aof.c 48.5 KB
Newer Older
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
/*
 * Copyright (c) 2009-2012, Salvatore Sanfilippo <antirez at gmail dot com>
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *
 *   * Redistributions of source code must retain the above copyright notice,
 *     this list of conditions and the following disclaimer.
 *   * Redistributions in binary form must reproduce the above copyright
 *     notice, this list of conditions and the following disclaimer in the
 *     documentation and/or other materials provided with the distribution.
 *   * Neither the name of Redis nor the names of its contributors may be used
 *     to endorse or promote products derived from this software without
 *     specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 * POSSIBILITY OF SUCH DAMAGE.
 */

30
#include "redis.h"
31
#include "bio.h"
A
antirez 已提交
32
#include "rio.h"
33 34 35 36

#include <signal.h>
#include <fcntl.h>
#include <sys/stat.h>
37 38 39 40
#include <sys/types.h>
#include <sys/time.h>
#include <sys/resource.h>
#include <sys/wait.h>
41

42 43
void aofUpdateCurrentSize(void);

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
/* ----------------------------------------------------------------------------
 * AOF rewrite buffer implementation.
 *
 * The following code implement a simple buffer used in order to accumulate
 * changes while the background process is rewriting the AOF file.
 *
 * We only need to append, but can't just use realloc with a large block
 * because 'huge' reallocs are not always handled as one could expect
 * (via remapping of pages at OS level) but may involve copying data.
 *
 * For this reason we use a list of blocks, every block is
 * AOF_RW_BUF_BLOCK_SIZE bytes.
 * ------------------------------------------------------------------------- */

#define AOF_RW_BUF_BLOCK_SIZE (1024*1024*10)    /* 10 MB per block */

typedef struct aofrwblock {
    unsigned long used, free;
    char buf[AOF_RW_BUF_BLOCK_SIZE];
} aofrwblock;

/* This function free the old AOF rewrite buffer if needed, and initialize
 * a fresh new one. It tests for server.aof_rewrite_buf_blocks equal to NULL
 * so can be used for the first initialization as well. */
void aofRewriteBufferReset(void) {
    if (server.aof_rewrite_buf_blocks)
        listRelease(server.aof_rewrite_buf_blocks);

    server.aof_rewrite_buf_blocks = listCreate();
    listSetFreeMethod(server.aof_rewrite_buf_blocks,zfree);
}

/* Return the current size of the AOF rerwite buffer. */
unsigned long aofRewriteBufferSize(void) {
    listNode *ln = listLast(server.aof_rewrite_buf_blocks);
    aofrwblock *block = ln ? ln->value : NULL;

    if (block == NULL) return 0;
    unsigned long size =
        (listLength(server.aof_rewrite_buf_blocks)-1) * AOF_RW_BUF_BLOCK_SIZE;
    size += block->used;
    return size;
}

/* Append data to the AOF rewrite buffer, allocating new blocks if needed. */
void aofRewriteBufferAppend(unsigned char *s, unsigned long len) {
    listNode *ln = listLast(server.aof_rewrite_buf_blocks);
    aofrwblock *block = ln ? ln->value : NULL;

    while(len) {
        /* If we already got at least an allocated block, try appending
         * at least some piece into it. */
        if (block) {
            unsigned long thislen = (block->free < len) ? block->free : len;
            if (thislen) {  /* The current block is not already full. */
                memcpy(block->buf+block->used, s, thislen);
                block->used += thislen;
                block->free -= thislen;
                s += thislen;
                len -= thislen;
            }
        }

        if (len) { /* First block to allocate, or need another block. */
            int numblocks;

            block = zmalloc(sizeof(*block));
            block->free = AOF_RW_BUF_BLOCK_SIZE;
            block->used = 0;
            listAddNodeTail(server.aof_rewrite_buf_blocks,block);

            /* Log every time we cross more 10 or 100 blocks, respectively
             * as a notice or warning. */
            numblocks = listLength(server.aof_rewrite_buf_blocks);
            if (((numblocks+1) % 10) == 0) {
                int level = ((numblocks+1) % 100) == 0 ? REDIS_WARNING :
                                                         REDIS_NOTICE;
                redisLog(level,"Background AOF buffer size: %lu MB",
                    aofRewriteBufferSize()/(1024*1024));
            }
        }
    }
}

/* Write the buffer (possibly composed of multiple blocks) into the specified
129
 * fd. If a short write or any other error happens -1 is returned,
130 131 132 133 134 135 136 137 138 139 140 141 142
 * otherwise the number of bytes written is returned. */
ssize_t aofRewriteBufferWrite(int fd) {
    listNode *ln;
    listIter li;
    ssize_t count = 0;

    listRewind(server.aof_rewrite_buf_blocks,&li);
    while((ln = listNext(&li))) {
        aofrwblock *block = listNodeValue(ln);
        ssize_t nwritten;

        if (block->used) {
            nwritten = write(fd,block->buf,block->used);
143
            if (nwritten != (ssize_t)block->used) {
144 145 146 147 148 149 150 151 152
                if (nwritten == 0) errno = EIO;
                return -1;
            }
            count += nwritten;
        }
    }
    return count;
}

153 154 155 156 157 158
/* ----------------------------------------------------------------------------
 * AOF file implementation
 * ------------------------------------------------------------------------- */

/* Starts a background task that performs fsync() against the specified
 * file descriptor (the one of the AOF file) in another thread. */
159
void aof_background_fsync(int fd) {
A
antirez 已提交
160
    bioCreateBackgroundJob(REDIS_BIO_AOF_FSYNC,(void*)(long)fd,NULL,NULL);
161 162
}

163 164 165
/* Called when the user switches from "appendonly yes" to "appendonly no"
 * at runtime using the CONFIG command. */
void stopAppendOnly(void) {
166
    redisAssert(server.aof_state != REDIS_AOF_OFF);
167
    flushAppendOnlyFile(1);
A
antirez 已提交
168 169
    aof_fsync(server.aof_fd);
    close(server.aof_fd);
170

A
antirez 已提交
171 172
    server.aof_fd = -1;
    server.aof_selected_db = -1;
173
    server.aof_state = REDIS_AOF_OFF;
174
    /* rewrite operation in progress? kill it, wait child exit */
A
antirez 已提交
175
    if (server.aof_child_pid != -1) {
176 177
        int statloc;

178 179
        redisLog(REDIS_NOTICE,"Killing running AOF rewrite child: %ld",
            (long) server.aof_child_pid);
180
        if (kill(server.aof_child_pid,SIGUSR1) != -1)
181 182
            wait3(&statloc,0,NULL);
        /* reset the buffer accumulating changes while the child saves */
183
        aofRewriteBufferReset();
A
antirez 已提交
184 185
        aofRemoveTempFile(server.aof_child_pid);
        server.aof_child_pid = -1;
186
        server.aof_rewrite_time_start = -1;
187 188 189 190 191 192
    }
}

/* Called when the user switches from "appendonly no" to "appendonly yes"
 * at runtime using the CONFIG command. */
int startAppendOnly(void) {
193
    server.aof_last_fsync = server.unixtime;
A
antirez 已提交
194
    server.aof_fd = open(server.aof_filename,O_WRONLY|O_APPEND|O_CREAT,0644);
195
    redisAssert(server.aof_state == REDIS_AOF_OFF);
A
antirez 已提交
196
    if (server.aof_fd == -1) {
197
        redisLog(REDIS_WARNING,"Redis needs to enable the AOF but can't open the append only file: %s",strerror(errno));
198 199 200
        return REDIS_ERR;
    }
    if (rewriteAppendOnlyFileBackground() == REDIS_ERR) {
A
antirez 已提交
201
        close(server.aof_fd);
202
        redisLog(REDIS_WARNING,"Redis needs to enable the AOF but can't trigger a background AOF rewrite operation. Check the above logs for more info about the error.");
203 204
        return REDIS_ERR;
    }
205 206
    /* We correctly switched on AOF, now wait for the rerwite to be complete
     * in order to append data on disk. */
207
    server.aof_state = REDIS_AOF_WAIT_REWRITE;
208 209 210 211 212 213 214 215 216
    return REDIS_OK;
}

/* Write the append only file buffer on disk.
 *
 * Since we are required to write the AOF before replying to the client,
 * and the only way the client socket can get a write is entering when the
 * the event loop, we accumulate all the AOF writes in a memory
 * buffer and write it on disk using this function just before entering
217 218 219 220 221 222 223 224 225 226 227 228
 * the event loop again.
 *
 * About the 'force' argument:
 *
 * When the fsync policy is set to 'everysec' we may delay the flush if there
 * is still an fsync() going on in the background thread, since for instance
 * on Linux write(2) will be blocked by the background fsync anyway.
 * When this happens we remember that there is some aof buffer to be
 * flushed ASAP, and will try to do that in the serverCron() function.
 *
 * However if force is set to 1 we'll write regardless of the background
 * fsync. */
229
#define AOF_WRITE_LOG_ERROR_RATE 30 /* Seconds between errors logging. */
230
void flushAppendOnlyFile(int force) {
231
    ssize_t nwritten;
232
    int sync_in_progress = 0;
233
    mstime_t latency;
234

A
antirez 已提交
235
    if (sdslen(server.aof_buf) == 0) return;
236

237
    if (server.aof_fsync == AOF_FSYNC_EVERYSEC)
238 239
        sync_in_progress = bioPendingJobsOfType(REDIS_BIO_AOF_FSYNC) != 0;

240
    if (server.aof_fsync == AOF_FSYNC_EVERYSEC && !force) {
241 242 243 244 245 246 247 248 249 250
        /* With this append fsync policy we do background fsyncing.
         * If the fsync is still in progress we can try to delay
         * the write for a couple of seconds. */
        if (sync_in_progress) {
            if (server.aof_flush_postponed_start == 0) {
                /* No previous write postponinig, remember that we are
                 * postponing the flush and return. */
                server.aof_flush_postponed_start = server.unixtime;
                return;
            } else if (server.unixtime - server.aof_flush_postponed_start < 2) {
A
antirez 已提交
251
                /* We were already waiting for fsync to finish, but for less
252 253 254 255 256
                 * than two seconds this is still ok. Postpone again. */
                return;
            }
            /* Otherwise fall trough, and go write since we can't wait
             * over two seconds. */
257
            server.aof_delayed_fsync++;
258
            redisLog(REDIS_NOTICE,"Asynchronous AOF fsync is taking too long (disk is busy?). Writing the AOF buffer without waiting for fsync to complete, this may slow down Redis.");
259 260
        }
    }
261 262 263 264 265
    /* We want to perform a single write. This should be guaranteed atomic
     * at least if the filesystem we are writing is a real physical one.
     * While this will save us against the server being killed I don't think
     * there is much to do about the whole server stopping for power problems
     * or alike */
266

267
    latencyStartMonitor(latency);
A
antirez 已提交
268
    nwritten = write(server.aof_fd,server.aof_buf,sdslen(server.aof_buf));
269
    latencyEndMonitor(latency);
270 271 272 273 274
    /* We want to capture different events for delayed writes:
     * when the delay happens with a pending fsync, or with a saving child
     * active, and when the above two conditions are missing.
     * We also use an additional event name to save all samples which is
     * useful for graphing / monitoring purposes. */
275
    if (sync_in_progress) {
276 277 278 279 280 281
        latencyAddSampleIfNeeded("aof-write-pending-fsync",latency);
    } else if (server.aof_child_pid != -1 || server.rdb_child_pid != -1) {
        latencyAddSampleIfNeeded("aof-write-active-child",latency);
    } else {
        latencyAddSampleIfNeeded("aof-write-alone",latency);
    }
282 283
    latencyAddSampleIfNeeded("aof-write",latency);

284 285 286
    /* We performed the write so reset the postponed flush sentinel to zero. */
    server.aof_flush_postponed_start = 0;

A
antirez 已提交
287
    if (nwritten != (signed)sdslen(server.aof_buf)) {
288 289 290 291 292 293 294 295 296 297
        static time_t last_write_error_log = 0;
        int can_log = 0;

        /* Limit logging rate to 1 line per AOF_WRITE_LOG_ERROR_RATE seconds. */
        if ((server.unixtime - last_write_error_log) > AOF_WRITE_LOG_ERROR_RATE) {
            can_log = 1;
            last_write_error_log = server.unixtime;
        }

        /* Lof the AOF write error and record the error code. */
P
Pieter Noordhuis 已提交
298
        if (nwritten == -1) {
299 300 301 302 303
            if (can_log) {
                redisLog(REDIS_WARNING,"Error writing to the AOF file: %s",
                    strerror(errno));
                server.aof_last_write_errno = errno;
            }
P
Pieter Noordhuis 已提交
304
        } else {
305 306 307 308 309 310 311
            if (can_log) {
                redisLog(REDIS_WARNING,"Short write while writing to "
                                       "the AOF file: (nwritten=%lld, "
                                       "expected=%lld)",
                                       (long long)nwritten,
                                       (long long)sdslen(server.aof_buf));
            }
312 313

            if (ftruncate(server.aof_fd, server.aof_current_size) == -1) {
314 315 316 317 318 319 320 321 322 323
                if (can_log) {
                    redisLog(REDIS_WARNING, "Could not remove short write "
                             "from the append-only file.  Redis may refuse "
                             "to load the AOF the next time it starts.  "
                             "ftruncate: %s", strerror(errno));
                }
            } else {
                /* If the ftrunacate() succeeded we can set nwritten to
                 * -1 since there is no longer partial data into the AOF. */
                nwritten = -1;
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
            server.aof_last_write_errno = ENOSPC;
        }

        /* Handle the AOF write error. */
        if (server.aof_fsync == AOF_FSYNC_ALWAYS) {
            /* We can't recover when the fsync policy is ALWAYS since the
             * reply for the client is already in the output buffers, and we
             * have the contract with the user that on acknowledged write data
             * is synched on disk. */
            redisLog(REDIS_WARNING,"Can't recover from AOF write error when the AOF fsync policy is 'always'. Exiting...");
            exit(1);
        } else {
            /* Recover from failed write leaving data into the buffer. However
             * set an error to stop accepting writes as long as the error
             * condition is not cleared. */
            server.aof_last_write_status = REDIS_ERR;

            /* Trim the sds buffer if there was a partial write, and there
             * was no way to undo it with ftruncate(2). */
            if (nwritten > 0) {
                server.aof_current_size += nwritten;
                sdsrange(server.aof_buf,nwritten,-1);
            }
            return; /* We'll try again on the next call... */
        }
    } else {
        /* Successful write(2). If AOF was in error state, restore the
         * OK state and log the event. */
        if (server.aof_last_write_status == REDIS_ERR) {
            redisLog(REDIS_WARNING,
                "AOF write error looks solved, Redis can write again.");
            server.aof_last_write_status = REDIS_OK;
P
Pieter Noordhuis 已提交
357
        }
358
    }
359
    server.aof_current_size += nwritten;
360

361 362
    /* Re-use AOF buffer when it is small enough. The maximum comes from the
     * arena size of 4k minus some overhead (but is otherwise arbitrary). */
A
antirez 已提交
363 364
    if ((sdslen(server.aof_buf)+sdsavail(server.aof_buf)) < 4000) {
        sdsclear(server.aof_buf);
365
    } else {
A
antirez 已提交
366 367
        sdsfree(server.aof_buf);
        server.aof_buf = sdsempty();
368 369
    }

370 371
    /* Don't fsync if no-appendfsync-on-rewrite is set to yes and there are
     * children doing I/O in the background. */
372
    if (server.aof_no_fsync_on_rewrite &&
A
antirez 已提交
373
        (server.aof_child_pid != -1 || server.rdb_child_pid != -1))
374
            return;
375 376

    /* Perform the fsync if needed. */
377
    if (server.aof_fsync == AOF_FSYNC_ALWAYS) {
378 379
        /* aof_fsync is defined as fdatasync() for Linux in order to avoid
         * flushing metadata. */
380
        latencyStartMonitor(latency);
A
antirez 已提交
381
        aof_fsync(server.aof_fd); /* Let's try to get this data on the disk */
382 383
        latencyEndMonitor(latency);
        latencyAddSampleIfNeeded("aof-fsync-always",latency);
A
antirez 已提交
384
        server.aof_last_fsync = server.unixtime;
385
    } else if ((server.aof_fsync == AOF_FSYNC_EVERYSEC &&
A
antirez 已提交
386 387 388
                server.unixtime > server.aof_last_fsync)) {
        if (!sync_in_progress) aof_background_fsync(server.aof_fd);
        server.aof_last_fsync = server.unixtime;
389 390 391
    }
}

392 393 394 395 396 397 398 399 400 401 402
sds catAppendOnlyGenericCommand(sds dst, int argc, robj **argv) {
    char buf[32];
    int len, j;
    robj *o;

    buf[0] = '*';
    len = 1+ll2string(buf+1,sizeof(buf)-1,argc);
    buf[len++] = '\r';
    buf[len++] = '\n';
    dst = sdscatlen(dst,buf,len);

403
    for (j = 0; j < argc; j++) {
404 405 406 407 408 409 410 411
        o = getDecodedObject(argv[j]);
        buf[0] = '$';
        len = 1+ll2string(buf+1,sizeof(buf)-1,sdslen(o->ptr));
        buf[len++] = '\r';
        buf[len++] = '\n';
        dst = sdscatlen(dst,buf,len);
        dst = sdscatlen(dst,o->ptr,sdslen(o->ptr));
        dst = sdscatlen(dst,"\r\n",2);
412 413
        decrRefCount(o);
    }
414
    return dst;
415 416
}

417 418 419 420 421 422 423 424 425
/* Create the sds representation of an PEXPIREAT command, using
 * 'seconds' as time to live and 'cmd' to understand what command
 * we are translating into a PEXPIREAT.
 *
 * This command is used in order to translate EXPIRE and PEXPIRE commands
 * into PEXPIREAT command so that we retain precision in the append only
 * file, and the time is always absolute and not relative. */
sds catAppendOnlyExpireAtCommand(sds buf, struct redisCommand *cmd, robj *key, robj *seconds) {
    long long when;
426 427 428 429
    robj *argv[3];

    /* Make sure we can use strtol */
    seconds = getDecodedObject(seconds);
430 431 432 433 434 435 436 437 438 439 440 441 442
    when = strtoll(seconds->ptr,NULL,10);
    /* Convert argument into milliseconds for EXPIRE, SETEX, EXPIREAT */
    if (cmd->proc == expireCommand || cmd->proc == setexCommand ||
        cmd->proc == expireatCommand)
    {
        when *= 1000;
    }
    /* Convert into absolute time for EXPIRE, PEXPIRE, SETEX, PSETEX */
    if (cmd->proc == expireCommand || cmd->proc == pexpireCommand ||
        cmd->proc == setexCommand || cmd->proc == psetexCommand)
    {
        when += mstime();
    }
443 444
    decrRefCount(seconds);

445
    argv[0] = createStringObject("PEXPIREAT",9);
446
    argv[1] = key;
447 448
    argv[2] = createStringObjectFromLongLong(when);
    buf = catAppendOnlyGenericCommand(buf, 3, argv);
449 450 451 452 453 454
    decrRefCount(argv[0]);
    decrRefCount(argv[2]);
    return buf;
}

void feedAppendOnlyFile(struct redisCommand *cmd, int dictid, robj **argv, int argc) {
455
    sds buf = sdsempty();
456 457
    robj *tmpargv[3];

G
guiquanz 已提交
458
    /* The DB this command was targeting is not the same as the last command
459
     * we appendend. To issue a SELECT command is needed. */
A
antirez 已提交
460
    if (dictid != server.aof_selected_db) {
461 462 463 464 465
        char seldb[64];

        snprintf(seldb,sizeof(seldb),"%d",dictid);
        buf = sdscatprintf(buf,"*2\r\n$6\r\nSELECT\r\n$%lu\r\n%s\r\n",
            (unsigned long)strlen(seldb),seldb);
A
antirez 已提交
466
        server.aof_selected_db = dictid;
467 468
    }

469 470 471 472 473 474
    if (cmd->proc == expireCommand || cmd->proc == pexpireCommand ||
        cmd->proc == expireatCommand) {
        /* Translate EXPIRE/PEXPIRE/EXPIREAT into PEXPIREAT */
        buf = catAppendOnlyExpireAtCommand(buf,cmd,argv[1],argv[2]);
    } else if (cmd->proc == setexCommand || cmd->proc == psetexCommand) {
        /* Translate SETEX/PSETEX to SET and PEXPIREAT */
475 476 477 478 479
        tmpargv[0] = createStringObject("SET",3);
        tmpargv[1] = argv[1];
        tmpargv[2] = argv[3];
        buf = catAppendOnlyGenericCommand(buf,3,tmpargv);
        decrRefCount(tmpargv[0]);
480
        buf = catAppendOnlyExpireAtCommand(buf,cmd,argv[1],argv[2]);
481
    } else {
482 483 484
        /* All the other commands don't need translation or need the
         * same translation already operated in the command vector
         * for the replication itself. */
485 486 487 488 489
        buf = catAppendOnlyGenericCommand(buf,argc,argv);
    }

    /* Append to the AOF buffer. This will be flushed on disk just before
     * of re-entering the event loop, so before the client will get a
490 491
     * positive reply about the operation performed. */
    if (server.aof_state == REDIS_AOF_ON)
A
antirez 已提交
492
        server.aof_buf = sdscatlen(server.aof_buf,buf,sdslen(buf));
493 494 495 496 497

    /* If a background append only file rewriting is in progress we want to
     * accumulate the differences between the child DB and the current one
     * in a buffer, so that when the child process will do its work we
     * can append the differences to the new append only file. */
A
antirez 已提交
498
    if (server.aof_child_pid != -1)
499
        aofRewriteBufferAppend((unsigned char*)buf,sdslen(buf));
500 501 502 503

    sdsfree(buf);
}

504 505 506 507
/* ----------------------------------------------------------------------------
 * AOF loading
 * ------------------------------------------------------------------------- */

508 509 510 511 512 513 514
/* In Redis commands are always executed in the context of a client, so in
 * order to load the append only file we need to create a fake client. */
struct redisClient *createFakeClient(void) {
    struct redisClient *c = zmalloc(sizeof(*c));

    selectDb(c,0);
    c->fd = -1;
515
    c->name = NULL;
516
    c->querybuf = sdsempty();
517
    c->querybuf_peak = 0;
518 519
    c->argc = 0;
    c->argv = NULL;
520
    c->bufpos = 0;
521 522 523 524 525
    c->flags = 0;
    /* We set the fake client as a slave waiting for the synchronization
     * so that Redis will not try to send replies to this client. */
    c->replstate = REDIS_REPL_WAIT_BGSAVE_START;
    c->reply = listCreate();
526
    c->reply_bytes = 0;
527
    c->obuf_soft_limit_reached_time = 0;
528
    c->watched_keys = listCreate();
529
    c->peerid = NULL;
530
    listSetFreeMethod(c->reply,decrRefCountVoid);
531 532 533 534 535 536 537 538
    listSetDupMethod(c->reply,dupClientReplyValue);
    initClientMultiState(c);
    return c;
}

void freeFakeClient(struct redisClient *c) {
    sdsfree(c->querybuf);
    listRelease(c->reply);
539
    listRelease(c->watched_keys);
540 541 542 543 544 545 546 547 548 549 550
    freeClientMultiState(c);
    zfree(c);
}

/* Replay the append log file. On error REDIS_OK is returned. On non fatal
 * error (the append only file is zero-length) REDIS_ERR is returned. On
 * fatal error an error message is logged and the program exists. */
int loadAppendOnlyFile(char *filename) {
    struct redisClient *fakeClient;
    FILE *fp = fopen(filename,"r");
    struct redis_stat sb;
551
    int old_aof_state = server.aof_state;
552
    long loops = 0;
553

A
antirez 已提交
554
    if (fp && redis_fstat(fileno(fp),&sb) != -1 && sb.st_size == 0) {
555
        server.aof_current_size = 0;
A
antirez 已提交
556
        fclose(fp);
557
        return REDIS_ERR;
A
antirez 已提交
558
    }
559 560 561 562 563 564 565 566

    if (fp == NULL) {
        redisLog(REDIS_WARNING,"Fatal error: can't open the append log file for reading: %s",strerror(errno));
        exit(1);
    }

    /* Temporarily disable AOF, to prevent EXEC from feeding a MULTI
     * to the same file we're about to read. */
567
    server.aof_state = REDIS_AOF_OFF;
568 569

    fakeClient = createFakeClient();
570 571
    startLoading(fp);

572 573 574 575 576 577 578 579
    while(1) {
        int argc, j;
        unsigned long len;
        robj **argv;
        char buf[128];
        sds argsds;
        struct redisCommand *cmd;

580 581 582
        /* Serve the clients from time to time */
        if (!(loops++ % 1000)) {
            loadingProgress(ftello(fp));
583
            processEventsWhileBlocked();
584 585
        }

586 587 588 589 590 591 592 593
        if (fgets(buf,sizeof(buf),fp) == NULL) {
            if (feof(fp))
                break;
            else
                goto readerr;
        }
        if (buf[0] != '*') goto fmterr;
        argc = atoi(buf+1);
594 595
        if (argc < 1) goto fmterr;

596 597 598 599 600 601 602 603 604 605 606 607 608 609
        argv = zmalloc(sizeof(robj*)*argc);
        for (j = 0; j < argc; j++) {
            if (fgets(buf,sizeof(buf),fp) == NULL) goto readerr;
            if (buf[0] != '$') goto fmterr;
            len = strtol(buf+1,NULL,10);
            argsds = sdsnewlen(NULL,len);
            if (len && fread(argsds,len,1,fp) == 0) goto fmterr;
            argv[j] = createObject(REDIS_STRING,argsds);
            if (fread(buf,2,1,fp) == 0) goto fmterr; /* discard CRLF */
        }

        /* Command lookup */
        cmd = lookupCommand(argv[0]->ptr);
        if (!cmd) {
610
            redisLog(REDIS_WARNING,"Unknown command '%s' reading the append only file", (char*)argv[0]->ptr);
611 612 613 614 615 616
            exit(1);
        }
        /* Run the command in the context of a fake client */
        fakeClient->argc = argc;
        fakeClient->argv = argv;
        cmd->proc(fakeClient);
617 618 619

        /* The fake client should not have a reply */
        redisAssert(fakeClient->bufpos == 0 && listLength(fakeClient->reply) == 0);
620 621
        /* The fake client should never get blocked */
        redisAssert((fakeClient->flags & REDIS_BLOCKED) == 0);
622

623 624 625 626 627
        /* Clean up. Command code may have changed argv/argc so we use the
         * argv/argc of the client instead of the local variables. */
        for (j = 0; j < fakeClient->argc; j++)
            decrRefCount(fakeClient->argv[j]);
        zfree(fakeClient->argv);
628 629 630 631 632 633 634 635
    }

    /* This point can only be reached when EOF is reached without errors.
     * If the client is in the middle of a MULTI/EXEC, log error and quit. */
    if (fakeClient->flags & REDIS_MULTI) goto readerr;

    fclose(fp);
    freeFakeClient(fakeClient);
636
    server.aof_state = old_aof_state;
637
    stopLoading();
638
    aofUpdateCurrentSize();
639
    server.aof_rewrite_base_size = server.aof_current_size;
640 641 642 643 644 645 646 647 648 649
    return REDIS_OK;

readerr:
    if (feof(fp)) {
        redisLog(REDIS_WARNING,"Unexpected end of file reading the append only file");
    } else {
        redisLog(REDIS_WARNING,"Unrecoverable error reading the append only file: %s", strerror(errno));
    }
    exit(1);
fmterr:
A
antirez 已提交
650
    redisLog(REDIS_WARNING,"Bad file format reading the append only file: make a backup of your AOF file, then use ./redis-check-aof --fix <filename>");
651 652 653
    exit(1);
}

654 655 656 657
/* ----------------------------------------------------------------------------
 * AOF rewrite
 * ------------------------------------------------------------------------- */

P
Pieter Noordhuis 已提交
658 659 660 661 662 663 664 665 666 667 668 669 670 671
/* Delegate writing an object to writing a bulk string or bulk long long.
 * This is not placed in rio.c since that adds the redis.h dependency. */
int rioWriteBulkObject(rio *r, robj *obj) {
    /* Avoid using getDecodedObject to help copy-on-write (we are often
     * in a child process when this function is called). */
    if (obj->encoding == REDIS_ENCODING_INT) {
        return rioWriteBulkLongLong(r,(long)obj->ptr);
    } else if (obj->encoding == REDIS_ENCODING_RAW) {
        return rioWriteBulkString(r,obj->ptr,sdslen(obj->ptr));
    } else {
        redisPanic("Unknown string encoding");
    }
}

672 673 674 675 676 677 678 679 680 681 682 683 684 685
/* Emit the commands needed to rebuild a list object.
 * The function returns 0 on error, 1 on success. */
int rewriteListObject(rio *r, robj *key, robj *o) {
    long long count = 0, items = listTypeLength(o);

    if (o->encoding == REDIS_ENCODING_ZIPLIST) {
        unsigned char *zl = o->ptr;
        unsigned char *p = ziplistIndex(zl,0);
        unsigned char *vstr;
        unsigned int vlen;
        long long vlong;

        while(ziplistGet(p,&vstr,&vlen,&vlong)) {
            if (count == 0) {
686 687
                int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
                    REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
688

689 690 691 692 693 694 695 696 697 698
                if (rioWriteBulkCount(r,'*',2+cmd_items) == 0) return 0;
                if (rioWriteBulkString(r,"RPUSH",5) == 0) return 0;
                if (rioWriteBulkObject(r,key) == 0) return 0;
            }
            if (vstr) {
                if (rioWriteBulkString(r,(char*)vstr,vlen) == 0) return 0;
            } else {
                if (rioWriteBulkLongLong(r,vlong) == 0) return 0;
            }
            p = ziplistNext(zl,p);
699
            if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
700 701 702 703 704 705 706 707 708 709 710
            items--;
        }
    } else if (o->encoding == REDIS_ENCODING_LINKEDLIST) {
        list *list = o->ptr;
        listNode *ln;
        listIter li;

        listRewind(list,&li);
        while((ln = listNext(&li))) {
            robj *eleobj = listNodeValue(ln);

711
            if (count == 0) {
712 713
                int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
                    REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
714

715 716 717 718
                if (rioWriteBulkCount(r,'*',2+cmd_items) == 0) return 0;
                if (rioWriteBulkString(r,"RPUSH",5) == 0) return 0;
                if (rioWriteBulkObject(r,key) == 0) return 0;
            }
719
            if (rioWriteBulkObject(r,eleobj) == 0) return 0;
720
            if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
721
            items--;
722 723 724 725 726 727 728
        }
    } else {
        redisPanic("Unknown list encoding");
    }
    return 1;
}

729 730 731 732 733 734 735 736 737 738 739
/* Emit the commands needed to rebuild a set object.
 * The function returns 0 on error, 1 on success. */
int rewriteSetObject(rio *r, robj *key, robj *o) {
    long long count = 0, items = setTypeSize(o);

    if (o->encoding == REDIS_ENCODING_INTSET) {
        int ii = 0;
        int64_t llval;

        while(intsetGet(o->ptr,ii++,&llval)) {
            if (count == 0) {
740 741
                int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
                    REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
742 743 744 745 746 747

                if (rioWriteBulkCount(r,'*',2+cmd_items) == 0) return 0;
                if (rioWriteBulkString(r,"SADD",4) == 0) return 0;
                if (rioWriteBulkObject(r,key) == 0) return 0;
            }
            if (rioWriteBulkLongLong(r,llval) == 0) return 0;
748
            if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
749 750 751 752 753 754 755 756 757
            items--;
        }
    } else if (o->encoding == REDIS_ENCODING_HT) {
        dictIterator *di = dictGetIterator(o->ptr);
        dictEntry *de;

        while((de = dictNext(di)) != NULL) {
            robj *eleobj = dictGetKey(de);
            if (count == 0) {
758 759
                int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
                    REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
760 761 762 763 764 765

                if (rioWriteBulkCount(r,'*',2+cmd_items) == 0) return 0;
                if (rioWriteBulkString(r,"SADD",4) == 0) return 0;
                if (rioWriteBulkObject(r,key) == 0) return 0;
            }
            if (rioWriteBulkObject(r,eleobj) == 0) return 0;
766
            if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
767 768 769 770 771 772 773 774 775
            items--;
        }
        dictReleaseIterator(di);
    } else {
        redisPanic("Unknown set encoding");
    }
    return 1;
}

776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798
/* Emit the commands needed to rebuild a sorted set object.
 * The function returns 0 on error, 1 on success. */
int rewriteSortedSetObject(rio *r, robj *key, robj *o) {
    long long count = 0, items = zsetLength(o);

    if (o->encoding == REDIS_ENCODING_ZIPLIST) {
        unsigned char *zl = o->ptr;
        unsigned char *eptr, *sptr;
        unsigned char *vstr;
        unsigned int vlen;
        long long vll;
        double score;

        eptr = ziplistIndex(zl,0);
        redisAssert(eptr != NULL);
        sptr = ziplistNext(zl,eptr);
        redisAssert(sptr != NULL);

        while (eptr != NULL) {
            redisAssert(ziplistGet(eptr,&vstr,&vlen,&vll));
            score = zzlGetScore(sptr);

            if (count == 0) {
799 800
                int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
                    REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
801 802 803 804 805 806 807 808 809 810 811 812

                if (rioWriteBulkCount(r,'*',2+cmd_items*2) == 0) return 0;
                if (rioWriteBulkString(r,"ZADD",4) == 0) return 0;
                if (rioWriteBulkObject(r,key) == 0) return 0;
            }
            if (rioWriteBulkDouble(r,score) == 0) return 0;
            if (vstr != NULL) {
                if (rioWriteBulkString(r,(char*)vstr,vlen) == 0) return 0;
            } else {
                if (rioWriteBulkLongLong(r,vll) == 0) return 0;
            }
            zzlNext(zl,&eptr,&sptr);
813
            if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
814 815 816 817 818 819 820 821 822 823 824 825
            items--;
        }
    } else if (o->encoding == REDIS_ENCODING_SKIPLIST) {
        zset *zs = o->ptr;
        dictIterator *di = dictGetIterator(zs->dict);
        dictEntry *de;

        while((de = dictNext(di)) != NULL) {
            robj *eleobj = dictGetKey(de);
            double *score = dictGetVal(de);

            if (count == 0) {
826 827
                int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
                    REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
828 829 830 831 832 833 834

                if (rioWriteBulkCount(r,'*',2+cmd_items*2) == 0) return 0;
                if (rioWriteBulkString(r,"ZADD",4) == 0) return 0;
                if (rioWriteBulkObject(r,key) == 0) return 0;
            }
            if (rioWriteBulkDouble(r,*score) == 0) return 0;
            if (rioWriteBulkObject(r,eleobj) == 0) return 0;
835
            if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
836 837 838 839 840 841 842 843 844
            items--;
        }
        dictReleaseIterator(di);
    } else {
        redisPanic("Unknown sorted zset encoding");
    }
    return 1;
}

845
/* Write either the key or the value of the currently selected item of a hash.
846 847 848 849 850
 * The 'hi' argument passes a valid Redis hash iterator.
 * The 'what' filed specifies if to write a key or a value and can be
 * either REDIS_HASH_KEY or REDIS_HASH_VALUE.
 *
 * The function returns 0 on error, non-zero on success. */
851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874
static int rioWriteHashIteratorCursor(rio *r, hashTypeIterator *hi, int what) {
    if (hi->encoding == REDIS_ENCODING_ZIPLIST) {
        unsigned char *vstr = NULL;
        unsigned int vlen = UINT_MAX;
        long long vll = LLONG_MAX;

        hashTypeCurrentFromZiplist(hi, what, &vstr, &vlen, &vll);
        if (vstr) {
            return rioWriteBulkString(r, (char*)vstr, vlen);
        } else {
            return rioWriteBulkLongLong(r, vll);
        }

    } else if (hi->encoding == REDIS_ENCODING_HT) {
        robj *value;

        hashTypeCurrentFromHashTable(hi, what, &value);
        return rioWriteBulkObject(r, value);
    }

    redisPanic("Unknown hash encoding");
    return 0;
}

A
antirez 已提交
875 876 877
/* Emit the commands needed to rebuild a hash object.
 * The function returns 0 on error, 1 on success. */
int rewriteHashObject(rio *r, robj *key, robj *o) {
878
    hashTypeIterator *hi;
A
antirez 已提交
879 880
    long long count = 0, items = hashTypeLength(o);

881 882 883 884 885
    hi = hashTypeInitIterator(o);
    while (hashTypeNext(hi) != REDIS_ERR) {
        if (count == 0) {
            int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
                REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
A
antirez 已提交
886

887 888 889
            if (rioWriteBulkCount(r,'*',2+cmd_items*2) == 0) return 0;
            if (rioWriteBulkString(r,"HMSET",5) == 0) return 0;
            if (rioWriteBulkObject(r,key) == 0) return 0;
A
antirez 已提交
890 891
        }

892 893 894 895 896
        if (rioWriteHashIteratorCursor(r, hi, REDIS_HASH_KEY) == 0) return 0;
        if (rioWriteHashIteratorCursor(r, hi, REDIS_HASH_VALUE) == 0) return 0;
        if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
        items--;
    }
A
antirez 已提交
897

898
    hashTypeReleaseIterator(hi);
A
antirez 已提交
899 900 901 902

    return 1;
}

903
/* Write a sequence of commands able to fully rebuild the dataset into
904 905 906 907
 * "filename". Used both by REWRITEAOF and BGREWRITEAOF.
 *
 * In order to minimize the number of commands needed in the rewritten
 * log Redis uses variadic commands when possible, such as RPUSH, SADD
908
 * and ZADD. However at max REDIS_AOF_REWRITE_ITEMS_PER_CMD items per time
909
 * are inserted using a single command. */
910 911 912
int rewriteAppendOnlyFile(char *filename) {
    dictIterator *di = NULL;
    dictEntry *de;
P
Pieter Noordhuis 已提交
913
    rio aof;
914 915 916
    FILE *fp;
    char tmpfile[256];
    int j;
917
    long long now = mstime();
918 919 920 921 922 923

    /* Note that we have to use a different temp name here compared to the
     * one used by rewriteAppendOnlyFileBackground() function. */
    snprintf(tmpfile,256,"temp-rewriteaof-%d.aof", (int) getpid());
    fp = fopen(tmpfile,"w");
    if (!fp) {
A
antirez 已提交
924
        redisLog(REDIS_WARNING, "Opening the temp file for AOF rewrite in rewriteAppendOnlyFile(): %s", strerror(errno));
925 926
        return REDIS_ERR;
    }
P
Pieter Noordhuis 已提交
927

928
    rioInitWithFile(&aof,fp);
929 930
    if (server.aof_rewrite_incremental_fsync)
        rioSetAutoSync(&aof,REDIS_AOF_AUTOSYNC_BYTES);
931 932 933 934 935
    for (j = 0; j < server.dbnum; j++) {
        char selectcmd[] = "*2\r\n$6\r\nSELECT\r\n";
        redisDb *db = server.db+j;
        dict *d = db->dict;
        if (dictSize(d) == 0) continue;
936
        di = dictGetSafeIterator(d);
937 938 939 940 941 942
        if (!di) {
            fclose(fp);
            return REDIS_ERR;
        }

        /* SELECT the new DB */
P
Pieter Noordhuis 已提交
943 944
        if (rioWrite(&aof,selectcmd,sizeof(selectcmd)-1) == 0) goto werr;
        if (rioWriteBulkLongLong(&aof,j) == 0) goto werr;
945 946 947

        /* Iterate this DB writing every entry */
        while((de = dictNext(di)) != NULL) {
A
antirez 已提交
948
            sds keystr;
949
            robj key, *o;
950
            long long expiretime;
951

952 953
            keystr = dictGetKey(de);
            o = dictGetVal(de);
954
            initStaticStringObject(key,keystr);
955

956 957
            expiretime = getExpire(db,&key);

958 959 960
            /* If this key is already expired skip it */
            if (expiretime != -1 && expiretime < now) continue;

961 962 963 964
            /* Save the key and associated value */
            if (o->type == REDIS_STRING) {
                /* Emit a SET command */
                char cmd[]="*3\r\n$3\r\nSET\r\n";
P
Pieter Noordhuis 已提交
965
                if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr;
966
                /* Key and value */
P
Pieter Noordhuis 已提交
967 968
                if (rioWriteBulkObject(&aof,&key) == 0) goto werr;
                if (rioWriteBulkObject(&aof,o) == 0) goto werr;
969
            } else if (o->type == REDIS_LIST) {
970
                if (rewriteListObject(&aof,&key,o) == 0) goto werr;
971
            } else if (o->type == REDIS_SET) {
972
                if (rewriteSetObject(&aof,&key,o) == 0) goto werr;
973
            } else if (o->type == REDIS_ZSET) {
974
                if (rewriteSortedSetObject(&aof,&key,o) == 0) goto werr;
975
            } else if (o->type == REDIS_HASH) {
A
antirez 已提交
976
                if (rewriteHashObject(&aof,&key,o) == 0) goto werr;
977 978 979 980 981
            } else {
                redisPanic("Unknown object type");
            }
            /* Save the expire time */
            if (expiretime != -1) {
982
                char cmd[]="*3\r\n$9\r\nPEXPIREAT\r\n";
P
Pieter Noordhuis 已提交
983 984
                if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr;
                if (rioWriteBulkObject(&aof,&key) == 0) goto werr;
985
                if (rioWriteBulkLongLong(&aof,expiretime) == 0) goto werr;
986 987 988 989 990 991
            }
        }
        dictReleaseIterator(di);
    }

    /* Make sure data will not remain on the OS's output buffers */
992
    if (fflush(fp) == EOF) goto werr;
993
    if (fsync(fileno(fp)) == -1) goto werr;
994
    if (fclose(fp) == EOF) goto werr;
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018

    /* Use RENAME to make sure the DB file is changed atomically only
     * if the generate DB file is ok. */
    if (rename(tmpfile,filename) == -1) {
        redisLog(REDIS_WARNING,"Error moving temp append only file on the final destination: %s", strerror(errno));
        unlink(tmpfile);
        return REDIS_ERR;
    }
    redisLog(REDIS_NOTICE,"SYNC append only file rewrite performed");
    return REDIS_OK;

werr:
    fclose(fp);
    unlink(tmpfile);
    redisLog(REDIS_WARNING,"Write error writing append only file on disk: %s", strerror(errno));
    if (di) dictReleaseIterator(di);
    return REDIS_ERR;
}

/* This is how rewriting of the append only file in background works:
 *
 * 1) The user calls BGREWRITEAOF
 * 2) Redis calls this function, that forks():
 *    2a) the child rewrite the append only file in a temp file.
A
antirez 已提交
1019
 *    2b) the parent accumulates differences in server.aof_rewrite_buf.
1020 1021
 * 3) When the child finished '2a' exists.
 * 4) The parent will trap the exit code, if it's OK, will append the
A
antirez 已提交
1022
 *    data accumulated into server.aof_rewrite_buf into the temp file, and
1023 1024 1025 1026 1027
 *    finally will rename(2) the temp file in the actual file name.
 *    The the new file is reopened as the new append only file. Profit!
 */
int rewriteAppendOnlyFileBackground(void) {
    pid_t childpid;
1028
    long long start;
1029

A
antirez 已提交
1030
    if (server.aof_child_pid != -1) return REDIS_ERR;
1031
    start = ustime();
1032 1033 1034
    if ((childpid = fork()) == 0) {
        char tmpfile[256];

1035
        /* Child */
1036
        closeListeningSockets(0);
1037
        redisSetProcTitle("redis-aof-rewrite");
1038 1039
        snprintf(tmpfile,256,"temp-rewriteaof-bg-%d.aof", (int) getpid());
        if (rewriteAppendOnlyFile(tmpfile) == REDIS_OK) {
1040 1041 1042 1043
            size_t private_dirty = zmalloc_get_private_dirty();

            if (private_dirty) {
                redisLog(REDIS_NOTICE,
1044
                    "AOF rewrite: %zu MB of memory used by copy-on-write",
1045 1046
                    private_dirty/(1024*1024));
            }
1047
            exitFromChild(0);
1048
        } else {
1049
            exitFromChild(1);
1050 1051 1052
        }
    } else {
        /* Parent */
1053
        server.stat_fork_time = ustime()-start;
1054
        server.stat_fork_rate = (double) zmalloc_used_memory() * 1000000 / server.stat_fork_time / (1024*1024*1024); /* GB per second. */
1055
        latencyAddSampleIfNeeded("fork",server.stat_fork_time/1000);
1056 1057 1058 1059 1060 1061 1062 1063
        if (childpid == -1) {
            redisLog(REDIS_WARNING,
                "Can't rewrite append only file in background: fork: %s",
                strerror(errno));
            return REDIS_ERR;
        }
        redisLog(REDIS_NOTICE,
            "Background append only file rewriting started by pid %d",childpid);
1064
        server.aof_rewrite_scheduled = 0;
1065
        server.aof_rewrite_time_start = time(NULL);
A
antirez 已提交
1066
        server.aof_child_pid = childpid;
1067 1068 1069
        updateDictResizePolicy();
        /* We set appendseldb to -1 in order to force the next call to the
         * feedAppendOnlyFile() to issue a SELECT command, so the differences
A
antirez 已提交
1070
         * accumulated by the parent into server.aof_rewrite_buf will start
1071
         * with a SELECT statement and it will be safe to merge. */
A
antirez 已提交
1072
        server.aof_selected_db = -1;
1073
        replicationScriptCacheFlush();
1074 1075 1076 1077 1078 1079
        return REDIS_OK;
    }
    return REDIS_OK; /* unreached */
}

void bgrewriteaofCommand(redisClient *c) {
A
antirez 已提交
1080
    if (server.aof_child_pid != -1) {
1081
        addReplyError(c,"Background append only file rewriting already in progress");
A
antirez 已提交
1082
    } else if (server.rdb_child_pid != -1) {
1083
        server.aof_rewrite_scheduled = 1;
1084
        addReplyStatus(c,"Background append only file rewriting scheduled");
1085
    } else if (rewriteAppendOnlyFileBackground() == REDIS_OK) {
1086
        addReplyStatus(c,"Background append only file rewriting started");
1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098
    } else {
        addReply(c,shared.err);
    }
}

void aofRemoveTempFile(pid_t childpid) {
    char tmpfile[256];

    snprintf(tmpfile,256,"temp-rewriteaof-bg-%d.aof", (int) childpid);
    unlink(tmpfile);
}

1099
/* Update the server.aof_current_size filed explicitly using stat(2)
1100 1101
 * to check the size of the file. This is useful after a rewrite or after
 * a restart, normally the size is updated just adding the write length
A
antirez 已提交
1102
 * to the current length, that is much faster. */
1103 1104
void aofUpdateCurrentSize(void) {
    struct redis_stat sb;
1105
    mstime_t latency;
1106

1107
    latencyStartMonitor(latency);
A
antirez 已提交
1108
    if (redis_fstat(server.aof_fd,&sb) == -1) {
A
antirez 已提交
1109
        redisLog(REDIS_WARNING,"Unable to obtain the AOF file length. stat: %s",
1110 1111
            strerror(errno));
    } else {
1112
        server.aof_current_size = sb.st_size;
1113
    }
1114 1115
    latencyEndMonitor(latency);
    latencyAddSampleIfNeeded("aof-fstat",latency);
1116 1117
}

1118 1119
/* A background append only file rewriting (BGREWRITEAOF) terminated its work.
 * Handle this. */
1120
void backgroundRewriteDoneHandler(int exitcode, int bysignal) {
1121
    if (!bysignal && exitcode == 0) {
1122
        int newfd, oldfd;
1123
        char tmpfile[256];
1124
        long long now = ustime();
1125
        mstime_t latency;
1126 1127

        redisLog(REDIS_NOTICE,
1128 1129
            "Background AOF rewrite terminated with success");

1130 1131
        /* Flush the differences accumulated by the parent to the
         * rewritten AOF. */
1132
        latencyStartMonitor(latency);
1133
        snprintf(tmpfile,256,"temp-rewriteaof-bg-%d.aof",
A
antirez 已提交
1134
            (int)server.aof_child_pid);
1135 1136 1137 1138
        newfd = open(tmpfile,O_WRONLY|O_APPEND);
        if (newfd == -1) {
            redisLog(REDIS_WARNING,
                "Unable to open the temporary AOF produced by the child: %s", strerror(errno));
1139 1140
            goto cleanup;
        }
1141

1142 1143 1144
        if (aofRewriteBufferWrite(newfd) == -1) {
            redisLog(REDIS_WARNING,
                "Error trying to flush the parent diff to the rewritten AOF: %s", strerror(errno));
1145
            close(newfd);
1146 1147
            goto cleanup;
        }
1148 1149
        latencyEndMonitor(latency);
        latencyAddSampleIfNeeded("aof-rewrite-diff-write",latency);
1150 1151

        redisLog(REDIS_NOTICE,
1152
            "Parent diff successfully flushed to the rewritten AOF (%lu bytes)", aofRewriteBufferSize());
1153 1154 1155

        /* The only remaining thing to do is to rename the temporary file to
         * the configured file and switch the file descriptor used to do AOF
1156 1157 1158 1159
         * writes. We don't want close(2) or rename(2) calls to block the
         * server on old file deletion.
         *
         * There are two possible scenarios:
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173
         *
         * 1) AOF is DISABLED and this was a one time rewrite. The temporary
         * file will be renamed to the configured file. When this file already
         * exists, it will be unlinked, which may block the server.
         *
         * 2) AOF is ENABLED and the rewritten AOF will immediately start
         * receiving writes. After the temporary file is renamed to the
         * configured file, the original AOF file descriptor will be closed.
         * Since this will be the last reference to that file, closing it
         * causes the underlying file to be unlinked, which may block the
         * server.
         *
         * To mitigate the blocking effect of the unlink operation (either
         * caused by rename(2) in scenario 1, or by close(2) in scenario 2), we
1174
         * use a background thread to take care of this. First, we
1175 1176 1177 1178 1179 1180
         * make scenario 1 identical to scenario 2 by opening the target file
         * when it exists. The unlink operation after the rename(2) will then
         * be executed upon calling close(2) for its descriptor. Everything to
         * guarantee atomicity for this switch has already happened by then, so
         * we don't care what the outcome or duration of that close operation
         * is, as long as the file descriptor is released again. */
A
antirez 已提交
1181
        if (server.aof_fd == -1) {
1182 1183
            /* AOF disabled */

1184 1185 1186
             /* Don't care if this fails: oldfd will be -1 and we handle that.
              * One notable case of -1 return is if the old file does
              * not exist. */
1187
             oldfd = open(server.aof_filename,O_RDONLY|O_NONBLOCK);
1188 1189
        } else {
            /* AOF enabled */
1190
            oldfd = -1; /* We'll set this to the current AOF filedes later. */
1191 1192 1193 1194
        }

        /* Rename the temporary file. This will not unlink the target file if
         * it exists, because we reference it with "oldfd". */
1195
        latencyStartMonitor(latency);
1196
        if (rename(tmpfile,server.aof_filename) == -1) {
1197
            redisLog(REDIS_WARNING,
A
antirez 已提交
1198
                "Error trying to rename the temporary AOF file: %s", strerror(errno));
1199
            close(newfd);
1200
            if (oldfd != -1) close(oldfd);
1201 1202
            goto cleanup;
        }
1203 1204
        latencyEndMonitor(latency);
        latencyAddSampleIfNeeded("aof-rename",latency);
1205

A
antirez 已提交
1206
        if (server.aof_fd == -1) {
1207 1208
            /* AOF disabled, we don't need to set the AOF file descriptor
             * to this new file, so we can close it. */
1209 1210
            close(newfd);
        } else {
1211
            /* AOF enabled, replace the old fd with the new one. */
A
antirez 已提交
1212 1213
            oldfd = server.aof_fd;
            server.aof_fd = newfd;
1214
            if (server.aof_fsync == AOF_FSYNC_ALWAYS)
1215
                aof_fsync(newfd);
1216
            else if (server.aof_fsync == AOF_FSYNC_EVERYSEC)
1217
                aof_background_fsync(newfd);
A
antirez 已提交
1218
            server.aof_selected_db = -1; /* Make sure SELECT is re-issued */
1219
            aofUpdateCurrentSize();
1220
            server.aof_rewrite_base_size = server.aof_current_size;
1221 1222 1223

            /* Clear regular AOF buffer since its contents was just written to
             * the new AOF from the background rewrite buffer. */
A
antirez 已提交
1224 1225
            sdsfree(server.aof_buf);
            server.aof_buf = sdsempty();
1226
        }
1227

1228 1229
        server.aof_lastbgrewrite_status = REDIS_OK;

A
antirez 已提交
1230
        redisLog(REDIS_NOTICE, "Background AOF rewrite finished successfully");
1231 1232 1233
        /* Change state from WAIT_REWRITE to ON if needed */
        if (server.aof_state == REDIS_AOF_WAIT_REWRITE)
            server.aof_state = REDIS_AOF_ON;
1234 1235

        /* Asynchronously close the overwritten AOF. */
1236
        if (oldfd != -1) bioCreateBackgroundJob(REDIS_BIO_CLOSE_FILE,(void*)(long)oldfd,NULL,NULL);
1237 1238 1239

        redisLog(REDIS_VERBOSE,
            "Background AOF rewrite signal handler took %lldus", ustime()-now);
1240
    } else if (!bysignal && exitcode != 0) {
1241 1242
        server.aof_lastbgrewrite_status = REDIS_ERR;

1243 1244
        redisLog(REDIS_WARNING,
            "Background AOF rewrite terminated with error");
1245
    } else {
1246 1247
        server.aof_lastbgrewrite_status = REDIS_ERR;

1248
        redisLog(REDIS_WARNING,
1249
            "Background AOF rewrite terminated by signal %d", bysignal);
1250
    }
1251

1252
cleanup:
1253
    aofRewriteBufferReset();
A
antirez 已提交
1254 1255
    aofRemoveTempFile(server.aof_child_pid);
    server.aof_child_pid = -1;
1256 1257
    server.aof_rewrite_time_last = time(NULL)-server.aof_rewrite_time_start;
    server.aof_rewrite_time_start = -1;
1258 1259
    /* Schedule a new rewrite if we are waiting for it to switch the AOF ON. */
    if (server.aof_state == REDIS_AOF_WAIT_REWRITE)
1260
        server.aof_rewrite_scheduled = 1;
1261
}