summaryrefslogtreecommitdiff
path: root/ovsdb/storage.c
blob: d727b1eacd5da0f74cba67a7b2961421d04e8ed1 (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
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649

/* Copyright (c) 2009, 2010, 2011, 2016, 2017 Nicira, Inc.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this storage except in compliance with the License.
 * You may obtain a copy of the License at:
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#include <config.h>

#include "storage.h"
#include <string.h>
#include "log.h"
#include "ovsdb-error.h"
#include "openvswitch/json.h"
#include "openvswitch/poll-loop.h"
#include "openvswitch/vlog.h"
#include "ovsdb.h"
#include "raft.h"
#include "random.h"
#include "simap.h"
#include "timeval.h"
#include "util.h"

VLOG_DEFINE_THIS_MODULE(storage);

struct ovsdb_storage {
    /* There are three kinds of storage:
     *
     *    - Standalone, backed by a disk file.  'log' is nonnull, 'raft' is
     *      null.
     *
     *    - Clustered, backed by a Raft cluster.  'log' is null, 'raft' is
     *      nonnull.
     *
     *    - Memory only, unbacked.  'log' and 'raft' are null. */
    struct ovsdb_log *log;
    struct raft *raft;

    char *unbacked_name; /* Name of the unbacked storage. */

    /* All kinds of storage. */
    struct ovsdb_error *error;  /* If nonnull, a permanent error. */
    long long next_snapshot_min; /* Earliest time to take next snapshot. */
    long long next_snapshot_max; /* Latest time to take next snapshot. */

    /* Standalone only. */
    unsigned int n_read;
    unsigned int n_written;
};

static void schedule_next_snapshot(struct ovsdb_storage *, bool quick);

static struct ovsdb_error * OVS_WARN_UNUSED_RESULT
ovsdb_storage_open__(const char *filename, bool rw, bool allow_clustered,
                     struct ovsdb_storage **storagep)
{
    *storagep = NULL;

    struct ovsdb_log *log;
    struct ovsdb_error *error;
    error = ovsdb_log_open(filename, OVSDB_MAGIC"|"RAFT_MAGIC,
                           rw ? OVSDB_LOG_READ_WRITE : OVSDB_LOG_READ_ONLY,
                           -1, &log);
    if (error) {
        return error;
    }

    struct raft *raft = NULL;
    if (!strcmp(ovsdb_log_get_magic(log), RAFT_MAGIC)) {
        if (!allow_clustered) {
            ovsdb_log_close(log);
            return ovsdb_error(NULL, "%s: cannot apply this operation to "
                               "clustered database file", filename);
        }
        error = raft_open(log, &raft);
        log = NULL;
        if (error) {
            return error;
        }
    }

    struct ovsdb_storage *storage = xzalloc(sizeof *storage);
    storage->log = log;
    storage->raft = raft;
    schedule_next_snapshot(storage, false);
    *storagep = storage;
    return NULL;
}

/* Opens 'filename' for use as storage.  If 'rw', opens it for read/write
 * access, otherwise read-only.  If successful, stores the new storage in
 * '*storagep' and returns NULL; on failure, stores NULL in '*storagep' and
 * returns the error.
 *
 * The returned storage might be clustered or standalone, depending on what the
 * disk file contains. */
struct ovsdb_error * OVS_WARN_UNUSED_RESULT
ovsdb_storage_open(const char *filename, bool rw,
                   struct ovsdb_storage **storagep)
{
    return ovsdb_storage_open__(filename, rw, true, storagep);
}

struct ovsdb_storage *
ovsdb_storage_open_standalone(const char *filename, bool rw)
{
    struct ovsdb_storage *storage;
    struct ovsdb_error *error = ovsdb_storage_open__(filename, rw, false,
                                                     &storage);
    if (error) {
        ovs_fatal(0, "%s", ovsdb_error_to_string_free(error));
    }
    return storage;
}

/* Creates and returns new storage without any backing.  Nothing will be read
 * from the storage, and writes are discarded.  If 'name' is nonnull, it will
 * be used as a storage name. */
struct ovsdb_storage *
ovsdb_storage_create_unbacked(const char *name)
{
    struct ovsdb_storage *storage = xzalloc(sizeof *storage);
    schedule_next_snapshot(storage, false);
    storage->unbacked_name = nullable_xstrdup(name);
    return storage;
}

void
ovsdb_storage_close(struct ovsdb_storage *storage)
{
    if (storage) {
        ovsdb_log_close(storage->log);
        raft_close(storage->raft);
        ovsdb_error_destroy(storage->error);
        free(storage->unbacked_name);
        free(storage);
    }
}

const char *
ovsdb_storage_get_model(const struct ovsdb_storage *storage)
{
    return storage->raft ? "clustered" : "standalone";
}

bool
ovsdb_storage_is_clustered(const struct ovsdb_storage *storage)
{
    return storage->raft != NULL;
}

bool
ovsdb_storage_is_connected(const struct ovsdb_storage *storage)
{
    return !storage->raft || raft_is_connected(storage->raft);
}

bool
ovsdb_storage_is_dead(const struct ovsdb_storage *storage)
{
    return storage->raft && raft_left(storage->raft);
}

bool
ovsdb_storage_is_leader(const struct ovsdb_storage *storage)
{
    return !storage->raft || raft_is_leader(storage->raft);
}

const struct uuid *
ovsdb_storage_get_cid(const struct ovsdb_storage *storage)
{
    return storage->raft ? raft_get_cid(storage->raft) : NULL;
}

const struct uuid *
ovsdb_storage_get_sid(const struct ovsdb_storage *storage)
{
    return storage->raft ? raft_get_sid(storage->raft) : NULL;
}

uint64_t
ovsdb_storage_get_applied_index(const struct ovsdb_storage *storage)
{
    return storage->raft ? raft_get_applied_index(storage->raft) : 0;
}

void
ovsdb_storage_get_memory_usage(const struct ovsdb_storage *storage,
                               struct simap *usage)
{
    if (storage->raft) {
        raft_get_memory_usage(storage->raft, usage);
    }
}

char *
ovsdb_storage_get_error(const struct ovsdb_storage *storage)
{
    if (storage->error) {
        return ovsdb_error_to_string(storage->error);
    }

    return NULL;
}

void
ovsdb_storage_run(struct ovsdb_storage *storage)
{
    if (storage->raft) {
        raft_run(storage->raft);
    }
}

void
ovsdb_storage_wait(struct ovsdb_storage *storage)
{
    if (storage->raft) {
        raft_wait(storage->raft);
    }
}

/* Returns 'storage''s embedded name, if it has one, otherwise null.
 *
 * Only clustered storage has a built-in name.  */
const char *
ovsdb_storage_get_name(const struct ovsdb_storage *storage)
{
    return storage->unbacked_name ? storage->unbacked_name
           : storage->raft ? raft_get_name(storage->raft)
           : NULL;
}

/* Attempts to read a log record from 'storage'.
 *
 * If successful, returns NULL and stores the transaction information in
 * '*schemap', '*txnp', and '*txnid'.  At least one of these will be nonnull.
 * The caller owns the data and must eventually free it (with json_destroy()).
 *
 * If 'storage' is not clustered, 'txnid' may be null.
 *
 * If a read error occurs, returns the error and stores NULL in '*jsonp'.
 *
 * If the read reaches end of file, returns NULL and stores NULL in
 * '*jsonp'. */
struct ovsdb_error * OVS_WARN_UNUSED_RESULT
ovsdb_storage_read(struct ovsdb_storage *storage,
                   struct ovsdb_schema **schemap,
                   struct json **txnp,
                   struct uuid *txnid)
{
    *schemap = NULL;
    *txnp = NULL;
    if (txnid) {
        *txnid = UUID_ZERO;
    }

    struct json *json;
    struct json *schema_json = NULL;
    struct json *txn_json = NULL;
    if (storage->raft) {
        bool is_snapshot;
        json = json_nullable_clone(
            raft_next_entry(storage->raft, txnid, &is_snapshot));
        if (!json) {
            return NULL;
        } else if (json->type != JSON_ARRAY || json->array.n != 2) {
            json_destroy(json);
            return ovsdb_error(NULL, "invalid commit format");
        }

        struct json **e = json->array.elems;
        schema_json = e[0]->type != JSON_NULL ? e[0] : NULL;
        txn_json = e[1]->type != JSON_NULL ? e[1] : NULL;
    } else if (storage->log) {
        struct ovsdb_error *error = ovsdb_log_read(storage->log, &json);
        if (error || !json) {
            return error;
        }

        unsigned int n = storage->n_read++;
        struct json **jsonp = !n ? &schema_json : &txn_json;
        *jsonp = json;
        if (n == 1) {
            ovsdb_log_mark_base(storage->log);
        }
    } else {
        /* Unbacked.  Nothing to do. */
        return NULL;
    }

    /* If we got this far then we must have at least a schema or a
     * transaction. */
    ovs_assert(schema_json || txn_json);

    if (schema_json) {
        struct ovsdb_schema *schema;
        struct ovsdb_error *error = ovsdb_schema_from_json(schema_json,
                                                           &schema);
        if (error) {
            json_destroy(json);
            return error;
        }

        const char *storage_name = ovsdb_storage_get_name(storage);
        const char *schema_name = schema->name;
        if (storage_name && strcmp(storage_name, schema_name)) {
            error = ovsdb_error(NULL, "name %s in header does not match "
                                "name %s in schema",
                                storage_name, schema_name);
            json_destroy(json);
            ovsdb_schema_destroy(schema);
            return error;
        }

        *schemap = schema;
    }

    if (txn_json) {
        *txnp = json_clone(txn_json);
    }

    json_destroy(json);
    return NULL;
}

/* Reads and returns the schema from standalone storage 'storage'.  Terminates
 * with an error on failure. */
struct ovsdb_schema *
ovsdb_storage_read_schema(struct ovsdb_storage *storage)
{
    ovs_assert(storage->log);

    struct json *txn_json;
    struct ovsdb_schema *schema;
    struct ovsdb_error *error = ovsdb_storage_read(storage, &schema,
                                                   &txn_json, NULL);
    if (error) {
        ovs_fatal(0, "%s", ovsdb_error_to_string_free(error));
    }
    if (!schema && !txn_json) {
        ovs_fatal(0, "unexpected end of file reading schema");
    }
    ovs_assert(schema && !txn_json);

    return schema;
}

bool
ovsdb_storage_read_wait(struct ovsdb_storage *storage)
{
    return (storage->raft
            ? raft_has_next_entry(storage->raft)
            : false);
}

void
ovsdb_storage_unread(struct ovsdb_storage *storage)
{
    if (storage->error) {
        return;
    }

    if (storage->raft) {
        if (!storage->error) {
            storage->error = ovsdb_error(NULL, "inconsistent data");
        }
    } else if (storage->log) {
        ovsdb_log_unread(storage->log);
    }
}

struct ovsdb_write {
    struct ovsdb_error *error;
    struct raft_command *command;
};

/* Not suitable for writing transactions that change the schema. */
struct ovsdb_write * OVS_WARN_UNUSED_RESULT
ovsdb_storage_write(struct ovsdb_storage *storage, const struct json *data,
                    const struct uuid *prereq, struct uuid *resultp,
                    bool durable)
{
    struct ovsdb_write *w = xzalloc(sizeof *w);
    struct uuid result = UUID_ZERO;
    if (storage->error) {
        w->error = ovsdb_error_clone(storage->error);
    } else if (storage->raft) {
        struct json *txn_json = json_array_create_2(json_null_create(),
                                                    json_clone(data));
        w->command = raft_command_execute(storage->raft, txn_json,
                                          prereq, &result);
        json_destroy(txn_json);
    } else if (storage->log) {
        w->error = ovsdb_log_write(storage->log, data);
        if (!w->error) {
            storage->n_written++;
            if (durable) {
                w->error = ovsdb_log_commit_block(storage->log);
            }
        }
    } else {
        /* When 'error' and 'command' are both null, it indicates that the
         * command is complete.  This is fine since this unbacked storage drops
         * writes. */
    }
    if (resultp) {
        *resultp = result;
    }
    return w;
}

/* Not suitable for writing transactions that change the schema. */
struct ovsdb_error * OVS_WARN_UNUSED_RESULT
ovsdb_storage_write_block(struct ovsdb_storage *storage,
                          const struct json *data, const struct uuid *prereq,
                          struct uuid *resultp, bool durable)
{
    struct ovsdb_write *w = ovsdb_storage_write(storage, data,
                                                prereq, resultp, durable);
    while (!ovsdb_write_is_complete(w)) {
        if (storage->raft) {
            raft_run(storage->raft);
        }

        ovsdb_write_wait(w);
        if (storage->raft) {
            raft_wait(storage->raft);
        }
        poll_block();
    }

    struct ovsdb_error *error = ovsdb_error_clone(ovsdb_write_get_error(w));
    ovsdb_write_destroy(w);
    return error;
}

bool
ovsdb_write_is_complete(const struct ovsdb_write *w)
{
    return (w->error
            || !w->command
            || raft_command_get_status(w->command) != RAFT_CMD_INCOMPLETE);
}

const struct ovsdb_error *
ovsdb_write_get_error(const struct ovsdb_write *w_)
{
    struct ovsdb_write *w = CONST_CAST(struct ovsdb_write *, w_);
    ovs_assert(ovsdb_write_is_complete(w));

    if (w->command && !w->error) {
        enum raft_command_status status = raft_command_get_status(w->command);
        if (status != RAFT_CMD_SUCCESS) {
            w->error = ovsdb_error("cluster error", "%s",
                                   raft_command_status_to_string(status));
        }
    }

    return w->error;
}

uint64_t
ovsdb_write_get_commit_index(const struct ovsdb_write *w)
{
    ovs_assert(ovsdb_write_is_complete(w));
    return (w->command && !w->error
            ? raft_command_get_commit_index(w->command)
            : 0);
}

void
ovsdb_write_wait(const struct ovsdb_write *w)
{
    if (ovsdb_write_is_complete(w)) {
        poll_immediate_wake();
    }
}

void
ovsdb_write_destroy(struct ovsdb_write *w)
{
    if (w) {
        raft_command_unref(w->command);
        ovsdb_error_destroy(w->error);
        free(w);
    }
}

static void
schedule_next_snapshot(struct ovsdb_storage *storage, bool quick)
{
    if (storage->log || storage->raft) {
        unsigned int base = 10 * 60 * 1000;  /* 10 minutes */
        unsigned int range = 10 * 60 * 1000; /* 10 minutes */
        if (quick) {
            base /= 10;
            range /= 10;
        }

        long long int now = time_msec();
        storage->next_snapshot_min = now + base + random_range(range);
        storage->next_snapshot_max = now + 60LL * 60 * 24 * 1000; /* 1 day */
    } else {
        storage->next_snapshot_min = LLONG_MAX;
        storage->next_snapshot_max = LLONG_MAX;
    }
}

bool
ovsdb_storage_should_snapshot(const struct ovsdb_storage *storage)
{
    if (storage->raft || storage->log) {
        /* If we haven't reached the minimum snapshot time, don't snapshot. */
        long long int now = time_msec();
        if (now < storage->next_snapshot_min) {
            return false;
        }

        uint64_t log_len = (storage->raft
                            ? raft_get_log_length(storage->raft)
                            : storage->n_read + storage->n_written);
        bool snapshot_recommended = false;

        if (now < storage->next_snapshot_max) {
            /* Maximum snapshot time not yet reached.  Take a snapshot if there
             * have been at least 100 log entries and the log file size has
             * grown a lot. */
            bool grew_lots = (storage->raft
                              ? raft_grew_lots(storage->raft)
                              : ovsdb_log_grew_lots(storage->log));
            snapshot_recommended = (log_len >= 100 && grew_lots);
        } else {
            /* We have reached the maximum snapshot time.  Take a snapshot if
             * there have been any log entries at all. */
            snapshot_recommended = (log_len > 0);
        }

        if (!snapshot_recommended) {
            return false;
        }

        /* If we can't snapshot right now, don't. */
        if (storage->raft && !raft_may_snapshot(storage->raft)) {
            /* Notifying the storage that it needs to make a snapshot soon. */
            raft_notify_snapshot_recommended(storage->raft);
            return false;
        }

        return true;
    }

    return false;
}

static struct ovsdb_error * OVS_WARN_UNUSED_RESULT
ovsdb_storage_store_snapshot__(struct ovsdb_storage *storage,
                               const struct json *schema,
                               const struct json *data)
{
    if (storage->raft) {
        struct json *entries = json_array_create_empty();
        if (schema) {
            json_array_add(entries, json_clone(schema));
        }
        if (data) {
            json_array_add(entries, json_clone(data));
        }
        struct ovsdb_error *error = raft_store_snapshot(storage->raft,
                                                        entries);
        json_destroy(entries);
        return error;
    } else if (storage->log) {
        struct json *entries[2];
        size_t n = 0;
        if (schema) {
            entries[n++] = CONST_CAST(struct json *, schema);
        }
        if (data) {
            entries[n++] = CONST_CAST(struct json *, data);
        }
        return ovsdb_log_replace(storage->log, entries, n);
    } else {
        return NULL;
    }
}

/* 'schema' and 'data' should faithfully represent the current schema and data,
 * otherwise the two storing backing formats will yield divergent results.  Use
 * ovsdb_storage_write_schema_change() to change the schema. */
struct ovsdb_error * OVS_WARN_UNUSED_RESULT
ovsdb_storage_store_snapshot(struct ovsdb_storage *storage,
                             const struct json *schema,
                             const struct json *data)
{
    struct ovsdb_error *error = ovsdb_storage_store_snapshot__(storage,
                                                               schema, data);
    bool retry_quickly = error != NULL;
    schedule_next_snapshot(storage, retry_quickly);
    return error;
}

struct ovsdb_write * OVS_WARN_UNUSED_RESULT
ovsdb_storage_write_schema_change(struct ovsdb_storage *storage,
                                  const struct json *schema,
                                  const struct json *data,
                                  const struct uuid *prereq,
                                  struct uuid *resultp)
{
    struct ovsdb_write *w = xzalloc(sizeof *w);
    struct uuid result = UUID_ZERO;
    if (storage->error) {
        w->error = ovsdb_error_clone(storage->error);
    } else if (storage->raft) {
        struct json *txn_json = json_array_create_2(json_clone(schema),
                                                    json_clone(data));
        w->command = raft_command_execute(storage->raft, txn_json,
                                          prereq, &result);
        json_destroy(txn_json);
    } else if (storage->log) {
        w->error = ovsdb_storage_store_snapshot__(storage, schema, data);
    } else {
        /* When 'error' and 'command' are both null, it indicates that the
         * command is complete.  This is fine since this unbacked storage drops
         * writes. */
    }
    if (resultp) {
        *resultp = result;
    }
    return w;
}

const struct uuid *
ovsdb_storage_peek_last_eid(struct ovsdb_storage *storage)
{
    if (!storage->raft) {
        return NULL;
    }
    return raft_current_eid(storage->raft);
}