Coverage Report

Created: 2025-11-14 03:48

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/root/doris/be/src/olap/schema_change.h
Line
Count
Source
1
// Licensed to the Apache Software Foundation (ASF) under one
2
// or more contributor license agreements.  See the NOTICE file
3
// distributed with this work for additional information
4
// regarding copyright ownership.  The ASF licenses this file
5
// to you under the Apache License, Version 2.0 (the
6
// "License"); you may not use this file except in compliance
7
// with the License.  You may obtain a copy of the License at
8
//
9
//   http://www.apache.org/licenses/LICENSE-2.0
10
//
11
// Unless required by applicable law or agreed to in writing,
12
// software distributed under the License is distributed on an
13
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14
// KIND, either express or implied.  See the License for the
15
// specific language governing permissions and limitations
16
// under the License.
17
18
#pragma once
19
20
#include <butil/macros.h>
21
#include <fmt/format.h>
22
#include <gen_cpp/Descriptors_types.h>
23
#include <glog/logging.h>
24
#include <stddef.h>
25
#include <stdint.h>
26
27
#include <memory>
28
#include <ostream>
29
#include <set>
30
#include <shared_mutex>
31
#include <string>
32
#include <unordered_map>
33
#include <unordered_set>
34
#include <utility>
35
#include <vector>
36
37
#include "common/config.h"
38
#include "common/object_pool.h"
39
#include "common/status.h"
40
#include "olap/column_mapping.h"
41
#include "olap/olap_common.h"
42
#include "olap/rowset/pending_rowset_helper.h"
43
#include "olap/rowset/rowset.h"
44
#include "olap/rowset/rowset_reader.h"
45
#include "olap/rowset/rowset_writer.h"
46
#include "olap/rowset/segment_v2/inverted_index_writer.h"
47
#include "olap/storage_engine.h"
48
#include "olap/tablet.h"
49
#include "olap/tablet_fwd.h"
50
#include "olap/tablet_schema.h"
51
#include "runtime/descriptors.h"
52
#include "runtime/memory/mem_tracker.h"
53
#include "vec/data_types/data_type.h"
54
55
namespace doris {
56
class DeleteHandler;
57
class Field;
58
class TAlterInvertedIndexReq;
59
class TAlterTabletReqV2;
60
class TExpr;
61
enum AlterTabletType : int;
62
enum RowsetTypePB : int;
63
enum SegmentsOverlapPB : int;
64
65
namespace vectorized {
66
class Block;
67
class OlapBlockDataConvertor;
68
} // namespace vectorized
69
70
class BlockChanger {
71
public:
72
    BlockChanger(TabletSchemaSPtr tablet_schema, DescriptorTbl desc_tbl);
73
74
    ~BlockChanger();
75
76
    ColumnMapping* get_mutable_column_mapping(size_t column_index);
77
78
    Status change_block(vectorized::Block* ref_block, vectorized::Block* new_block) const;
79
80
0
    void set_where_expr(const std::shared_ptr<TExpr>& where_expr) { _where_expr = where_expr; }
81
82
0
    void set_type(AlterTabletType type) { _type = type; }
83
84
0
    void set_compatible_version(int32_t version) noexcept { _fe_compatible_version = version; }
85
86
0
    bool has_where() const { return _where_expr != nullptr; }
87
88
private:
89
    static Status _check_cast_valid(vectorized::ColumnPtr ref_column,
90
                                    vectorized::ColumnPtr new_column);
91
92
    // @brief column-mapping specification of new schema
93
    SchemaMapping _schema_mapping;
94
95
    DescriptorTbl _desc_tbl;
96
97
    std::shared_ptr<TExpr> _where_expr;
98
99
    AlterTabletType _type;
100
101
    int32_t _fe_compatible_version = -1;
102
};
103
104
class SchemaChange {
105
public:
106
0
    SchemaChange() = default;
107
0
    virtual ~SchemaChange() = default;
108
109
    virtual Status process(RowsetReaderSharedPtr rowset_reader, RowsetWriter* rowset_writer,
110
                           BaseTabletSPtr new_tablet, BaseTabletSPtr base_tablet,
111
                           TabletSchemaSPtr base_tablet_schema,
112
0
                           TabletSchemaSPtr new_tablet_schema) {
113
0
        if (rowset_reader->rowset()->empty() || rowset_reader->rowset()->num_rows() == 0) {
114
0
            RETURN_IF_ERROR(rowset_writer->flush());
115
0
            return Status::OK();
116
0
        }
117
118
0
        _filtered_rows = 0;
119
0
        _merged_rows = 0;
120
0
        RETURN_IF_ERROR_OR_CATCH_EXCEPTION(_inner_process(rowset_reader, rowset_writer, new_tablet,
121
0
                                                          base_tablet_schema, new_tablet_schema));
122
123
        // Check row num changes
124
0
        if (!_check_row_nums(rowset_reader, *rowset_writer)) {
125
0
            return Status::Error<ErrorCode::ALTER_STATUS_ERR>("SchemaChange check row nums failed");
126
0
        }
127
128
0
        LOG(INFO) << "all row nums. source_rows=" << rowset_reader->rowset()->num_rows()
129
0
                  << ", source_filtered_rows=" << rowset_reader->filtered_rows()
130
0
                  << ", source_merged_rows=" << rowset_reader->merged_rows()
131
0
                  << ", merged_rows=" << merged_rows() << ", filtered_rows=" << filtered_rows()
132
0
                  << ", new_index_rows=" << rowset_writer->num_rows()
133
0
                  << ", writer_filtered_rows=" << rowset_writer->num_rows_filtered();
134
0
        return Status::OK();
135
0
    }
136
137
0
    uint64_t filtered_rows() const { return _filtered_rows; }
138
139
0
    uint64_t merged_rows() const { return _merged_rows; }
140
141
protected:
142
0
    void _add_filtered_rows(uint64_t filtered_rows) { _filtered_rows += filtered_rows; }
143
144
0
    void _add_merged_rows(uint64_t merged_rows) { _merged_rows += merged_rows; }
145
146
    virtual Status _inner_process(RowsetReaderSharedPtr rowset_reader, RowsetWriter* rowset_writer,
147
                                  BaseTabletSPtr new_tablet, TabletSchemaSPtr base_tablet_schema,
148
0
                                  TabletSchemaSPtr new_tablet_schema) {
149
0
        return Status::NotSupported("inner process unsupported.");
150
0
    }
151
152
0
    virtual bool _check_row_nums(RowsetReaderSharedPtr reader, const RowsetWriter& writer) const {
153
0
        if (reader->rowset()->num_rows() - reader->filtered_rows() - reader->merged_rows() !=
154
0
            writer.num_rows() + writer.num_rows_filtered() + _merged_rows + _filtered_rows) {
155
0
            LOG(WARNING) << "fail to check row num! "
156
0
                         << "source_rows=" << reader->rowset()->num_rows()
157
0
                         << ", source_filtered_rows=" << reader->filtered_rows()
158
0
                         << ", source_merged_rows=" << reader->merged_rows()
159
0
                         << ", written_rows=" << writer.num_rows()
160
0
                         << ", writer_filtered_rows=" << writer.num_rows_filtered()
161
0
                         << ", merged_rows=" << merged_rows()
162
0
                         << ", filtered_rows=" << filtered_rows();
163
0
            if (!config::ignore_schema_change_check) {
164
0
                return false;
165
0
            }
166
0
        }
167
0
        return true;
168
0
    }
169
170
private:
171
    uint64_t _filtered_rows {};
172
    uint64_t _merged_rows {};
173
174
protected:
175
    std::vector<bool> _row_same_bit;
176
};
177
178
class LinkedSchemaChange : public SchemaChange {
179
public:
180
0
    LinkedSchemaChange() = default;
181
0
    ~LinkedSchemaChange() override = default;
182
183
    Status process(RowsetReaderSharedPtr rowset_reader, RowsetWriter* rowset_writer,
184
                   BaseTabletSPtr new_tablet, BaseTabletSPtr base_tablet,
185
                   TabletSchemaSPtr base_tablet_schema,
186
                   TabletSchemaSPtr new_tablet_schema) override;
187
188
private:
189
    DISALLOW_COPY_AND_ASSIGN(LinkedSchemaChange);
190
};
191
192
class VSchemaChangeDirectly : public SchemaChange {
193
public:
194
0
    VSchemaChangeDirectly(const BlockChanger& changer) : _changer(changer) {}
195
196
private:
197
    Status _inner_process(RowsetReaderSharedPtr rowset_reader, RowsetWriter* rowset_writer,
198
                          BaseTabletSPtr new_tablet, TabletSchemaSPtr base_tablet_schema,
199
                          TabletSchemaSPtr new_tablet_schema) override;
200
201
0
    bool _check_row_nums(RowsetReaderSharedPtr reader, const RowsetWriter& writer) const override {
202
0
        return _changer.has_where() || SchemaChange::_check_row_nums(reader, writer);
203
0
    }
204
205
    const BlockChanger& _changer;
206
};
207
208
class VBaseSchemaChangeWithSorting : public SchemaChange {
209
public:
210
    VBaseSchemaChangeWithSorting(const BlockChanger& changer, size_t memory_limitation);
211
0
    ~VBaseSchemaChangeWithSorting() override = default;
212
213
    Status _inner_process(RowsetReaderSharedPtr rowset_reader, RowsetWriter* rowset_writer,
214
                          BaseTabletSPtr new_tablet, TabletSchemaSPtr base_tablet_schema,
215
                          TabletSchemaSPtr new_tablet_schema) override;
216
217
    virtual Result<RowsetSharedPtr> _internal_sorting(
218
            const std::vector<std::unique_ptr<vectorized::Block>>& blocks,
219
            const Version& temp_delta_versions, int64_t newest_write_timestamp,
220
            BaseTabletSPtr new_tablet, RowsetTypePB new_rowset_type,
221
            SegmentsOverlapPB segments_overlap, TabletSchemaSPtr new_tablet_schema);
222
223
    Status _external_sorting(std::vector<RowsetSharedPtr>& src_rowsets, RowsetWriter* rowset_writer,
224
                             BaseTabletSPtr new_tablet, TabletSchemaSPtr new_tablet_schema);
225
226
protected:
227
    // for external sorting
228
    // src_rowsets to store the rowset generated by internal sorting
229
    std::vector<RowsetSharedPtr> _src_rowsets;
230
231
private:
232
0
    bool _check_row_nums(RowsetReaderSharedPtr reader, const RowsetWriter& writer) const override {
233
0
        return _changer.has_where() || SchemaChange::_check_row_nums(reader, writer);
234
0
    }
235
236
    const BlockChanger& _changer;
237
    size_t _memory_limitation;
238
    Version _temp_delta_versions;
239
    std::unique_ptr<MemTracker> _mem_tracker;
240
};
241
242
// @breif schema change with sorting
243
// Mixin for local StorageEngine
244
class VLocalSchemaChangeWithSorting final : public VBaseSchemaChangeWithSorting {
245
public:
246
    VLocalSchemaChangeWithSorting(const BlockChanger& changer, size_t memory_limitation,
247
                                  StorageEngine& local_storage_engine)
248
0
            : VBaseSchemaChangeWithSorting(changer, memory_limitation),
249
0
              _local_storage_engine(local_storage_engine) {}
250
0
    ~VLocalSchemaChangeWithSorting() override = default;
251
252
    Status _inner_process(RowsetReaderSharedPtr rowset_reader, RowsetWriter* rowset_writer,
253
                          BaseTabletSPtr new_tablet, TabletSchemaSPtr base_tablet_schema,
254
                          TabletSchemaSPtr new_tablet_schema) override;
255
256
    Result<RowsetSharedPtr> _internal_sorting(
257
            const std::vector<std::unique_ptr<vectorized::Block>>& blocks,
258
            const Version& temp_delta_versions, int64_t newest_write_timestamp,
259
            BaseTabletSPtr new_tablet, RowsetTypePB new_rowset_type,
260
            SegmentsOverlapPB segments_overlap, TabletSchemaSPtr new_tablet_schema) override;
261
262
private:
263
    StorageEngine& _local_storage_engine;
264
    std::vector<PendingRowsetGuard> _pending_rs_guards;
265
};
266
267
struct AlterMaterializedViewParam {
268
    std::string column_name;
269
    std::string origin_column_name;
270
    std::shared_ptr<TExpr> expr;
271
};
272
273
struct SchemaChangeParams {
274
    AlterTabletType alter_tablet_type;
275
    bool enable_unique_key_merge_on_write = false;
276
    std::vector<RowsetReaderSharedPtr> ref_rowset_readers;
277
    DeleteHandler* delete_handler = nullptr;
278
    std::unordered_map<std::string, AlterMaterializedViewParam> materialized_params_map;
279
    DescriptorTbl* desc_tbl = nullptr;
280
    ObjectPool pool;
281
    int32_t be_exec_version;
282
    std::string vault_id;
283
};
284
285
class SchemaChangeJob {
286
public:
287
    SchemaChangeJob(StorageEngine& local_storage_engine, const TAlterTabletReqV2& request,
288
                    const std::string& job_id);
289
    Status process_alter_tablet(const TAlterTabletReqV2& request);
290
291
    bool tablet_in_converting(int64_t tablet_id);
292
293
    static Status parse_request(const SchemaChangeParams& sc_params,
294
                                TabletSchema* base_tablet_schema, TabletSchema* new_tablet_schema,
295
                                BlockChanger* changer, bool* sc_sorting, bool* sc_directly);
296
297
private:
298
    std::unique_ptr<SchemaChange> _get_sc_procedure(const BlockChanger& changer, bool sc_sorting,
299
0
                                                    bool sc_directly, int64_t mem_limit) {
300
0
        if (sc_sorting) {
301
0
            return std::make_unique<VLocalSchemaChangeWithSorting>(changer, mem_limit,
302
0
                                                                   _local_storage_engine);
303
0
        }
304
305
0
        if (sc_directly) {
306
0
            return std::make_unique<VSchemaChangeDirectly>(changer);
307
0
        }
308
309
0
        return std::make_unique<LinkedSchemaChange>();
310
0
    }
311
312
    Status _get_versions_to_be_changed(std::vector<Version>* versions_to_be_changed,
313
                                       RowsetSharedPtr* max_rowset);
314
315
    Status _do_process_alter_tablet(const TAlterTabletReqV2& request);
316
317
    Status _validate_alter_result(const TAlterTabletReqV2& request);
318
319
    Status _convert_historical_rowsets(const SchemaChangeParams& sc_params,
320
                                       int64_t* real_alter_version);
321
322
    // Initialization Settings for creating a default value
323
    static Status _init_column_mapping(ColumnMapping* column_mapping,
324
                                       const TabletColumn& column_schema, const std::string& value);
325
326
    Status _calc_delete_bitmap_for_mow_table(int64_t alter_version);
327
328
    StorageEngine& _local_storage_engine;
329
    TabletSharedPtr _base_tablet;
330
    TabletSharedPtr _new_tablet;
331
    TabletSchemaSPtr _base_tablet_schema;
332
    TabletSchemaSPtr _new_tablet_schema;
333
    std::shared_mutex _mutex;
334
    std::unordered_set<int64_t> _tablet_ids_in_converting;
335
    std::set<std::string> _supported_functions;
336
    std::string _job_id;
337
};
338
} // namespace doris