X-Git-Url: https://git.saurik.com/redis.git/blobdiff_plain/b454056d9480b91fd8935a7719ac3ffa40445ddf..f14479c796fbf00c41ceefdcb4bec04df46f53ce:/src/aof.c?ds=sidebyside diff --git a/src/aof.c b/src/aof.c index 1a6abb98..4a463bde 100644 --- a/src/aof.c +++ b/src/aof.c @@ -1,4 +1,6 @@ #include "redis.h" +#include "bio.h" +#include "rio.h" #include #include @@ -10,10 +12,14 @@ void aofUpdateCurrentSize(void); +void aof_background_fsync(int fd) { + bioCreateBackgroundJob(REDIS_BIO_AOF_FSYNC,(void*)(long)fd,NULL,NULL); +} + /* Called when the user switches from "appendonly yes" to "appendonly no" * at runtime using the CONFIG command. */ void stopAppendOnly(void) { - flushAppendOnlyFile(); + flushAppendOnlyFile(1); aof_fsync(server.appendfd); close(server.appendfd); @@ -46,7 +52,7 @@ int startAppendOnly(void) { if (rewriteAppendOnlyFileBackground() == REDIS_ERR) { server.appendonly = 0; close(server.appendfd); - redisLog(REDIS_WARNING,"Used tried to switch on AOF via CONFIG, I can't trigger a background AOF rewrite operation. Check the above logs for more info about the error.",strerror(errno)); + redisLog(REDIS_WARNING,"User tried turning on AOF with CONFIG SET but I can't trigger a background AOF rewrite operation. Check the above logs for more info about the error."); return REDIS_ERR; } return REDIS_OK; @@ -58,12 +64,51 @@ int startAppendOnly(void) { * 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 - * the event loop again. */ -void flushAppendOnlyFile(void) { + * 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. */ +void flushAppendOnlyFile(int force) { ssize_t nwritten; + int sync_in_progress = 0; if (sdslen(server.aofbuf) == 0) return; + if (server.appendfsync == APPENDFSYNC_EVERYSEC) + sync_in_progress = bioPendingJobsOfType(REDIS_BIO_AOF_FSYNC) != 0; + + if (server.appendfsync == APPENDFSYNC_EVERYSEC && !force) { + /* 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) { + /* We were already waiting for fsync to finish, but for less + * than two seconds this is still ok. Postpone again. */ + return; + } + /* Otherwise fall trough, and go write since we can't wait + * over two seconds. */ + 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."); + } + } + /* If you are following this code path, then we are going to write so + * set reset the postponed flush sentinel to zero. */ + server.aof_flush_postponed_start = 0; + /* 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 @@ -99,14 +144,15 @@ void flushAppendOnlyFile(void) { return; /* Perform the fsync if needed. */ - if (server.appendfsync == APPENDFSYNC_ALWAYS || - (server.appendfsync == APPENDFSYNC_EVERYSEC && - server.unixtime > server.lastfsync)) - { + if (server.appendfsync == APPENDFSYNC_ALWAYS) { /* aof_fsync is defined as fdatasync() for Linux in order to avoid * flushing metadata. */ aof_fsync(server.appendfd); /* Let's try to get this data on the disk */ server.lastfsync = server.unixtime; + } else if ((server.appendfsync == APPENDFSYNC_EVERYSEC && + server.unixtime > server.lastfsync)) { + if (!sync_in_progress) aof_background_fsync(server.appendfd); + server.lastfsync = server.unixtime; } } @@ -135,21 +181,38 @@ sds catAppendOnlyGenericCommand(sds dst, int argc, robj **argv) { return dst; } -sds catAppendOnlyExpireAtCommand(sds buf, robj *key, robj *seconds) { - int argc = 3; - long when; +/* 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; robj *argv[3]; /* Make sure we can use strtol */ seconds = getDecodedObject(seconds); - when = time(NULL)+strtol(seconds->ptr,NULL,10); + 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(); + } decrRefCount(seconds); - argv[0] = createStringObject("EXPIREAT",8); + argv[0] = createStringObject("PEXPIREAT",9); argv[1] = key; - argv[2] = createObject(REDIS_STRING, - sdscatprintf(sdsempty(),"%ld",when)); - buf = catAppendOnlyGenericCommand(buf, argc, argv); + argv[2] = createStringObjectFromLongLong(when); + buf = catAppendOnlyGenericCommand(buf, 3, argv); decrRefCount(argv[0]); decrRefCount(argv[2]); return buf; @@ -170,18 +233,22 @@ void feedAppendOnlyFile(struct redisCommand *cmd, int dictid, robj **argv, int a server.appendseldb = dictid; } - if (cmd->proc == expireCommand) { - /* Translate EXPIRE into EXPIREAT */ - buf = catAppendOnlyExpireAtCommand(buf,argv[1],argv[2]); - } else if (cmd->proc == setexCommand) { - /* Translate SETEX to SET and EXPIREAT */ + 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 */ tmpargv[0] = createStringObject("SET",3); tmpargv[1] = argv[1]; tmpargv[2] = argv[3]; buf = catAppendOnlyGenericCommand(buf,3,tmpargv); decrRefCount(tmpargv[0]); - buf = catAppendOnlyExpireAtCommand(buf,argv[1],argv[2]); + buf = catAppendOnlyExpireAtCommand(buf,cmd,argv[1],argv[2]); } else { + /* All the other commands don't need translation or need the + * same translation already operated in the command vector + * for the replication itself. */ buf = catAppendOnlyGenericCommand(buf,argc,argv); } @@ -281,6 +348,8 @@ int loadAppendOnlyFile(char *filename) { } if (buf[0] != '*') goto fmterr; argc = atoi(buf+1); + if (argc < 1) goto fmterr; + argv = zmalloc(sizeof(robj*)*argc); for (j = 0; j < argc; j++) { if (fgets(buf,sizeof(buf),fp) == NULL) goto readerr; @@ -339,15 +408,30 @@ fmterr: exit(1); } +/* 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"); + } +} + /* Write a sequence of commands able to fully rebuild the dataset into * "filename". Used both by REWRITEAOF and BGREWRITEAOF. */ int rewriteAppendOnlyFile(char *filename) { dictIterator *di = NULL; dictEntry *de; + rio aof; FILE *fp; char tmpfile[256]; int j; - time_t now = time(NULL); + long long now = mstime(); /* Note that we have to use a different temp name here compared to the * one used by rewriteAppendOnlyFileBackground() function. */ @@ -357,6 +441,8 @@ int rewriteAppendOnlyFile(char *filename) { redisLog(REDIS_WARNING, "Failed rewriting the append only file: %s", strerror(errno)); return REDIS_ERR; } + + rioInitWithFile(&aof,fp); for (j = 0; j < server.dbnum; j++) { char selectcmd[] = "*2\r\n$6\r\nSELECT\r\n"; redisDb *db = server.db+j; @@ -369,17 +455,17 @@ int rewriteAppendOnlyFile(char *filename) { } /* SELECT the new DB */ - if (fwrite(selectcmd,sizeof(selectcmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkLongLong(fp,j) == 0) goto werr; + if (rioWrite(&aof,selectcmd,sizeof(selectcmd)-1) == 0) goto werr; + if (rioWriteBulkLongLong(&aof,j) == 0) goto werr; /* Iterate this DB writing every entry */ while((de = dictNext(di)) != NULL) { sds keystr; robj key, *o; - time_t expiretime; + long long expiretime; - keystr = dictGetEntryKey(de); - o = dictGetEntryVal(de); + keystr = dictGetKey(de); + o = dictGetVal(de); initStaticStringObject(key,keystr); expiretime = getExpire(db,&key); @@ -388,10 +474,10 @@ int rewriteAppendOnlyFile(char *filename) { if (o->type == REDIS_STRING) { /* Emit a SET command */ char cmd[]="*3\r\n$3\r\nSET\r\n"; - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; /* Key and value */ - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkObject(fp,o) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkObject(&aof,o) == 0) goto werr; } else if (o->type == REDIS_LIST) { /* Emit the RPUSHes needed to rebuild the list */ char cmd[]="*3\r\n$5\r\nRPUSH\r\n"; @@ -403,13 +489,13 @@ int rewriteAppendOnlyFile(char *filename) { long long vlong; while(ziplistGet(p,&vstr,&vlen,&vlong)) { - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; if (vstr) { - if (fwriteBulkString(fp,(char*)vstr,vlen) == 0) + if (rioWriteBulkString(&aof,(char*)vstr,vlen) == 0) goto werr; } else { - if (fwriteBulkLongLong(fp,vlong) == 0) + if (rioWriteBulkLongLong(&aof,vlong) == 0) goto werr; } p = ziplistNext(zl,p); @@ -423,9 +509,9 @@ int rewriteAppendOnlyFile(char *filename) { while((ln = listNext(&li))) { robj *eleobj = listNodeValue(ln); - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkObject(fp,eleobj) == 0) goto werr; + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkObject(&aof,eleobj) == 0) goto werr; } } else { redisPanic("Unknown list encoding"); @@ -438,18 +524,18 @@ int rewriteAppendOnlyFile(char *filename) { int ii = 0; int64_t llval; while(intsetGet(o->ptr,ii++,&llval)) { - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkLongLong(fp,llval) == 0) goto werr; + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkLongLong(&aof,llval) == 0) goto werr; } } else if (o->encoding == REDIS_ENCODING_HT) { dictIterator *di = dictGetIterator(o->ptr); dictEntry *de; while((de = dictNext(di)) != NULL) { - robj *eleobj = dictGetEntryKey(de); - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkObject(fp,eleobj) == 0) goto werr; + robj *eleobj = dictGetKey(de); + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkObject(&aof,eleobj) == 0) goto werr; } dictReleaseIterator(di); } else { @@ -476,14 +562,14 @@ int rewriteAppendOnlyFile(char *filename) { redisAssert(ziplistGet(eptr,&vstr,&vlen,&vll)); score = zzlGetScore(sptr); - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkDouble(fp,score) == 0) goto werr; + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkDouble(&aof,score) == 0) goto werr; if (vstr != NULL) { - if (fwriteBulkString(fp,(char*)vstr,vlen) == 0) + if (rioWriteBulkString(&aof,(char*)vstr,vlen) == 0) goto werr; } else { - if (fwriteBulkLongLong(fp,vll) == 0) + if (rioWriteBulkLongLong(&aof,vll) == 0) goto werr; } zzlNext(zl,&eptr,&sptr); @@ -494,13 +580,13 @@ int rewriteAppendOnlyFile(char *filename) { dictEntry *de; while((de = dictNext(di)) != NULL) { - robj *eleobj = dictGetEntryKey(de); - double *score = dictGetEntryVal(de); + robj *eleobj = dictGetKey(de); + double *score = dictGetVal(de); - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkDouble(fp,*score) == 0) goto werr; - if (fwriteBulkObject(fp,eleobj) == 0) goto werr; + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkDouble(&aof,*score) == 0) goto werr; + if (rioWriteBulkObject(&aof,eleobj) == 0) goto werr; } dictReleaseIterator(di); } else { @@ -516,11 +602,11 @@ int rewriteAppendOnlyFile(char *filename) { unsigned int flen, vlen; while((p = zipmapNext(p,&field,&flen,&val,&vlen)) != NULL) { - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkString(fp,(char*)field,flen) == 0) + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkString(&aof,(char*)field,flen) == 0) goto werr; - if (fwriteBulkString(fp,(char*)val,vlen) == 0) + if (rioWriteBulkString(&aof,(char*)val,vlen) == 0) goto werr; } } else { @@ -528,13 +614,13 @@ int rewriteAppendOnlyFile(char *filename) { dictEntry *de; while((de = dictNext(di)) != NULL) { - robj *field = dictGetEntryKey(de); - robj *val = dictGetEntryVal(de); + robj *field = dictGetKey(de); + robj *val = dictGetVal(de); - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkObject(fp,field) == 0) goto werr; - if (fwriteBulkObject(fp,val) == 0) goto werr; + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkObject(&aof,field) == 0) goto werr; + if (rioWriteBulkObject(&aof,val) == 0) goto werr; } dictReleaseIterator(di); } @@ -543,12 +629,12 @@ int rewriteAppendOnlyFile(char *filename) { } /* Save the expire time */ if (expiretime != -1) { - char cmd[]="*3\r\n$8\r\nEXPIREAT\r\n"; + char cmd[]="*3\r\n$9\r\nPEXPIREAT\r\n"; /* If this key is already expired skip it */ if (expiretime < now) continue; - if (fwrite(cmd,sizeof(cmd)-1,1,fp) == 0) goto werr; - if (fwriteBulkObject(fp,&key) == 0) goto werr; - if (fwriteBulkLongLong(fp,expiretime) == 0) goto werr; + if (rioWrite(&aof,cmd,sizeof(cmd)-1) == 0) goto werr; + if (rioWriteBulkObject(&aof,&key) == 0) goto werr; + if (rioWriteBulkLongLong(&aof,expiretime) == 0) goto werr; } } dictReleaseIterator(di); @@ -618,6 +704,7 @@ int rewriteAppendOnlyFileBackground(void) { } redisLog(REDIS_NOTICE, "Background append only file rewriting started by pid %d",childpid); + server.aofrewrite_scheduled = 0; server.bgrewritechildpid = childpid; updateDictResizePolicy(); /* We set appendseldb to -1 in order to force the next call to the @@ -677,8 +764,10 @@ void backgroundRewriteDoneHandler(int exitcode, int bysignal) { redisLog(REDIS_NOTICE, "Background AOF rewrite terminated with success"); - /* Flush the differences accumulated by the parent to the rewritten AOF. */ - snprintf(tmpfile,256,"temp-rewriteaof-bg-%d.aof", (int)server.bgrewritechildpid); + /* Flush the differences accumulated by the parent to the + * rewritten AOF. */ + snprintf(tmpfile,256,"temp-rewriteaof-bg-%d.aof", + (int)server.bgrewritechildpid); newfd = open(tmpfile,O_WRONLY|O_APPEND); if (newfd == -1) { redisLog(REDIS_WARNING, @@ -704,7 +793,10 @@ void backgroundRewriteDoneHandler(int exitcode, int bysignal) { /* 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 - * writes. There are two possible scenarios: + * writes. We don't want close(2) or rename(2) calls to block the + * server on old file deletion. + * + * There are two possible scenarios: * * 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 @@ -719,7 +811,7 @@ void backgroundRewriteDoneHandler(int exitcode, int bysignal) { * * 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 - * use a background thread in libeio to take care of this. First, we + * use a background thread to take care of this. First, we * 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 @@ -728,19 +820,14 @@ void backgroundRewriteDoneHandler(int exitcode, int bysignal) { * is, as long as the file descriptor is released again. */ if (server.appendfd == -1) { /* AOF disabled */ - struct stat st; - /* Check if the configured filename exists. If so, we need to open - * it to prevent rename(2) from unlinking it. */ - if (stat(server.appendfilename, &st) == ENOENT) { - oldfd = -1; - } else { - /* Don't care if this fails: oldfd will be -1. */ - oldfd = open(server.appendfilename,O_RDONLY|O_NONBLOCK); - } + /* 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. */ + oldfd = open(server.appendfilename,O_RDONLY|O_NONBLOCK); } else { /* AOF enabled */ - oldfd = -1; + oldfd = -1; /* We'll set this to the current AOF filedes later. */ } /* Rename the temporary file. This will not unlink the target file if @@ -749,28 +836,36 @@ void backgroundRewriteDoneHandler(int exitcode, int bysignal) { redisLog(REDIS_WARNING, "Error trying to rename the temporary AOF: %s", strerror(errno)); close(newfd); - if (oldfd != -1) - close(oldfd); + if (oldfd != -1) close(oldfd); goto cleanup; } if (server.appendfd == -1) { - /* AOF disabled */ + /* AOF disabled, we don't need to set the AOF file descriptor + * to this new file, so we can close it. */ close(newfd); } else { - /* AOF enabled */ + /* AOF enabled, replace the old fd with the new one. */ oldfd = server.appendfd; server.appendfd = newfd; - if (server.appendfsync != APPENDFSYNC_NO) aof_fsync(newfd); + if (server.appendfsync == APPENDFSYNC_ALWAYS) + aof_fsync(newfd); + else if (server.appendfsync == APPENDFSYNC_EVERYSEC) + aof_background_fsync(newfd); server.appendseldb = -1; /* Make sure SELECT is re-issued */ aofUpdateCurrentSize(); server.auto_aofrewrite_base_size = server.appendonly_current_size; + + /* Clear regular AOF buffer since its contents was just written to + * the new AOF from the background rewrite buffer. */ + sdsfree(server.aofbuf); + server.aofbuf = sdsempty(); } redisLog(REDIS_NOTICE, "Background AOF rewrite successful"); /* Asynchronously close the overwritten AOF. */ - if (oldfd != -1) eio_close(oldfd, 0, NULL, 0); + if (oldfd != -1) bioCreateBackgroundJob(REDIS_BIO_CLOSE_FILE,(void*)(long)oldfd,NULL,NULL); redisLog(REDIS_VERBOSE, "Background AOF rewrite signal handler took %lldus", ustime()-now);