X-Git-Url: https://git.saurik.com/redis.git/blobdiff_plain/890a2ed989274cb09b5cde1def3935e110ec3cb9..087f41409052da35270417c85010c6739fd28200:/src/replication.c diff --git a/src/replication.c b/src/replication.c index 87d575ca..2be03c1b 100644 --- a/src/replication.c +++ b/src/replication.c @@ -10,38 +10,8 @@ void replicationFeedSlaves(list *slaves, int dictid, robj **argv, int argc) { listNode *ln; listIter li; - int outc = 0, j; - robj **outv; - /* We need 1+(ARGS*3) objects since commands are using the new protocol - * and we one 1 object for the first "*\r\n" multibulk count, then - * for every additional object we have "$\r\n" + object + "\r\n". */ - robj *static_outv[REDIS_STATIC_ARGS*3+1]; - robj *lenobj; - - if (argc <= REDIS_STATIC_ARGS) { - outv = static_outv; - } else { - outv = zmalloc(sizeof(robj*)*(argc*3+1)); - } - - lenobj = createObject(REDIS_STRING, - sdscatprintf(sdsempty(), "*%d\r\n", argc)); - lenobj->refcount = 0; - outv[outc++] = lenobj; - for (j = 0; j < argc; j++) { - lenobj = createObject(REDIS_STRING, - sdscatprintf(sdsempty(),"$%lu\r\n", - (unsigned long) stringObjectLen(argv[j]))); - lenobj->refcount = 0; - outv[outc++] = lenobj; - outv[outc++] = argv[j]; - outv[outc++] = shared.crlf; - } + int j; - /* Increment all the refcounts at start and decrement at end in order to - * be sure to free objects if there is no slave in a replication state - * able to be feed with commands */ - for (j = 0; j < outc; j++) incrRefCount(outv[j]); listRewind(slaves,&li); while((ln = listNext(&li))) { redisClient *slave = ln->value; @@ -49,7 +19,9 @@ void replicationFeedSlaves(list *slaves, int dictid, robj **argv, int argc) { /* Don't feed slaves that are still waiting for BGSAVE to start */ if (slave->replstate == REDIS_REPL_WAIT_BGSAVE_START) continue; - /* Feed all the other slaves, MONITORs and so on */ + /* Feed slaves that are waiting for the initial SYNC (so these commands + * are queued in the output buffer until the intial SYNC completes), + * or are already in sync with the master. */ if (slave->slaveseldb != dictid) { robj *selectcmd; @@ -73,10 +45,9 @@ void replicationFeedSlaves(list *slaves, int dictid, robj **argv, int argc) { addReply(slave,selectcmd); slave->slaveseldb = dictid; } - for (j = 0; j < outc; j++) addReply(slave,outv[j]); + addReplyMultiBulkLen(slave,argc); + for (j = 0; j < argc; j++) addReplyBulk(slave,argv[j]); } - for (j = 0; j < outc; j++) decrRefCount(outv[j]); - if (outv != static_outv) zfree(outv); } void replicationFeedMonitors(list *monitors, int dictid, robj **argv, int argc) { @@ -387,6 +358,22 @@ void readSyncBulkPayload(aeEventLoop *el, int fd, void *privdata, int mask) { server.master->authenticated = 1; server.replstate = REDIS_REPL_CONNECTED; redisLog(REDIS_NOTICE, "MASTER <-> SLAVE sync: Finished with success"); + /* Restart the AOF subsystem now that we finished the sync. This + * will trigger an AOF rewrite, and when done will start appending + * to the new file. */ + if (server.appendonly) { + int retry = 10; + + stopAppendOnly(); + while (retry-- && startAppendOnly() == REDIS_ERR) { + redisLog(REDIS_WARNING,"Failed enabling the AOF after successful master synchrnization! Trying it again in one second."); + sleep(1); + } + if (!retry) { + redisLog(REDIS_WARNING,"FATAL: this slave instance finished the synchronization with its master, but the AOF can't be turned on. Exiting now."); + exit(1); + } + } } return; @@ -403,10 +390,18 @@ void syncWithMaster(aeEventLoop *el, int fd, void *privdata, int mask) { REDIS_NOTUSED(privdata); REDIS_NOTUSED(mask); + /* If this event fired after the user turned the instance into a master + * with SLAVEOF NO ONE we must just return ASAP. */ + if (server.replstate == REDIS_REPL_NONE) { + close(fd); + return; + } + + redisLog(REDIS_NOTICE,"Non blocking connect for SYNC fired the event."); /* This event should only be triggered once since it is used to have a * non-blocking connect(2) to the master. It has been triggered when this * function is called, so we can delete it. */ - aeDeleteFileEvent(server.el,fd,AE_WRITABLE); + aeDeleteFileEvent(server.el,fd,AE_READABLE|AE_WRITABLE); /* AUTH with the master if required. */ if(server.masterauth) { @@ -482,7 +477,7 @@ int connectWithMaster(void) { return REDIS_ERR; } - if (aeCreateFileEvent(server.el,fd,AE_WRITABLE,syncWithMaster,NULL) == + if (aeCreateFileEvent(server.el,fd,AE_READABLE|AE_WRITABLE,syncWithMaster,NULL) == AE_ERR) { close(fd); @@ -490,11 +485,24 @@ int connectWithMaster(void) { return REDIS_ERR; } + server.repl_transfer_lastio = time(NULL); server.repl_transfer_s = fd; server.replstate = REDIS_REPL_CONNECTING; return REDIS_OK; } +/* This function can be called when a non blocking connection is currently + * in progress to undo it. */ +void undoConnectWithMaster(void) { + int fd = server.repl_transfer_s; + + redisAssert(server.replstate == REDIS_REPL_CONNECTING); + aeDeleteFileEvent(server.el,fd,AE_READABLE|AE_WRITABLE); + close(fd); + server.repl_transfer_s = -1; + server.replstate = REDIS_REPL_CONNECT; +} + void slaveofCommand(redisClient *c) { if (!strcasecmp(c->argv[1]->ptr,"no") && !strcasecmp(c->argv[2]->ptr,"one")) { @@ -504,6 +512,8 @@ void slaveofCommand(redisClient *c) { if (server.master) freeClient(server.master); if (server.replstate == REDIS_REPL_TRANSFER) replicationAbortSyncTransfer(); + else if (server.replstate == REDIS_REPL_CONNECTING) + undoConnectWithMaster(); server.replstate = REDIS_REPL_NONE; redisLog(REDIS_NOTICE,"MASTER MODE enabled (user request)"); } @@ -523,13 +533,18 @@ void slaveofCommand(redisClient *c) { /* --------------------------- REPLICATION CRON ---------------------------- */ -#define REDIS_REPL_TIMEOUT 60 -#define REDIS_REPL_PING_SLAVE_PERIOD 10 - void replicationCron(void) { + /* Non blocking connection timeout? */ + if (server.masterhost && server.replstate == REDIS_REPL_CONNECTING && + (time(NULL)-server.repl_transfer_lastio) > server.repl_timeout) + { + redisLog(REDIS_WARNING,"Timeout connecting to the MASTER..."); + undoConnectWithMaster(); + } + /* Bulk transfer I/O timeout? */ if (server.masterhost && server.replstate == REDIS_REPL_TRANSFER && - (time(NULL)-server.repl_transfer_lastio) > REDIS_REPL_TIMEOUT) + (time(NULL)-server.repl_transfer_lastio) > server.repl_timeout) { redisLog(REDIS_WARNING,"Timeout receiving bulk data from MASTER..."); replicationAbortSyncTransfer(); @@ -537,7 +552,7 @@ void replicationCron(void) { /* Timed out master when we are an already connected slave? */ if (server.masterhost && server.replstate == REDIS_REPL_CONNECTED && - (time(NULL)-server.master->lastinteraction) > REDIS_REPL_TIMEOUT) + (time(NULL)-server.master->lastinteraction) > server.repl_timeout) { redisLog(REDIS_WARNING,"MASTER time out: no data nor PING received..."); freeClient(server.master); @@ -548,7 +563,6 @@ void replicationCron(void) { redisLog(REDIS_NOTICE,"Connecting to MASTER..."); if (connectWithMaster() == REDIS_OK) { redisLog(REDIS_NOTICE,"MASTER <-> SLAVE sync started"); - if (server.appendonly) rewriteAppendOnlyFileBackground(); } } @@ -556,7 +570,7 @@ void replicationCron(void) { * So slaves can implement an explicit timeout to masters, and will * be able to detect a link disconnection even if the TCP connection * will not actually go down. */ - if (!(server.cronloops % (REDIS_REPL_PING_SLAVE_PERIOD*10))) { + if (!(server.cronloops % (server.repl_ping_slave_period*10))) { listIter li; listNode *ln;