summaryrefslogtreecommitdiff
path: root/src/partitions.erl
blob: 3e024264afae43d40e41e6b9285dc02ab239f90d (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
-module(partitions).
-author('brad@cloudant.com').

%% API
-export([fullmap/2, fullmap/3, hash/1, install_fullmap/4]).
-export([for_key/2, all_parts/1]).
-export([shard_name/2]).

-define(RINGTOP, trunc(math:pow(2,160))).  % SHA-1 space

-include("../../couch/src/couch_db.hrl").
-include("../../dynomite/include/membership.hrl").
-include_lib("eunit/include/eunit.hrl").

%%====================================================================
%% API
%%====================================================================

%% @doc build a full partition map
fullmap(DbName, Options) ->
    {ok, Nodes} = mem3:nodes(),
    fullmap(DbName, Nodes, Options).

fullmap(DbName, Nodes, Options) ->
    {N,Q} = db_init_constants(Options),
    NewNodes = ordered_nodes(DbName, Nodes),
    Pmap = pmap(Q, NewNodes),
    int_fullmap(DbName, N, Pmap, NewNodes).

%% @spec hash(term()) -> Digest::binary()
%% @doc uses SHA-1 as its hash
hash(Item) when is_binary(Item) ->
    crypto:sha(Item);
hash(Item) ->
    crypto:sha(term_to_binary(Item)).

install_fullmap(DbName, Fullmap, FullNodes, Options) ->
    {N,Q} = db_init_constants(Options),
    Doc = {[{<<"_id">>,DbName},
            {<<"map">>, jsonify(<<"map">>, Fullmap)},
            {<<"nodes">>, jsonify(<<"nodes">>, FullNodes)},
            {<<"n">>,N},
            {<<"q">>,Q}]},
    write_db_doc(Doc).

for_key(DbName, Key) ->
    <<HashKey:160/integer>> = hash(Key),
    Head = #shard{
        name = '_',
        node = '_',
        dbname = DbName,
        range = ['$1','$2'],
        ref = '_'
    },
    % TODO these conditions assume A < B, which we don't require
    Conditions = [{'<', '$1', HashKey}, {'<', HashKey, '$2'}],
    case ets:select(partitions, [{Head, Conditions, ['$_']}]) of
    [] ->
        erlang:error(database_does_not_exist);
    Shards ->
        Shards
    end.

all_parts(DbName) ->
    case ets:lookup(partitions, DbName) of
    [] ->
        erlang:error(database_does_not_exist);
    Else ->
        Else
    end.

%%====================================================================
%% Internal functions
%%====================================================================

%% @doc get cluster constants from options or config
db_init_constants(Options) ->
    {const(n, Options), const(q, Options)}.

%% @doc get individual constant
const(Const, Options) ->
    ListResult = case couch_util:get_value(Const, Options) of
    undefined -> couch_config:get("cluster", atom_to_list(Const));
    Val -> Val
    end,
    list_to_integer(ListResult).

%% @doc hash the dbname, and return the corresponding node for seeding a ring
seednode(DbName, Nodes) ->
    <<HashInt:160/integer>> = hash(DbName),
    Size = partition_range(length(Nodes)),
    Factor = (HashInt div Size),
    lists:nth(Factor+1, Nodes).

%% @doc take the list of nodes, and rearrange it, starting with the node that
%%      results from hashing the Term
ordered_nodes(Term, Nodes) ->
    SeedNode = seednode(Term, Nodes),
    {A, B} = lists:splitwith(fun(N) -> N /= SeedNode end, Nodes),
    lists:append(B,A).

%% @doc create a partition map
pmap(NumPartitions, Nodes) ->
    Increment = ?RINGTOP div NumPartitions,
    Parts = parts(?RINGTOP, Increment, 0, []),
    make_map(Nodes, Nodes, Parts, []).

%% @doc makes a {beg, end} list of partition ranges
%%      last range may have an extra few values, because Increment is created
%%      with Ringtop 'div' NumPartitions above.
parts(Top, _, Beg, Acc) when Beg > Top -> Acc;
parts(Top, Increment, Beg, Acc) ->
    End = case Beg + 2*Increment of
    Over when Over > Top -> Top;
    _ -> Beg + Increment - 1
    end,
    NewAcc = [{Beg, End} | Acc],
    parts(Top, Increment, End+1, NewAcc).

%% @doc create a full map, which is a pmap with N-1 replication partner nodes
%%      added per partition
int_fullmap(DbName, N, Pmap, Nodes) ->
    Full = lists:foldl(fun({Node,{B,E} = Part}, AccIn) ->
        Primary = [#shard{dbname=DbName, node=Node, range=[B,E],
                          name=shard_name(B,DbName)}],
        Partners = partners(DbName, N, Node, Nodes, Part),
        lists:append([Primary, Partners, AccIn])
    end, [], Pmap),
    lists:reverse(Full).

partners(DbName, N, Node, Nodes, {Beg,End}) ->
    {A, [Node|B]} = lists:splitwith(fun(Nd) -> Nd /= Node end, Nodes),
    Nodes1 = lists:append(B,A),
    Partners = lists:sublist(Nodes1, N-1), % N-1 replication partner nodes
    lists:map(fun(Partner) ->
        #shard{dbname=DbName, node=Partner, range=[Beg,End],
               name=shard_name(Beg,DbName)}
    end, Partners).

%% @doc size of one partition in the ring
partition_range(Q) ->
    trunc( ?RINGTOP / Q ).  % SHA-1 space / Q

%% @doc assign nodes to each of the partitions.  When you run out of nodes,
%%      start at the beginning of the node list again.
%%      The provided node list starts with the seed node (seednode fun)
make_map(_,_,[], Acc) ->
    lists:keysort(2,Acc);
make_map(AllNodes, [], Parts, Acc) ->
    % start back at beginning of node list
    make_map(AllNodes, AllNodes, Parts, Acc);
make_map(AllNodes, [Node|RestNodes], [Part|RestParts], Acc) ->
    % add a node/part combo to the Acc
    make_map(AllNodes, RestNodes, RestParts, [{Node,Part}|Acc]).

jsonify(<<"map">>, Map) ->
    lists:map(fun(#shard{node=Node, range=[Beg,End]}) ->
        {[{<<"node">>, Node}, {<<"b">>, Beg}, {<<"e">>, End}]}
    end, Map);
jsonify(<<"nodes">>, Nodes) ->
    lists:map(fun({Order, Node, Options}) ->
        {[{<<"order">>, Order}, {<<"node">>, Node}, {<<"options">>, Options}]}
    end, Nodes).

write_db_doc(EDoc) ->
    {ok, Db} = couch_db:open(<<"dbs">>, []),
    try
        update_db_doc(Db, couch_doc:from_json_obj(EDoc))
    catch {conflict, _} ->
        ?LOG_ERROR("conflict writing db doc, must be a race", [])
    after
        couch_db:close(Db)
    end.

update_db_doc(Db, #doc{id=Id, body=Body} = Doc) ->
    case couch_db:open_doc(Db, Id, []) of
    {not_found, _} ->
        {ok, _} = couch_db:update_doc(Db, Doc, []);
    {ok, #doc{body=Body}} ->
        ok;
    {ok, OldDoc} ->
        {ok, _} = couch_db:update_doc(Db, OldDoc#doc{body=Body}, [])
    end.

shard_name(Part, DbName) when is_list(DbName) ->
    shard_name(Part, ?l2b(DbName));
shard_name(Part, DbName) ->
  PartHex = ?l2b(showroom_utils:int_to_hexstr(Part)),
  <<"x", PartHex/binary, "/", DbName/binary, "_", PartHex/binary>>.