summaryrefslogtreecommitdiff
path: root/src/mongo/db/s/shard_metadata_util.h
blob: cc9f96097ba3b3b86952d28029c29d0d8bc9c369 (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
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
/**
 *    Copyright (C) 2018-present MongoDB, Inc.
 *
 *    This program is free software: you can redistribute it and/or modify
 *    it under the terms of the Server Side Public License, version 1,
 *    as published by MongoDB, Inc.
 *
 *    This program is distributed in the hope that it will be useful,
 *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *    Server Side Public License for more details.
 *
 *    You should have received a copy of the Server Side Public License
 *    along with this program. If not, see
 *    <http://www.mongodb.com/licensing/server-side-public-license>.
 *
 *    As a special exception, the copyright holders give permission to link the
 *    code of portions of this program with the OpenSSL library under certain
 *    conditions as described in each individual source file and distribute
 *    linked combinations including the program with the OpenSSL library. You
 *    must comply with the Server Side Public License in all respects for
 *    all of the code used other than as permitted herein. If you modify file(s)
 *    with this exception, you may extend this exception to your version of the
 *    file(s), but you are not obligated to do so. If you do not wish to do so,
 *    delete this exception statement from your version. If you delete this
 *    exception statement from all source files in the program, then also delete
 *    it in the license file.
 */

#pragma once

#include <string>
#include <vector>

#include "mongo/base/status_with.h"
#include "mongo/bson/bsonobj.h"
#include "mongo/bson/oid.h"
#include "mongo/s/chunk_version.h"

namespace mongo {

class ChunkType;
class NamespaceString;
class OperationContext;
class ShardCollectionType;
class ShardDatabaseType;

namespace shardmetadatautil {

/**
 * Structure representing the generated query and sort order for a chunk diffing operation.
 */
struct QueryAndSort {
    const BSONObj query;
    const BSONObj sort;
};

/**
 * Returns the query needed to find incremental changes to the chunks collection on a shard server.
 *
 * The query has to find all the chunks $gte the current max version. Currently, any splits, merges
 * and moves will increment the current max version. Querying by lastmod is essential because we
 * want to use the {lastmod} index on the chunks collection. This makes potential cursor yields to
 * apply split/merge/move updates safe: updates always move or insert documents at the end of the
 * index (because the document updates always have higher lastmod), so changed always come *after*
 * our current cursor position and are seen when the cursor recommences.
 *
 * The sort must be by ascending version so that the updates can be applied in-memory in order. This
 * is important because it is possible for a cursor to read updates to the same _id document twice,
 * due to the yield described above. If updates are applied in ascending version order, the newer
 * update is applied last.
 */
QueryAndSort createShardChunkDiffQuery(const ChunkVersion& collectionPlacementVersion);

/**
 * Writes a persisted signal to indicate that it is once again safe to read from the chunks
 * collection for 'nss' and updates the collection's collection placement version to
 * 'refreshedVersion'. It is essential to call this after updating the chunks collection so that
 * secondaries know they can safely use the chunk metadata again.
 *
 * It is safe to call this multiple times: it's an idempotent action.
 *
 * refreshedVersion - the new collection placement version for the completed refresh.
 *
 * Note: if there is no document present in the collections collection for 'nss', nothing is
 * updated.
 */
Status unsetPersistedRefreshFlags(OperationContext* opCtx,
                                  const NamespaceString& nss,
                                  const ChunkVersion& refreshedVersion);

/**
 * Represents a subset of a collection's config.cache.collections entry that relates to refresh
 * state.
 */
struct RefreshState {
    bool operator==(const RefreshState& other) const;

    std::string toString() const;

    // The current generation of the collection.
    CollectionGeneration generation;

    // Whether a refresh is currently in progress.
    bool refreshing;

    // The collection placement version after the last complete refresh. Indicates change if
    // refreshing has started and finished since last loaded.
    ChunkVersion lastRefreshedCollectionPlacementVersion;
};

/**
 * Reads the persisted refresh signal for 'nss' and returns those settings.
 */
StatusWith<RefreshState> getPersistedRefreshFlags(OperationContext* opCtx,
                                                  const NamespaceString& nss);

/**
 * Reads the shard server's collections collection entry identified by 'nss'.
 */
StatusWith<ShardCollectionType> readShardCollectionsEntry(OperationContext* opCtx,
                                                          const NamespaceString& nss);

/**
 * Reads the shard server's databases collection entry identified by 'dbName'.
 */
StatusWith<ShardDatabaseType> readShardDatabasesEntry(OperationContext* opCtx, StringData dbName);

/**
 * Updates the collections collection entry matching 'query' with 'update' using local write
 * concern.
 *
 * If 'upsert' is true, expects 'lastRefreshedCollectionPlacementVersion' to be absent in the
 * update: these refreshing fields should only be added to an existing document.
 */
Status updateShardCollectionsEntry(OperationContext* opCtx,
                                   const BSONObj& query,
                                   const BSONObj& update,
                                   bool upsert);

/**
 * Updates the databases collection entry matching 'query' with 'update' using local write
 * concern.
 *
 * Uses the $set operator on the update so that updates can be applied without resetting everything.
 * 'inc' can be used to specify fields and their increments: it will be assigned to the $inc
 * operator.
 *
 * 'inc' should not specify 'upsert' true.
 */
Status updateShardDatabasesEntry(OperationContext* opCtx,
                                 const BSONObj& query,
                                 const BSONObj& update,
                                 const BSONObj& inc,
                                 bool upsert);

/**
 * Reads the shard server's chunks collection corresponding to 'nss' for chunks matching 'query',
 * returning at most 'limit' chunks in 'sort' order. 'epoch' populates the returned chunks' version
 * fields, because we do not yet have UUIDs to replace epochs nor UUIDs associated with namespaces.
 */
StatusWith<std::vector<ChunkType>> readShardChunks(OperationContext* opCtx,
                                                   const NamespaceString& nss,
                                                   const BSONObj& query,
                                                   const BSONObj& sort,
                                                   boost::optional<long long> limit,
                                                   const OID& epoch,
                                                   const Timestamp& timestamp);

/**
 * Takes a vector of 'chunks' and updates the shard's chunks collection for 'nss'. Any chunk
 * documents in config.cache.chunks.<ns> that overlap with a chunk in 'chunks' is removed as the
 * updated chunk document is inserted. If the epoch of a chunk in 'chunks' does not match
 * 'currEpoch', a ConflictingOperationInProgress error is returned and no more updates are applied.
 *
 * Note: two threads running this function in parallel for the same collection can corrupt the
 * collection data!
 *
 * nss - the collection namespace for which chunk metadata is being updated.
 * chunks - chunks retrieved from the config server, sorted in ascending chunk version order.
 * currEpoch - what this shard server expects the collection epoch to be.
 *
 * Returns:
 * - ConflictingOperationInProgress if the chunk version epoch of any chunk in 'chunks' is different
 *   than 'currEpoch'.
 * - Other errors if unable to do local writes/reads to the config.chunks.ns collection.
 */
Status updateShardChunks(OperationContext* opCtx,
                         const NamespaceString& nss,
                         const std::vector<ChunkType>& chunks,
                         const OID& currEpoch);

/**
 * Deletes locally persisted chunk metadata associated with 'nss': drops the chunks collection
 * and removes the collections collection entry.
 *
 * The order is important because the secondary observes changes to the config.collections entries.
 * If the chunks were dropped first, the secondary would keep refreshing until it exceeded its
 * retries, rather than returning with a useful error message.
 */
Status dropChunksAndDeleteCollectionsEntry(OperationContext* opCtx, const NamespaceString& nss);

/**
 * Drops locally persisted chunk metadata associated with 'nss': only drops the chunks collection.
 */
void dropChunks(OperationContext* opCtx, const NamespaceString& nss);

/**
 * Deletes locally persisted database metadata associated with 'dbName': removes the databases
 * collection entry.
 */
Status deleteDatabasesEntry(OperationContext* opCtx, StringData dbName);

}  // namespace shardmetadatautil
}  // namespace mongo