).

-define(MICROS_PER_SECOND, 1000000.0).

%% We're sampling every 5s for RAM duration; a half life that is of
%% the same order of magnitude is probably about right.
-define(RATE_AVG_HALF_LIFE, 5.0).

%% We will recalculate the #rates{} every time we get asked for our
%% RAM duration, or every N messages published, whichever is
%% sooner. We do this since the priority calculations in
%% rabbit_amqqueue_process need fairly fresh rates.
-define(MSGS_PER_RATE_CALC, 100).

%%----------------------------------------------------------------------------
%% Public API
%%----------------------------------------------------------------------------

start(DurableQueues) ->
    {AllTerms, StartFunState} = rabbit_queue_index:start(DurableQueues),
    start_msg_store(
      [Ref || Terms <- AllTerms,
              Terms /= non_clean_shutdown,
              begin
                  Ref = proplists:get_value(persistent_ref, Terms),
                  Ref =/= undefined
              end],
      StartFunState),
    {ok, AllTerms}.

stop() ->
    ok = stop_msg_store(),
    ok = rabbit_queue_index:stop().

start_msg_store(Refs, StartFunState) ->
    ok = rabbit_sup:start_child(?TRANSIENT_MSG_STORE, rabbit_msg_store,
                                [?TRANSIENT_MSG_STORE, rabbit_mnesia:dir(),
                                 undefined,  {fun (ok) -> finished end, ok}]),
    ok = rabbit_sup:start_child(?PERSISTENT_MSG_STORE, rabbit_msg_store,
                                [?PERSISTENT_MSG_STORE, rabbit_mnesia:dir(),
                                 Refs, StartFunState]).

stop_msg_store() ->
    ok = rabbit_sup:stop_child(?PERSISTENT_MSG_STORE),
    ok = rabbit_sup:stop_child(?TRANSIENT_MSG_STORE).

init(Queue, Recover, AsyncCallback) ->
    init(Queue, Recover, AsyncCallback,
         fun (MsgIds, ActionTaken) ->
                 msgs_written_to_disk(AsyncCallback, MsgIds, ActionTaken)
         end,
         fun (MsgIds) -> msg_indices_written_to_disk(AsyncCallback, MsgIds) end).

init(#amqqueue { name = QueueName, durable = IsDurable }, new,
     AsyncCallback, MsgOnDiskFun, MsgIdxOnDiskFun) ->
    IndexState = rabbit_queue_index:init(QueueName, MsgIdxOnDiskFun),
    init(IsDurable, IndexState, 0, [],
         case IsDurable of
             true  -> msg_store_client_init(?PERSISTENT_MSG_STORE,
                                            MsgOnDiskFun, AsyncCallback);
             false -> undefined
         end,
         msg_store_client_init(?TRANSIENT_MSG_STORE, undefined, AsyncCallback));

init(#amqqueue { name = QueueName, durable = true }, Terms,
     AsyncCallback, MsgOnDiskFun, MsgIdxOnDiskFun) ->
    {PRef, RecoveryTerms} = process_recovery_terms(Terms),
    PersistentClient = msg_store_client_init(?PERSISTENT_MSG_STORE, PRef,
                                             MsgOnDiskFun, AsyncCallback),
    TransientClient  = msg_store_client_init(?TRANSIENT_MSG_STORE,
                                             undefined, AsyncCallback),
    {DeltaCount, IndexState} =
        rabbit_queue_index:recover(
          QueueName, RecoveryTerms,
          rabbit_msg_store:successfully_recovered_state(?PERSISTENT_MSG_STORE),
          fun (MsgId) ->
                  rabbit_msg_store:contains(MsgId, PersistentClient)
          end,
          MsgIdxOnDiskFun),
    init(true, IndexState, DeltaCount, RecoveryTerms,
         PersistentClient, TransientClient).

process_recovery_terms(Terms=non_clean_shutdown) ->
    {rabbit_guid:gen(), Terms};
process_recovery_terms(Terms) ->
    case proplists:get_value(persistent_ref, Terms) of
        undefined -> {rabbit_guid:gen(), []};
        PRef      -> {PRef, Terms}
    end.

terminate(_Reason, State) ->
    State1 = #vqstate { persistent_count  = PCount,
                        index_state       = IndexState,
                        msg_store_clients = {MSCStateP, MSCStateT} } =
        purge_pending_ack(true, State),
    PRef = case MSCStateP of
               undefined -> undefined;
               _         -> ok = rabbit_msg_store:client_terminate(MSCStateP),
                            rabbit_msg_store:client_ref(MSCStateP)
           end,
    ok = rabbit_msg_store:client_delete_and_terminate(MSCStateT),
    Terms = [{persistent_ref, PRef}, {persistent_count, PCount}],
    a(State1 #vqstate { index_state       = rabbit_queue_index:terminate(
                                              Terms, IndexState),
                        msg_store_clients = undefined }).

%% the only difference between purge and delete is that delete also
%% needs to delete everything that's been delivered and not ack'd.
delete_and_terminate(_Reason, State) ->
    %% TODO: there is no need to interact with qi at all - which we do
    %% as part of 'purge' and 'purge_pending_ack', other than
    %% deleting it.
    {_PurgeCount, State1} = purge(State),
    State2 = #vqstate { index_state         = IndexState,
                        msg_store_clients   = {MSCStateP, MSCStateT} } =
        purge_pending_ack(false, State1),
    IndexState1 = rabbit_queue_index:delete_and_terminate(IndexState),
    case MSCStateP of
        undefined -> ok;
        _         -> rabbit_msg_store:client_delete_and_terminate(MSCStateP)
    end,
    rabbit_msg_store:client_delete_and_terminate(MSCStateT),
    a(State2 #vqstate { index_state       = IndexState1,
                        msg_store_clients = undefined }).

purge(State = #vqstate { q4                = Q4,
                         index_state       = IndexState,
                         msg_store_clients = MSCState,
                         len               = Len,
                         persistent_count  = PCount }) ->
    %% TODO: when there are no pending acks, which is a common case,
    %% we could simply wipe the qi instead of issuing delivers and
    %% acks for all the messages.
    {LensByStore, IndexState1} = remove_queue_entries(
                                   fun ?QUEUE:foldl/3, Q4,
                                   orddict:new(), IndexState, MSCState),
    {LensByStore1, State1 = #vqstate { q1                = Q1,
                                       index_state       = IndexState2,
                                       msg_store_clients = MSCState1 }} =
        purge_betas_and_deltas(LensByStore,
                               State #vqstate { q4          = ?QUEUE:new(),
                                                index_state = IndexState1 }),
    {LensByStore2, IndexState3} = remove_queue_entries(
                                    fun ?QUEUE:foldl/3, Q1,
                                    LensByStore1, IndexState2, MSCState1),
    PCount1 = PCount - find_persistent_count(LensByStore2),
    {Len, a(State1 #vqstate { q1                = ?QUEUE:new(),
                              index_state       = IndexState3,
                              len               = 0,
                              ram_msg_count     = 0,
                              persistent_count  = PCount1 })}.

purge_acks(State) -> a(purge_pending_ack(false, State)).

publish(Msg = #basic_message { is_persistent = IsPersistent, id = MsgId },
        MsgProps = #message_properties { needs_confirming = NeedsConfirming },
        IsDelivered, _ChPid, State = #vqstate { q1 = Q1, q3 = Q3, q4 = Q4,
                                                next_seq_id      = SeqId,
                                                len              = Len,
                                                in_counter       = InCount,
                                                persistent_count = PCount,
                                                durable          = IsDurable,
                                                unconfirmed      = UC }) ->
    IsPersistent1 = IsDurable andalso IsPersistent,
    MsgStatus = msg_status(IsPersistent1, IsDelivered, SeqId, Msg, MsgProps),
    {MsgStatus1, State1} = maybe_write_to_disk(false, false, MsgStatus, State),
    State2 = case ?QUEUE:is_empty(Q3) of
                 false -> State1 #vqstate { q1 = ?QUEUE:in(m(MsgStatus1), Q1) };
                 true  -> State1 #vqstate { q4 = ?QUEUE:in(m(MsgStatus1), Q4) }
             end,
    InCount1 = InCount + 1,
    PCount1  = PCount  + one_if(IsPersistent1),
    UC1 = gb_sets_maybe_insert(NeedsConfirming, MsgId, UC),
    State3 = inc_ram_msg_count(State2 #vqstate { next_seq_id      = SeqId + 1,
                                                 len              = Len   + 1,
                                                 in_counter       = InCount1,
                                                 persistent_count = PCount1,
                                                 unconfirmed      = UC1 }),
    a(reduce_memory_use(maybe_update_rates(State3))).

publish_delivered(Msg = #basic_message { is_persistent = IsPersistent,
                                         id = MsgId },
                  MsgProps = #message_properties {
                    needs_confirming = NeedsConfirming },
                  _ChPid, State = #vqstate { next_seq_id      = SeqId,
                                             out_counter      = OutCount,
                                             in_counter       = InCount,
                                             persistent_count = PCount,
                                             durable          = IsDurable,
                                             unconfirmed      = UC }) ->
    IsPersistent1 = IsDurable andalso IsPersistent,
    MsgStatus = msg_status(IsPersistent1, true, SeqId, Msg, MsgProps),
    {MsgStatus1, State1} = maybe_write_to_disk(false, false, MsgStatus, State),
    State2 = record_pending_ack(m(MsgStatus1), State1),
    PCount1 = PCount + one_if(IsPersistent1),
    UC1 = gb_sets_maybe_insert(NeedsConfirming, MsgId, UC),
    State3 = State2 #vqstate { next_seq_id      = SeqId    + 1,
                               out_counter      = OutCount + 1,
                               in_counter       = InCount  + 1,
                               persistent_count = PCount1,
                               unconfirmed      = UC1 },
    {SeqId, a(reduce_memory_use(maybe_update_rates(State3)))}.

discard(_MsgId, _ChPid, State) -> State.

drain_confirmed(State = #vqstate { confirmed = C }) ->
    case gb_sets:is_empty(C) of
        true  -> {[], State}; %% common case
        false -> {gb_sets:to_list(C), State #vqstate {
                                        confirmed = gb_sets:new() }}
    end.

dropwhile(Pred, State) ->
    case queue_out(State) of
        {empty, State1} ->
            {undefined, a(State1)};
        {{value, MsgStatus = #msg_status { msg_props = MsgProps }}, State1} ->
            case Pred(MsgProps) of
                true  -> {_, State2} = remove(false, MsgStatus, State1),
                         dropwhile(Pred, State2);
                false -> {MsgProps, a(in_r(MsgStatus, State1))}
            end
    end.

fetchwhile(Pred, Fun, Acc, State) ->
    case queue_out(State) of
        {empty, State1} ->
            {undefined, Acc, a(State1)};
        {{value, MsgStatus = #msg_status { msg_props = MsgProps }}, State1} ->
            case Pred(MsgProps) of
                true  -> {Msg, State2} = read_msg(MsgStatus, State1),
                         {AckTag, State3} = remove(true, MsgStatus, State2),
                         fetchwhile(Pred, Fun, Fun(Msg, AckTag, Acc), State3);
                false -> {MsgProps, Acc, a(in_r(MsgStatus, State1))}
            end
    end.

fetch(AckRequired, State) ->
    case queue_out(State) of
        {empty, State1} ->
            {empty, a(State1)};
        {{value, MsgStatus}, State1} ->
            %% it is possible that the message wasn't read from disk
            %% at this point, so read it in.
            {Msg, State2} = read_msg(MsgStatus, State1),
            {AckTag, State3} = remove(AckRequired, MsgStatus, State2),
            {{Msg, MsgStatus#msg_status.is_delivered, AckTag}, a(State3)}
    end.

drop(AckRequired, State) ->
    case queue_out(State) of
        {empty, State1} ->
            {empty, a(State1)};
        {{value, MsgStatus}, State1} ->
            {AckTag, State2} = remove(AckRequired, MsgStatus, State1),
            {{MsgStatus#msg_status.msg_id, AckTag}, a(State2)}
    end.

ack([], State) ->
    {[], State};
%% optimisation: this head is essentially a partial evaluation of the
%% general case below, for the single-ack case.
ack([SeqId], State) ->
    {#msg_status { msg_id        = MsgId,
                   is_persistent = IsPersistent,
                   msg_on_disk   = MsgOnDisk,
                   index_on_disk = IndexOnDisk },
     State1 = #vqstate { index_state       = IndexState,
                         msg_store_clients = MSCState,
                         persistent_count  = PCount,
                         ack_out_counter   = AckOutCount }} =
        remove_pending_ack(SeqId, State),
    IndexState1 = case IndexOnDisk of
                      true  -> rabbit_queue_index:ack([SeqId], IndexState);
                      false -> IndexState
                  end,
    case MsgOnDisk of
        true  -> ok = msg_store_remove(MSCState, IsPersistent, [MsgId]);
        false -> ok
    end,
    PCount1 = PCount - one_if(IsPersistent),
    {[MsgId],
     a(State1 #vqstate { index_state      = IndexState1,
                         persistent_count = PCount1,
                         ack_out_counter  = AckOutCount + 1 })};
ack(AckTags, State) ->
    {{IndexOnDiskSeqIds, MsgIdsByStore, AllMsgIds},
     State1 = #vqstate { index_state       = IndexState,
                         msg_store_clients = MSCState,
                         persistent_count  = PCount,
                         ack_out_counter   = AckOutCount }} =
        lists:foldl(
          fun (SeqId, {Acc, State2}) ->
                  {MsgStatus, State3} = remove_pending_ack(SeqId, State2),
                  {accumulate_ack(MsgStatus, Acc), State3}
          end, {accumulate_ack_init(), State}, AckTags),
    IndexState1 = rabbit_queue_index:ack(IndexOnDiskSeqIds, IndexState),
    [ok = msg_store_remove(MSCState, IsPersistent, MsgIds)
     || {IsPersistent, MsgIds} <- orddict:to_list(MsgIdsByStore)],
    PCount1 = PCount - find_persistent_count(sum_msg_ids_by_store_to_len(
                                               orddict:new(), MsgIdsByStore)),
    {lists:reverse(AllMsgIds),
     a(State1 #vqstate { index_state      = IndexState1,
                         persistent_count = PCount1,
                         ack_out_counter  = AckOutCount + length(AckTags) })}.

requeue(AckTags, #vqstate { delta      = Delta,
                            q3         = Q3,
                            q4         = Q4,
                            in_counter = InCounter,
                            len        = Len } = State) ->
    {SeqIds,  Q4a, MsgIds,  State1} = queue_merge(lists:sort(AckTags), Q4, [],
                                                  beta_limit(Q3),
                                                  fun publish_alpha/2, State),
    {SeqIds1, Q3a, MsgIds1, State2} = queue_merge(SeqIds, Q3, MsgIds,
                                                  delta_limit(Delta),
                                                  fun publish_beta/2, State1),
    {Delta1, MsgIds2, State3}       = delta_merge(SeqIds1, Delta, MsgIds1,
                                                  State2),
    MsgCount = length(MsgIds2),
    {MsgIds2, a(reduce_memory_use(
                  maybe_update_rates(
                    State3 #vqstate { delta      = Delta1,
                                      q3         = Q3a,
                                      q4         = Q4a,
                                      in_counter = InCounter + MsgCount,
                                      len        = Len + MsgCount })))}.

ackfold(MsgFun, Acc, State, AckTags) ->
    {AccN, StateN} =
        lists:foldl(fun(SeqId, {Acc0, State0}) ->
                            MsgStatus = lookup_pending_ack(SeqId, State0),
                            {Msg, State1} = read_msg(MsgStatus, State0),
                            {MsgFun(Msg, SeqId, Acc0), State1}
                    end, {Acc, State}, AckTags),
    {AccN, a(StateN)}.

fold(Fun, Acc, State = #vqstate{index_state = IndexState}) ->
    {Its, IndexState1} = lists:foldl(fun inext/2, {[], IndexState},
                                     [msg_iterator(State),
                                      disk_ack_iterator(State),
                                      ram_ack_iterator(State)]),
    ifold(Fun, Acc, Its, State#vqstate{index_state = IndexState1}).

len(#vqstate { len = Len }) -> Len.

is_empty(State) -> 0 == len(State).

depth(State = #vqstate { ram_pending_ack = RPA, disk_pending_ack = DPA }) ->
    len(State) + gb_trees:size(RPA) + gb_trees:size(DPA).

set_ram_duration_target(
  DurationTarget, State = #vqstate {
                    rates = #rates { in      = AvgIngressRate,
                                     out     = AvgEgressRate,
                                     ack_in  = AvgAckIngressRate,
                                     ack_out = AvgAckEgressRate },
                    target_ram_count = TargetRamCount }) ->
    Rate =
        AvgEgressRate + AvgIngressRate + AvgAckEgressRate + AvgAckIngressRate,
    TargetRamCount1 =
        case DurationTarget of
            infinity  -> infinity;
            _         -> trunc(DurationTarget * Rate) %% msgs = sec * msgs/sec
        end,
    State1 = State #vqstate { target_ram_count = TargetRamCount1 },
    a(case TargetRamCount1 == infinity orelse
          (TargetRamCount =/= infinity andalso
           TargetRamCount1 >= TargetRamCount) of
          true  -> State1;
          false -> reduce_memory_use(State1)
      end).

maybe_update_rates(State = #vqstate{ in_counter  = InCount,
                                     out_counter = OutCount })
  when InCount + OutCount > ?MSGS_PER_RATE_CALC ->
    update_rates(State);
maybe_update_rates(State) ->
    State.

update_rates(State = #vqstate{ in_counter      =     InCount,
                               out_counter     =    OutCount,
                               ack_in_counter  =  AckInCount,
                               ack_out_counter = AckOutCount,
                               rates = #rates{ in        =     InRate,
                                               out       =    OutRate,
                                               ack_in    =  AckInRate,
                                               ack_out   = AckOutRate,
                                               timestamp = TS }}) ->
    Now = erlang:now(),

    Rates = #rates { in        = update_rate(Now, TS,     InCount,     InRate),
                     out       = update_rate(Now, TS,    OutCount,    OutRate),
                     ack_in    = update_rate(Now, TS,  AckInCount,  AckInRate),
                     ack_out   = update_rate(Now, TS, AckOutCount, AckOutRate),
                     timestamp = Now },

    State#vqstate{ in_counter      = 0,
                   out_counter     = 0,
                   ack_in_counter  = 0,
                   ack_out_counter = 0,
                   rates           = Rates }.

update_rate(Now, TS, Count, Rate) ->
    Time = timer:now_diff(Now, TS) / ?MICROS_PER_SECOND,
    rabbit_misc:moving_average(Time, ?RATE_AVG_HALF_LIFE, Count / Time, Rate).

ram_duration(State) ->
    State1 = #vqstate { rates = #rates { in      = AvgIngressRate,
                                         out     = AvgEgressRate,
                                         ack_in  = AvgAckIngressRate,
                                         ack_out = AvgAckEgressRate },
                        ram_msg_count      = RamMsgCount,
                        ram_msg_count_prev = RamMsgCountPrev,
                        ram_pending_ack    = RPA,
                        ram_ack_count_prev = RamAckCountPrev } =
        update_rates(State),

    RamAckCount = gb_trees:size(RPA),

    Duration = %% msgs+acks / (msgs+acks/sec) == sec
        case lists:all(fun (X) -> X < 0.01 end,
                       [AvgEgressRate, AvgIngressRate,
                        AvgAckEgressRate, AvgAckIngressRate]) of
            true  -> infinity;
            false -> (RamMsgCountPrev + RamMsgCount +
                          RamAckCount + RamAckCountPrev) /
                         (4 * (AvgEgressRate + AvgIngressRate +
                                   AvgAckEgressRate + AvgAckIngressRate))
        end,

    {Duration, State1}.

needs_timeout(#vqstate { index_state = IndexState }) ->
    case rabbit_queue_index:needs_sync(IndexState) of
        confirms -> timed;
        other    -> idle;
        false    -> false
    end.

timeout(State = #vqstate { index_state = IndexState }) ->
    State #vqstate { index_state = rabbit_queue_index:sync(IndexState) }.

handle_pre_hibernate(State = #vqstate { index_state = IndexState }) ->
    State #vqstate { index_state = rabbit_queue_index:flush(IndexState) }.

resume(State) -> a(reduce_memory_use(State)).

msg_rates(#vqstate { rates = #rates { in  = AvgIngressRate,
                                      out = AvgEgressRate } }) ->
    {AvgIngressRate, AvgEgressRate}.

status(#vqstate {
          q1 = Q1, q2 = Q2, delta = Delta, q3 = Q3, q4 = Q4,
          len              = Len,
          ram_pending_ack  = RPA,
          disk_pending_ack = DPA,
          target_ram_count = TargetRamCount,
          ram_msg_count    = RamMsgCount,
          next_seq_id      = NextSeqId,
          persistent_count = PersistentCount,
          rates            = #rates { in      = AvgIngressRate,
                                      out     = AvgEgressRate,
                                      ack_in  = AvgAckIngressRate,
                                      ack_out = AvgAckEgressRate }}) ->

    [ {q1                  , ?QUEUE:len(Q1)},
      {q2                  , ?QUEUE:len(Q2)},
      {delta               , Delta},
      {q3                  , ?QUEUE:len(Q3)},
      {q4                  , ?QUEUE:len(Q4)},
      {len                 , Len},
      {pending_acks        , gb_trees:size(RPA) + gb_trees:size(DPA)},
      {target_ram_count    , TargetRamCount},
      {ram_msg_count       , RamMsgCount},
      {ram_ack_count       , gb_trees:size(RPA)},
      {next_seq_id         , NextSeqId},
      {persistent_count    , PersistentCount},
      {avg_ingress_rate    , AvgIngressRate},
      {avg_egress_rate     , AvgEgressRate},
      {avg_ack_ingress_rate, AvgAckIngressRate},
      {avg_ack_egress_rate , AvgAckEgressRate} ].

invoke(?MODULE, Fun, State) -> Fun(?MODULE, State);
invoke(      _,   _, State) -> State.

is_duplicate(_Msg, State) -> {false, State}.

%%----------------------------------------------------------------------------
%% Minor helpers
%%----------------------------------------------------------------------------

a(State = #vqstate { q1 = Q1, q2 = Q2, delta = Delta, q3 = Q3, q4 = Q4,
                     len              = Len,
                     persistent_count = PersistentCount,
                     ram_msg_count    = RamMsgCount }) ->
    E1 = ?QUEUE:is_empty(Q1),
    E2 = ?QUEUE:is_empty(Q2),
    ED = Delta#delta.count == 0,
    E3 = ?QUEUE:is_empty(Q3),
    E4 = ?QUEUE:is_empty(Q4),
    LZ = Len == 0,

    true = E1 or not E3,
    true = E2 or not ED,
    true = ED or not E3,
    true = LZ == (E3 and E4),

    true = Len             >= 0,
    true = PersistentCount >= 0,
    true = RamMsgCount     >= 0,
    true = RamMsgCount     =< Len,

    State.

d(Delta = #delta { start_seq_id = Start, count = Count, end_seq_id = End })
  when Start + Count =< End ->
    Delta.

m(MsgStatus = #msg_status { msg           = Msg,
                            is_persistent = IsPersistent,
                            msg_on_disk   = MsgOnDisk,
                            index_on_disk = IndexOnDisk }) ->
    true = (not IsPersistent) or IndexOnDisk,
    true = (not IndexOnDisk) or MsgOnDisk,
    true = (Msg =/= undefined) or MsgOnDisk,

    MsgStatus.

one_if(true ) -> 1;
one_if(false) -> 0.

cons_if(true,   E, L) -> [E | L];
cons_if(false, _E, L) -> L.

gb_sets_maybe_insert(false, _Val, Set) -> Set;
gb_sets_maybe_insert(true,   Val, Set) -> gb_sets:add(Val, Set).

msg_status(IsPersistent, IsDelivered, SeqId,
           Msg = #basic_message {id = MsgId}, MsgProps) ->
    #msg_status{seq_id        = SeqId,
                msg_id        = MsgId,
                msg           = Msg,
                is_persistent = IsPersistent,
                is_delivered  = IsDelivered,
                msg_on_disk   = false,
                index_on_disk = false,
                msg_props     = MsgProps}.

beta_msg_status({MsgId, SeqId, MsgProps, IsPersistent, IsDelivered}) ->
  #msg_status{seq_id        = SeqId,
              msg_id        = MsgId,
              msg           = undefined,
              is_persistent = IsPersistent,
              is_delivered  = IsDelivered,
              msg_on_disk   = true,
              index_on_disk = true,
              msg_props     = MsgProps}.

trim_msg_status(MsgStatus) -> MsgStatus #msg_status { msg = undefined }.

with_msg_store_state({MSCStateP, MSCStateT},  true, Fun) ->
    {Result, MSCStateP1} = Fun(MSCStateP),
    {Result, {MSCStateP1, MSCStateT}};
with_msg_store_state({MSCStateP, MSCStateT}, false, Fun) ->
    {Result, MSCStateT1} = Fun(MSCStateT),
    {Result, {MSCStateP, MSCStateT1}}.

with_immutable_msg_store_state(MSCState, IsPersistent, Fun) ->
    {Res, MSCState} = with_msg_store_state(MSCState, IsPersistent,
                                           fun (MSCState1) ->
                                                   {Fun(MSCState1), MSCState1}
                                           end),
    Res.

msg_store_client_init(MsgStore, MsgOnDiskFun, Callback) ->
    msg_store_client_init(MsgStore, rabbit_guid:gen(), MsgOnDiskFun,
                          Callback).

msg_store_client_init(MsgStore, Ref, MsgOnDiskFun, Callback) ->
    CloseFDsFun = msg_store_close_fds_fun(MsgStore =:= ?PERSISTENT_MSG_STORE),
    rabbit_msg_store:client_init(MsgStore, Ref, MsgOnDiskFun,
                                 fun () -> Callback(?MODULE, CloseFDsFun) end).

msg_store_write(MSCState, IsPersistent, MsgId, Msg) ->
    with_immutable_msg_store_state(
      MSCState, IsPersistent,
      fun (MSCState1) ->
              rabbit_msg_store:write_flow(MsgId, Msg, MSCState1)
      end).

msg_store_read(MSCState, IsPersistent, MsgId) ->
    with_msg_store_state(
      MSCState, IsPersistent,
      fun (MSCState1) ->
              rabbit_msg_store:read(MsgId, MSCState1)
      end).

msg_store_remove(MSCState, IsPersistent, MsgIds) ->
    with_immutable_msg_store_state(
      MSCState, IsPersistent,
      fun (MCSState1) ->
              rabbit_msg_store:remove(MsgIds, MCSState1)
      end).

msg_store_close_fds(MSCState, IsPersistent) ->
    with_msg_store_state(
      MSCState, IsPersistent,
      fun (MSCState1) -> rabbit_msg_store:close_all_indicated(MSCState1) end).

msg_store_close_fds_fun(IsPersistent) ->
    fun (?MODULE, State = #vqstate { msg_store_clients = MSCState }) ->
            {ok, MSCState1} = msg_store_close_fds(MSCState, IsPersistent),
            State #vqstate { msg_store_clients = MSCState1 }
    end.

maybe_write_delivered(false, _SeqId, IndexState) ->
    IndexState;
maybe_write_delivered(true, SeqId, IndexState) ->
    rabbit_queue_index:deliver([SeqId], IndexState).

betas_from_index_entries(List, TransientThreshold, RPA, DPA, IndexState) ->
    {Filtered, Delivers, Acks} =
        lists:foldr(
          fun ({_MsgId, SeqId, _MsgProps, IsPersistent, IsDelivered} = M,
               {Filtered1, Delivers1, Acks1} = Acc) ->
                  case SeqId < TransientThreshold andalso not IsPersistent of
                      true  -> {Filtered1,
                                cons_if(not IsDelivered, SeqId, Delivers1),
                                [SeqId | Acks1]};
                      false -> case (gb_trees:is_defined(SeqId, RPA) orelse
                                     gb_trees:is_defined(SeqId, DPA)) of
                                   false -> {?QUEUE:in_r(m(beta_msg_status(M)),
                                                         Filtered1),
                                             Delivers1, Acks1};
                                   true  -> Acc
                           end
                  end
          end, {?QUEUE:new(), [], []}, List),
    {Filtered, rabbit_queue_index:ack(
                 Acks, rabbit_queue_index:deliver(Delivers, IndexState))}.

expand_delta(SeqId, ?BLANK_DELTA_PATTERN(X)) ->
    d(#delta { start_seq_id = SeqId, count = 1, end_seq_id = SeqId + 1 });
expand_delta(SeqId, #delta { start_seq_id = StartSeqId,
                             count        = Count } = Delta)
  when SeqId < StartSeqId ->
    d(Delta #delta { start_seq_id = SeqId, count = Count + 1 });
expand_delta(SeqId, #delta { count        = Count,
                             end_seq_id   = EndSeqId } = Delta)
  when SeqId >= EndSeqId ->
    d(Delta #delta { count = Count + 1, end_seq_id = SeqId + 1 });
expand_delta(_SeqId, #delta { count       = Count } = Delta) ->
    d(Delta #delta { count = Count + 1 }).

%%----------------------------------------------------------------------------
%% Internal major helpers for Public API
%%----------------------------------------------------------------------------

init(IsDurable, IndexState, DeltaCount, Terms,
     PersistentClient, TransientClient) ->
    {LowSeqId, NextSeqId, IndexState1} = rabbit_queue_index:bounds(IndexState),

    DeltaCount1 =
        case Terms of
            non_clean_shutdown -> DeltaCount;
            _                  -> proplists:get_value(persistent_count,
                                                      Terms, DeltaCount)
        end,
    Delta = case DeltaCount1 == 0 andalso DeltaCount /= undefined of
                true  -> ?BLANK_DELTA;
                false -> d(#delta { start_seq_id = LowSeqId,
                                    count        = DeltaCount1,
                                    end_seq_id   = NextSeqId })
            end,
    Now = now(),
    State = #vqstate {
      q1                  = ?QUEUE:new(),
      q2                  = ?QUEUE:new(),
      delta               = Delta,
      q3                  = ?QUEUE:new(),
      q4                  = ?QUEUE:new(),
      next_seq_id         = NextSeqId,
      ram_pending_ack     = gb_trees:empty(),
      disk_pending_ack    = gb_trees:empty(),
      index_state         = IndexState1,
      msg_store_clients   = {PersistentClient, TransientClient},
      durable             = IsDurable,
      transient_threshold = NextSeqId,

      len                 = DeltaCount1,
      persistent_count    = DeltaCount1,

      target_ram_count    = infinity,
      ram_msg_count       = 0,
      ram_msg_count_prev  = 0,
      ram_ack_count_prev  = 0,
      out_counter         = 0,
      in_counter          = 0,
      rates               = blank_rates(Now),
      msgs_on_disk        = gb_sets:new(),
      msg_indices_on_disk = gb_sets:new(),
      unconfirmed         = gb_sets:new(),
      confirmed           = gb_sets:new(),
      ack_out_counter     = 0,
      ack_in_counter      = 0 },
    a(maybe_deltas_to_betas(State)).

blank_rates(Now) ->
    #rates { in        = 0.0,
             out       = 0.0,
             ack_in    = 0.0,
             ack_out   = 0.0,
             timestamp = Now}.

in_r(MsgStatus = #msg_status { msg = undefined },
     State = #vqstate { q3 = Q3, q4 = Q4 }) ->
    case ?QUEUE:is_empty(Q4) of
        true  -> State #vqstate { q3 = ?QUEUE:in_r(MsgStatus, Q3) };
        false -> {Msg, State1 = #vqstate { q4 = Q4a }} =
                     read_msg(MsgStatus, State),
                 inc_ram_msg_count(
                   State1 #vqstate { q4 = ?QUEUE:in_r(MsgStatus#msg_status {
                                                        msg = Msg }, Q4a) })
    end;
in_r(MsgStatus, State = #vqstate { q4 = Q4 }) ->
    State #vqstate { q4 = ?QUEUE:in_r(MsgStatus, Q4) }.

queue_out(State = #vqstate { q4 = Q4 }) ->
    case ?QUEUE:out(Q4) of
        {empty, _Q4} ->
            case fetch_from_q3(State) of
                {empty, _State1} = Result     -> Result;
                {loaded, {MsgStatus, State1}} -> {{value, MsgStatus}, State1}
            end;
        {{value, MsgStatus}, Q4a} ->
            {{value, MsgStatus}, State #vqstate { q4 = Q4a }}
    end.

read_msg(#msg_status{msg           = undefined,
                     msg_id        = MsgId,
                     is_persistent = IsPersistent}, State) ->
    read_msg(MsgId, IsPersistent, State);
read_msg(#msg_status{msg = Msg}, State) ->
    {Msg, State}.

read_msg(MsgId, IsPersistent, State = #vqstate{msg_store_clients = MSCState}) ->
    {{ok, Msg = #basic_message {}}, MSCState1} =
        msg_store_read(MSCState, IsPersistent, MsgId),
    {Msg, State #vqstate {msg_store_clients = MSCState1}}.

inc_ram_msg_count(State = #vqstate{ram_msg_count = RamMsgCount}) ->
    State#vqstate{ram_msg_count = RamMsgCount + 1}.

remove(AckRequired, MsgStatus = #msg_status {
                      seq_id        = SeqId,
                      msg_id        = MsgId,
                      msg           = Msg,
                      is_persistent = IsPersistent,
                      is_delivered  = IsDelivered,
                      msg_on_disk   = MsgOnDisk,
                      index_on_disk = IndexOnDisk },
       State = #vqstate {ram_msg_count     = RamMsgCount,
                         out_counter       = OutCount,
                         index_state       = IndexState,
                         msg_store_clients = MSCState,
                         len               = Len,
                         persistent_count  = PCount}) ->
    %% 1. Mark it delivered if necessary
    IndexState1 = maybe_write_delivered(
                    IndexOnDisk andalso not IsDelivered,
                    SeqId, IndexState),

    %% 2. Remove from msg_store and queue index, if necessary
    Rem = fun () ->
                  ok = msg_store_remove(MSCState, IsPersistent, [MsgId])
          end,
    Ack = fun () -> rabbit_queue_index:ack([SeqId], IndexState1) end,
    IndexState2 = case {AckRequired, MsgOnDisk, IndexOnDisk} of
                      {false, true, false} -> Rem(), IndexState1;
                      {false, true,  true} -> Rem(), Ack();
                      _                    -> IndexState1
                  end,

    %% 3. If an ack is required, add something sensible to PA
    {AckTag, State1} = case AckRequired of
                           true  -> StateN = record_pending_ack(
                                               MsgStatus #msg_status {
                                                 is_delivered = true }, State),
                                    {SeqId, StateN};
                           false -> {undefined, State}
                       end,

    PCount1      = PCount      - one_if(IsPersistent andalso not AckRequired),
    RamMsgCount1 = RamMsgCount - one_if(Msg =/= undefined),

    {AckTag, maybe_update_rates(
               State1 #vqstate {ram_msg_count    = RamMsgCount1,
                                out_counter      = OutCount + 1,
                                index_state      = IndexState2,
                                len              = Len - 1,
                                persistent_count = PCount1})}.

purge_betas_and_deltas(LensByStore,
                       State = #vqstate { q3                = Q3,
                                          index_state       = IndexState,
                                          msg_store_clients = MSCState }) ->
    case ?QUEUE:is_empty(Q3) of
        true  -> {LensByStore, State};
        false -> {LensByStore1, IndexState1} =
                     remove_queue_entries(fun ?QUEUE:foldl/3, Q3,
                                          LensByStore, IndexState, MSCState),
                 purge_betas_and_deltas(LensByStore1,
                                        maybe_deltas_to_betas(
                                          State #vqstate {
                                            q3          = ?QUEUE:new(),
                                            index_state = IndexState1 }))
    end.

remove_queue_entries(Fold, Q, LensByStore, IndexState, MSCState) ->
    {MsgIdsByStore, Delivers, Acks} =
        Fold(fun remove_queue_entries1/2, {orddict:new(), [], []}, Q),
    ok = orddict:fold(fun (IsPersistent, MsgIds, ok) ->
                              msg_store_remove(MSCState, IsPersistent, MsgIds)
                      end, ok, MsgIdsByStore),
    {sum_msg_ids_by_store_to_len(LensByStore, MsgIdsByStore),
     rabbit_queue_index:ack(Acks,
                            rabbit_queue_index:deliver(Delivers, IndexState))}.

remove_queue_entries1(
  #msg_status { msg_id = MsgId, seq_id = SeqId,
                is_delivered = IsDelivered, msg_on_disk = MsgOnDisk,
                index_on_disk = IndexOnDisk, is_persistent = IsPersistent },
  {MsgIdsByStore, Delivers, Acks}) ->
    {case MsgOnDisk of
         true  -> rabbit_misc:orddict_cons(IsPersistent, MsgId, MsgIdsByStore);
         false -> MsgIdsByStore
     end,
     cons_if(IndexOnDisk andalso not IsDelivered, SeqId, Delivers),
     cons_if(IndexOnDisk, SeqId, Acks)}.

sum_msg_ids_by_store_to_len(LensByStore, MsgIdsByStore) ->
    orddict:fold(
      fun (IsPersistent, MsgIds, LensByStore1) ->
              orddict:update_counter(IsPersistent, length(MsgIds), LensByStore1)
      end, LensByStore, MsgIdsByStore).

%%----------------------------------------------------------------------------
%% Internal gubbins for publishing
%%----------------------------------------------------------------------------

maybe_write_msg_to_disk(_Force, MsgStatus = #msg_status {
                                  msg_on_disk = true }, _MSCState) ->
    MsgStatus;
maybe_write_msg_to_disk(Force, MsgStatus = #msg_status {
                                 msg = Msg, msg_id = MsgId,
                                 is_persistent = IsPersistent }, MSCState)
  when Force orelse IsPersistent ->
    Msg1 = Msg #basic_message {
             %% don't persist any recoverable decoded properties
             content = rabbit_binary_parser:clear_decoded_content(
                         Msg #basic_message.content)},
    ok = msg_store_write(MSCState, IsPersistent, MsgId, Msg1),
    MsgStatus #msg_status { msg_on_disk = true };
maybe_write_msg_to_disk(_Force, MsgStatus, _MSCState) ->
    MsgStatus.

maybe_write_index_to_disk(_Force, MsgStatus = #msg_status {
                                    index_on_disk = true }, IndexState) ->
    true = MsgStatus #msg_status.msg_on_disk, %% ASSERTION
    {MsgStatus, IndexState};
maybe_write_index_to_disk(Force, MsgStatus = #msg_status {
                                   msg_id        = MsgId,
                                   seq_id        = SeqId,
                                   is_persistent = IsPersistent,
                                   is_delivered  = IsDelivered,
                                   msg_props     = MsgProps}, IndexState)
  when Force orelse IsPersistent ->
    true = MsgStatus #msg_status.msg_on_disk, %% ASSERTION
    IndexState1 = rabbit_queue_index:publish(
                    MsgId, SeqId, MsgProps, IsPersistent, IndexState),
    {MsgStatus #msg_status { index_on_disk = true },
     maybe_write_delivered(IsDelivered, SeqId, IndexState1)};
maybe_write_index_to_disk(_Force, MsgStatus, IndexState) ->
    {MsgStatus, IndexState}.

maybe_write_to_disk(ForceMsg, ForceIndex, MsgStatus,
                    State = #vqstate { index_state       = IndexState,
                                       msg_store_clients = MSCState }) ->
    MsgStatus1 = maybe_write_msg_to_disk(ForceMsg, MsgStatus, MSCState),
    {MsgStatus2, IndexState1} =
        maybe_write_index_to_disk(ForceIndex, MsgStatus1, IndexState),
    {MsgStatus2, State #vqstate { index_state = IndexState1 }}.

%%----------------------------------------------------------------------------
%% Internal gubbins for acks
%%----------------------------------------------------------------------------

record_pending_ack(#msg_status { seq_id = SeqId, msg = Msg } = MsgStatus,
                   State = #vqstate { ram_pending_ack  = RPA,
                                      disk_pending_ack = DPA,
                                      ack_in_counter   = AckInCount}) ->
    {RPA1, DPA1} =
        case Msg of
            undefined -> {RPA, gb_trees:insert(SeqId, MsgStatus, DPA)};
            _         -> {gb_trees:insert(SeqId, MsgStatus, RPA), DPA}
        end,
    State #vqstate { ram_pending_ack  = RPA1,
                     disk_pending_ack = DPA1,
                     ack_in_counter   = AckInCount + 1}.

lookup_pending_ack(SeqId, #vqstate { ram_pending_ack  = RPA,
                                     disk_pending_ack = DPA }) ->
    case gb_trees:lookup(SeqId, RPA) of
        {value, V} -> V;
        none       -> gb_trees:get(SeqId, DPA)
    end.

remove_pending_ack(SeqId, State = #vqstate { ram_pending_ack  = RPA,
                                             disk_pending_ack = DPA }) ->
    case gb_trees:lookup(SeqId, RPA) of
        {value, V} -> RPA1 = gb_trees:delete(SeqId, RPA),
                      {V, State #vqstate { ram_pending_ack = RPA1 }};
        none       -> DPA1 = gb_trees:delete(SeqId, DPA),
                      {gb_trees:get(SeqId, DPA),
                       State #vqstate { disk_pending_ack = DPA1 }}
    end.

purge_pending_ack(KeepPersistent,
                  State = #vqstate { ram_pending_ack   = RPA,
                                     disk_pending_ack  = DPA,
                                     index_state       = IndexState,
                                     msg_store_clients = MSCState }) ->
    F = fun (_SeqId, MsgStatus, Acc) -> accumulate_ack(MsgStatus, Acc) end,
    {IndexOnDiskSeqIds, MsgIdsByStore, _AllMsgIds} =
        rabbit_misc:gb_trees_fold(
          F, rabbit_misc:gb_trees_fold(F, accumulate_ack_init(), RPA), DPA),
    State1 = State #vqstate { ram_pending_ack  = gb_trees:empty(),
                              disk_pending_ack = gb_trees:empty() },

    case KeepPersistent of
        true  -> case orddict:find(false, MsgIdsByStore) of
                     error        -> State1;
                     {ok, MsgIds} -> ok = msg_store_remove(MSCState, false,
                                                           MsgIds),
                                    State1
                 end;
        false -> IndexState1 =
                     rabbit_queue_index:ack(IndexOnDiskSeqIds, IndexState),
                 [ok = msg_store_remove(MSCState, IsPersistent, MsgIds)
                  || {IsPersistent, MsgIds} <- orddict:to_list(MsgIdsByStore)],
                 State1 #vqstate { index_state = IndexState1 }
    end.

accumulate_ack_init() -> {[], orddict:new(), []}.

accumulate_ack(#msg_status { seq_id        = SeqId,
                             msg_id        = MsgId,
                             is_persistent = IsPersistent,
                             msg_on_disk   = MsgOnDisk,
                             index_on_disk = IndexOnDisk },
               {IndexOnDiskSeqIdsAcc, MsgIdsByStore, AllMsgIds}) ->
    {cons_if(IndexOnDisk, SeqId, IndexOnDiskSeqIdsAcc),
     case MsgOnDisk of
         true  -> rabbit_misc:orddict_cons(IsPersistent, MsgId, MsgIdsByStore);
         false -> MsgIdsByStore
     end,
     [MsgId | AllMsgIds]}.

find_persistent_count(LensByStore) ->
    case orddict:find(true, LensByStore) of
        error     -> 0;
        {ok, Len} -> Len
    end.

%%----------------------------------------------------------------------------
%% Internal plumbing for confirms (aka publisher acks)
%%----------------------------------------------------------------------------

record_confirms(MsgIdSet, State = #vqstate { msgs_on_disk        = MOD,
                                             msg_indices_on_disk = MIOD,
                                             unconfirmed         = UC,
                                             confirmed           = C }) ->
    State #vqstate {
      msgs_on_disk        = rabbit_misc:gb_sets_difference(MOD,  MsgIdSet),
      msg_indices_on_disk = rabbit_misc:gb_sets_difference(MIOD, MsgIdSet),
      unconfirmed         = rabbit_misc:gb_sets_difference(UC,   MsgIdSet),
      confirmed           = gb_sets:union(C, MsgIdSet) }.

msgs_written_to_disk(Callback, MsgIdSet, ignored) ->
    Callback(?MODULE,
             fun (?MODULE, State) -> record_confirms(MsgIdSet, State) end);
msgs_written_to_disk(Callback, MsgIdSet, written) ->
    Callback(?MODULE,
             fun (?MODULE, State = #vqstate { msgs_on_disk        = MOD,
                                              msg_indices_on_disk = MIOD,
                                              unconfirmed         = UC }) ->
                     Confirmed = gb_sets:intersection(UC, MsgIdSet),
                     record_confirms(gb_sets:intersection(MsgIdSet, MIOD),
                                     State #vqstate {
                                       msgs_on_disk =
                                           gb_sets:union(MOD, Confirmed) })
             end).

msg_indices_written_to_disk(Callback, MsgIdSet) ->
    Callback(?MODULE,
             fun (?MODULE, State = #vqstate { msgs_on_disk        = MOD,
                                              msg_indices_on_disk = MIOD,
                                              unconfirmed         = UC }) ->
                     Confirmed = gb_sets:intersection(UC, MsgIdSet),
                     record_confirms(gb_sets:intersection(MsgIdSet, MOD),
                                     State #vqstate {
                                       msg_indices_on_disk =
                                           gb_sets:union(MIOD, Confirmed) })
             end).

%%----------------------------------------------------------------------------
%% Internal plumbing for requeue
%%----------------------------------------------------------------------------

publish_alpha(#msg_status { msg = undefined } = MsgStatus, State) ->
    {Msg, State1} = read_msg(MsgStatus, State),
    {MsgStatus#msg_status { msg = Msg }, inc_ram_msg_count(State1)};
publish_alpha(MsgStatus, State) ->
    {MsgStatus, inc_ram_msg_count(State)}.

publish_beta(MsgStatus, State) ->
    {MsgStatus1, State1} = maybe_write_to_disk(true, false, MsgStatus, State),
    {m(trim_msg_status(MsgStatus1)), State1}.

%% Rebuild queue, inserting sequence ids to maintain ordering
queue_merge(SeqIds, Q, MsgIds, Limit, PubFun, State) ->
    queue_merge(SeqIds, Q, ?QUEUE:new(), MsgIds,
                Limit, PubFun, State).

queue_merge([SeqId | Rest] = SeqIds, Q, Front, MsgIds,
            Limit, PubFun, State)
  when Limit == undefined orelse SeqId < Limit ->
    case ?QUEUE:out(Q) of
        {{value, #msg_status { seq_id = SeqIdQ } = MsgStatus}, Q1}
          when SeqIdQ < SeqId ->
            %% enqueue from the remaining queue
            queue_merge(SeqIds, Q1, ?QUEUE:in(MsgStatus, Front), MsgIds,
                        Limit, PubFun, State);
        {_, _Q1} ->
            %% enqueue from the remaining list of sequence ids
            {MsgStatus, State1} = msg_from_pending_ack(SeqId, State),
            {#msg_status { msg_id = MsgId } = MsgStatus1, State2} =
                PubFun(MsgStatus, State1),
            queue_merge(Rest, Q, ?QUEUE:in(MsgStatus1, Front), [MsgId | MsgIds],
                        Limit, PubFun, State2)
    end;
queue_merge(SeqIds, Q, Front, MsgIds,
            _Limit, _PubFun, State) ->
    {SeqIds, ?QUEUE:join(Front, Q), MsgIds, State}.

delta_merge([], Delta, MsgIds, State) ->
    {Delta, MsgIds, State};
delta_merge(SeqIds, Delta, MsgIds, State) ->
    lists:foldl(fun (SeqId, {Delta0, MsgIds0, State0}) ->
                        {#msg_status { msg_id = MsgId } = MsgStatus, State1} =
                            msg_from_pending_ack(SeqId, State0),
                        {_MsgStatus, State2} =
                            maybe_write_to_disk(true, true, MsgStatus, State1),
                        {expand_delta(SeqId, Delta0), [MsgId | MsgIds0], State2}
                end, {Delta, MsgIds, State}, SeqIds).

%% Mostly opposite of record_pending_ack/2
msg_from_pending_ack(SeqId, State) ->
    {#msg_status { msg_props = MsgProps } = MsgStatus, State1} =
        remove_pending_ack(SeqId, State),
    {MsgStatus #msg_status {
       msg_props = MsgProps #message_properties { needs_confirming = false } },
     State1}.

beta_limit(Q) ->
    case ?QUEUE:peek(Q) of
        {value, #msg_status { seq_id = SeqId }} -> SeqId;
        empty                                   -> undefined
    end.

delta_limit(?BLANK_DELTA_PATTERN(_X))             -> undefined;
delta_limit(#delta { start_seq_id = StartSeqId }) -> StartSeqId.

%%----------------------------------------------------------------------------
%% Iterator
%%----------------------------------------------------------------------------

ram_ack_iterator(State) ->
    {ack, gb_trees:iterator(State#vqstate.ram_pending_ack)}.

disk_ack_iterator(State) ->
    {ack, gb_trees:iterator(State#vqstate.disk_pending_ack)}.

msg_iterator(State) -> istate(start, State).

istate(start, State) -> {q4,    State#vqstate.q4,    State};
istate(q4,    State) -> {q3,    State#vqstate.q3,    State};
istate(q3,    State) -> {delta, State#vqstate.delta, State};
istate(delta, State) -> {q2,    State#vqstate.q2,    State};
istate(q2,    State) -> {q1,    State#vqstate.q1,    State};
istate(q1,   _State) -> done.

next({ack, It}, IndexState) ->
    case gb_trees:next(It) of
        none                     -> {empty, IndexState};
        {_SeqId, MsgStatus, It1} -> Next = {ack, It1},
                                    {value, MsgStatus, true, Next, IndexState}
    end;
next(done, IndexState) -> {empty, IndexState};
next({delta, #delta{start_seq_id = SeqId,
                    end_seq_id   = SeqId}, State}, IndexState) ->
    next(istate(delta, State), IndexState);
next({delta, #delta{start_seq_id = SeqId,
                    end_seq_id   = SeqIdEnd} = Delta, State}, IndexState) ->
    SeqIdB = rabbit_queue_index:next_segment_boundary(SeqId),
    SeqId1 = lists:min([SeqIdB, SeqIdEnd]),
    {List, IndexState1} = rabbit_queue_index:read(SeqId, SeqId1, IndexState),
    next({delta, Delta#delta{start_seq_id = SeqId1}, List, State}, IndexState1);
next({delta, Delta, [], State}, IndexState) ->
    next({delta, Delta, State}, IndexState);
next({delta, Delta, [{_, SeqId, _, _, _} = M | Rest], State}, IndexState) ->
    case (gb_trees:is_defined(SeqId, State#vqstate.ram_pending_ack) orelse
          gb_trees:is_defined(SeqId, State#vqstate.disk_pending_ack)) of
        false -> Next = {delta, Delta, Rest, State},
                 {value, beta_msg_status(M), false, Next, IndexState};
        true  -> next({delta, Delta, Rest, State}, IndexState)
    end;
next({Key, Q, State}, IndexState) ->
    case ?QUEUE:out(Q) of
        {empty, _Q}              -> next(istate(Key, State), IndexState);
        {{value, MsgStatus}, QN} -> Next = {Key, QN, State},
                                    {value, MsgStatus, false, Next, IndexState}
    end.

inext(It, {Its, IndexState}) ->
    case next(It, IndexState) of
        {empty, IndexState1} ->
            {Its, IndexState1};
        {value, MsgStatus1, Unacked, It1, IndexState1} ->
            {[{MsgStatus1, Unacked, It1} | Its], IndexState1}
    end.

ifold(_Fun, Acc, [], State) ->
    {Acc, State};
ifold(Fun, Acc, Its, State) ->
    [{MsgStatus, Unacked, It} | Rest] =
        lists:sort(fun ({#msg_status{seq_id = SeqId1}, _, _},
                        {#msg_status{seq_id = SeqId2}, _, _}) ->
                           SeqId1 =< SeqId2
                   end, Its),
    {Msg, State1} = read_msg(MsgStatus, State),
    case Fun(Msg, MsgStatus#msg_status.msg_props, Unacked, Acc) of
        {stop, Acc1} ->
            {Acc1, State};
        {cont, Acc1} ->
            {Its1, IndexState1} = inext(It, {Rest, State1#vqstate.index_state}),
            ifold(Fun, Acc1, Its1, State1#vqstate{index_state = IndexState1})
    end.

%%----------------------------------------------------------------------------
%% Phase changes
%%----------------------------------------------------------------------------

reduce_memory_use(State = #vqstate { target_ram_count = infinity }) ->
    State;
reduce_memory_use(State = #vqstate {
                    ram_pending_ack  = RPA,
                    ram_msg_count    = RamMsgCount,
                    target_ram_count = TargetRamCount,
                    rates            = #rates { in      = AvgIngress,
                                                out     = AvgEgress,
                                                ack_in  = AvgAckIngress,
                                                ack_out = AvgAckEgress } }) ->

    State1 = #vqstate { q2 = Q2, q3 = Q3 } =
        case chunk_size(RamMsgCount + gb_trees:size(RPA), TargetRamCount) of
            0  -> State;
            %% Reduce memory of pending acks and alphas. The order is
            %% determined based on which is growing faster. Whichever
            %% comes second may very well get a quota of 0 if the
            %% first manages to push out the max number of messages.
            S1 -> Funs = case ((AvgAckIngress - AvgAckEgress) >
                                   (AvgIngress - AvgEgress)) of
                             true  -> [fun limit_ram_acks/2,
                                       fun push_alphas_to_betas/2];
                             false -> [fun push_alphas_to_betas/2,
                                       fun limit_ram_acks/2]
                         end,
                  {_, State2} = lists:foldl(fun (ReduceFun, {QuotaN, StateN}) ->
                                                    ReduceFun(QuotaN, StateN)
                                            end, {S1, State}, Funs),
                  State2
        end,

    case chunk_size(?QUEUE:len(Q2) + ?QUEUE:len(Q3),
                    permitted_beta_count(State1)) of
        S2 when S2 >= ?IO_BATCH_SIZE ->
            %% There is an implicit, but subtle, upper bound here. We
            %% may shuffle a lot of messages from Q2/3 into delta, but
            %% the number of these that require any disk operation,
            %% namely index writing, i.e. messages that are genuine
            %% betas and not gammas, is bounded by the credit_flow
            %% limiting of the alpha->beta conversion above.
            push_betas_to_deltas(S2, State1);
        _  ->
            State1
    end.

limit_ram_acks(0, State) ->
    {0, State};
limit_ram_acks(Quota, State = #vqstate { ram_pending_ack  = RPA,
                                         disk_pending_ack = DPA }) ->
    case gb_trees:is_empty(RPA) of
        true ->
            {Quota, State};
        false ->
            {SeqId, MsgStatus, RPA1} = gb_trees:take_largest(RPA),
            {MsgStatus1, State1} =
                maybe_write_to_disk(true, false, MsgStatus, State),
            DPA1 = gb_trees:insert(SeqId, m(trim_msg_status(MsgStatus1)), DPA),
            limit_ram_acks(Quota - 1,
                           State1 #vqstate { ram_pending_ack  = RPA1,
                                             disk_pending_ack = DPA1 })
    end.

permitted_beta_count(#vqstate { len = 0 }) ->
    infinity;
permitted_beta_count(#vqstate { target_ram_count = 0, q3 = Q3 }) ->
    lists:min([?QUEUE:len(Q3), rabbit_queue_index:next_segment_boundary(0)]);
permitted_beta_count(#vqstate { q1               = Q1,
                                q4               = Q4,
                                target_ram_count = TargetRamCount,
                                len              = Len }) ->
    BetaDelta = Len - ?QUEUE:len(Q1) - ?QUEUE:len(Q4),
    lists:max([rabbit_queue_index:next_segment_boundary(0),
               BetaDelta - ((BetaDelta * BetaDelta) div
                                (BetaDelta + TargetRamCount))]).

chunk_size(Current, Permitted)
  when Permitted =:= infinity orelse Permitted >= Current ->
    0;
chunk_size(Current, Permitted) ->
    Current - Permitted.

fetch_from_q3(State = #vqstate { q1    = Q1,
                                 q2    = Q2,
                                 delta = #delta { count = DeltaCount },
                                 q3    = Q3,
                                 q4    = Q4 }) ->
    case ?QUEUE:out(Q3) of
        {empty, _Q3} ->
            {empty, State};
        {{value, MsgStatus}, Q3a} ->
            State1 = State #vqstate { q3 = Q3a },
            State2 = case {?QUEUE:is_empty(Q3a), 0 == DeltaCount} of
                         {true, true} ->
                             %% q3 is now empty, it wasn't before;
                             %% delta is still empty. So q2 must be
                             %% empty, and we know q4 is empty
                             %% otherwise we wouldn't be loading from
                             %% q3. As such, we can just set q4 to Q1.
                             true = ?QUEUE:is_empty(Q2), %% ASSERTION
                             true = ?QUEUE:is_empty(Q4), %% ASSERTION
                             State1 #vqstate { q1 = ?QUEUE:new(), q4 = Q1 };
                         {true, false} ->
                             maybe_deltas_to_betas(State1);
                         {false, _} ->
                             %% q3 still isn't empty, we've not
                             %% touched delta, so the invariants
                             %% between q1, q2, delta and q3 are
                             %% maintained
                             State1
                     end,
            {loaded, {MsgStatus, State2}}
    end.

maybe_deltas_to_betas(State = #vqstate { delta = ?BLANK_DELTA_PATTERN(X) }) ->
    State;
maybe_deltas_to_betas(State = #vqstate {
                        q2                   = Q2,
                        delta                = Delta,
                        q3                   = Q3,
                        index_state          = IndexState,
                        ram_pending_ack      = RPA,
                        disk_pending_ack     = DPA,
                        transient_threshold  = TransientThreshold }) ->
    #delta { start_seq_id = DeltaSeqId,
             count        = DeltaCount,
             end_seq_id   = DeltaSeqIdEnd } = Delta,
    DeltaSeqId1 =
        lists:min([rabbit_queue_index:next_segment_boundary(DeltaSeqId),
                   DeltaSeqIdEnd]),
    {List, IndexState1} = rabbit_queue_index:read(DeltaSeqId, DeltaSeqId1,
                                                  IndexState),
    {Q3a, IndexState2} = betas_from_index_entries(List, TransientThreshold,
                                                  RPA, DPA, IndexState1),
    State1 = State #vqstate { index_state = IndexState2 },
    case ?QUEUE:len(Q3a) of
        0 ->
            %% we ignored every message in the segment due to it being
            %% transient and below the threshold
            maybe_deltas_to_betas(
              State1 #vqstate {
                delta = d(Delta #delta { start_seq_id = DeltaSeqId1 })});
        Q3aLen ->
            Q3b = ?QUEUE:join(Q3, Q3a),
            case DeltaCount - Q3aLen of
                0 ->
                    %% delta is now empty, but it wasn't before, so
                    %% can now join q2 onto q3
                    State1 #vqstate { q2    = ?QUEUE:new(),
                                      delta = ?BLANK_DELTA,
                                      q3    = ?QUEUE:join(Q3b, Q2) };
                N when N > 0 ->
                    Delta1 = d(#delta { start_seq_id = DeltaSeqId1,
                                        count        = N,
                                        end_seq_id   = DeltaSeqIdEnd }),
                    State1 #vqstate { delta = Delta1,
                                      q3    = Q3b }
            end
    end.

push_alphas_to_betas(Quota, State) ->
    {Quota1, State1} =
        push_alphas_to_betas(
          fun ?QUEUE:out/1,
          fun (MsgStatus, Q1a,
               State0 = #vqstate { q3 = Q3, delta = #delta { count = 0 } }) ->
                  State0 #vqstate { q1 = Q1a, q3 = ?QUEUE:in(MsgStatus, Q3) };
              (MsgStatus, Q1a, State0 = #vqstate { q2 = Q2 }) ->
                  State0 #vqstate { q1 = Q1a, q2 = ?QUEUE:in(MsgStatus, Q2) }
          end, Quota, State #vqstate.q1, State),
    {Quota2, State2} =
        push_alphas_to_betas(
          fun ?QUEUE:out_r/1,
          fun (MsgStatus, Q4a, State0 = #vqstate { q3 = Q3 }) ->
                  State0 #vqstate { q3 = ?QUEUE:in_r(MsgStatus, Q3), q4 = Q4a }
          end, Quota1, State1 #vqstate.q4, State1),
    {Quota2, State2}.

push_alphas_to_betas(_Generator, _Consumer, Quota, _Q,
                     State = #vqstate { ram_msg_count    = RamMsgCount,
                                        target_ram_count = TargetRamCount })
  when Quota =:= 0 orelse
       TargetRamCount =:= infinity orelse
       TargetRamCount >= RamMsgCount ->
    {Quota, State};
push_alphas_to_betas(Generator, Consumer, Quota, Q, State) ->
    case credit_flow:blocked() of
        true  -> {Quota, State};
        false -> case Generator(Q) of
                     {empty, _Q} ->
                         {Quota, State};
                     {{value, MsgStatus}, Qa} ->
                         {MsgStatus1 = #msg_status { msg_on_disk = true },
                          State1 = #vqstate { ram_msg_count = RamMsgCount }} =
                             maybe_write_to_disk(true, false, MsgStatus, State),
                         MsgStatus2 = m(trim_msg_status(MsgStatus1)),
                         State2 = Consumer(MsgStatus2, Qa,
                                           State1 #vqstate {
                                             ram_msg_count = RamMsgCount - 1 }),
                         push_alphas_to_betas(Generator, Consumer, Quota - 1,
                                              Qa, State2)
                 end
    end.

push_betas_to_deltas(Quota, State = #vqstate { q2          = Q2,
                                               delta       = Delta,
                                               q3          = Q3,
                                               index_state = IndexState }) ->
    PushState = {Quota, Delta, IndexState},
    {Q3a, PushState1} = push_betas_to_deltas(
                          fun ?QUEUE:out_r/1,
                          fun rabbit_queue_index:next_segment_boundary/1,
                          Q3, PushState),
    {Q2a, PushState2} = push_betas_to_deltas(
                          fun ?QUEUE:out/1,
                          fun (Q2MinSeqId) -> Q2MinSeqId end,
                          Q2, PushState1),
    {_, Delta1, IndexState1} = PushState2,
    State #vqstate { q2          = Q2a,
                     delta       = Delta1,
                     q3          = Q3a,
                     index_state = IndexState1 }.

push_betas_to_deltas(Generator, LimitFun, Q, PushState) ->
    case ?QUEUE:is_empty(Q) of
        true ->
            {Q, PushState};
        false ->
            {value, #msg_status { seq_id = MinSeqId }} = ?QUEUE:peek(Q),
            {value, #msg_status { seq_id = MaxSeqId }} = ?QUEUE:peek_r(Q),
            Limit = LimitFun(MinSeqId),
            case MaxSeqId < Limit of
                true  -> {Q, PushState};
                false -> push_betas_to_deltas1(Generator, Limit, Q, PushState)
            end
    end.

push_betas_to_deltas1(_Generator, _Limit, Q,
                      {0, _Delta, _IndexState} = PushState) ->
    {Q, PushState};
push_betas_to_deltas1(Generator, Limit, Q,
                      {Quota, Delta, IndexState} = PushState) ->
    case Generator(Q) of
        {empty, _Q} ->
            {Q, PushState};
        {{value, #msg_status { seq_id = SeqId }}, _Qa}
          when SeqId < Limit ->
            {Q, PushState};
        {{value, MsgStatus = #msg_status { seq_id = SeqId }}, Qa} ->
            {#msg_status { index_on_disk = true }, IndexState1} =
                maybe_write_index_to_disk(true, MsgStatus, IndexState),
            Delta1 = expand_delta(SeqId, Delta),
            push_betas_to_deltas1(Generator, Limit, Qa,
                                  {Quota - 1, Delta1, IndexState1})
    end.

%%----------------------------------------------------------------------------
%% Upgrading
%%----------------------------------------------------------------------------

multiple_routing_keys() ->
    transform_storage(
      fun ({basic_message, ExchangeName, Routing_Key, Content,
            MsgId, Persistent}) ->
              {ok, {basic_message, ExchangeName, [Routing_Key], Content,
                    MsgId, Persistent}};
          (_) -> {error, corrupt_message}
      end),
    ok.


%% Assumes message store is not running
transform_storage(TransformFun) ->
    transform_store(?PERSISTENT_MSG_STORE, TransformFun),
    transform_store(?TRANSIENT_MSG_STORE, TransformFun).

transform_store(Store, TransformFun) ->
    rabbit_msg_store:force_recovery(rabbit_mnesia:dir(), Store),
    rabbit_msg_store:transform_dir(rabbit_mnesia:dir(), Store, TransformFun).