#include <sys/time.h>
#include <unistd.h>
#include <fcntl.h>
+#include <sys/socket.h>
#include <sys/stat.h>
/* ---------------------------------- MASTER -------------------------------- */
gettimeofday(&tv,NULL);
cmdrepr = sdscatprintf(cmdrepr,"%ld.%06ld ",(long)tv.tv_sec,(long)tv.tv_usec);
if (c->flags & REDIS_LUA_CLIENT) {
- cmdrepr = sdscatprintf(cmdrepr,"[%d lua] ", dictid);
+ cmdrepr = sdscatprintf(cmdrepr,"[%d lua] ",dictid);
+ } else if (c->flags & REDIS_UNIX_SOCKET) {
+ cmdrepr = sdscatprintf(cmdrepr,"[%d unix:%s] ",dictid,server.unixsocket);
} else {
anetPeerToString(c->fd,ip,&port);
- cmdrepr = sdscatprintf(cmdrepr,"[%d %s:%d] ", dictid,ip,port);
+ cmdrepr = sdscatprintf(cmdrepr,"[%d %s:%d] ",dictid,ip,port);
}
for (j = 0; j < argc; j++) {
return;
}
+/* REPLCONF <option> <value> <option> <value> ...
+ * This command is used by a slave in order to configure the replication
+ * process before starting it with the SYNC command.
+ *
+ * Currently the only use of this command is to communicate to the master
+ * what is the listening port of the Slave redis instance, so that the
+ * master can accurately list slaves and their listening ports in
+ * the INFO output.
+ *
+ * In the future the same command can be used in order to configure
+ * the replication to initiate an incremental replication instead of a
+ * full resync. */
+void replconfCommand(redisClient *c) {
+ int j;
+
+ if ((c->argc % 2) == 0) {
+ /* Number of arguments must be odd to make sure that every
+ * option has a corresponding value. */
+ addReply(c,shared.syntaxerr);
+ return;
+ }
+
+ /* Process every option-value pair. */
+ for (j = 1; j < c->argc; j+=2) {
+ if (!strcasecmp(c->argv[j]->ptr,"listening-port")) {
+ long port;
+
+ if ((getLongFromObjectOrReply(c,c->argv[j+1],
+ &port,NULL) != REDIS_OK))
+ return;
+ c->slave_listening_port = port;
+ } else {
+ addReplyErrorFormat(c,"Unrecognized REPLCONF option: %s",
+ (char*)c->argv[j]->ptr);
+ return;
+ }
+ }
+ addReply(c,shared.ok);
+}
+
void sendBulkToSlave(aeEventLoop *el, int fd, void *privdata, int mask) {
redisClient *slave = privdata;
REDIS_NOTUSED(el);
}
/* Asynchronously read the SYNC payload we receive from a master */
+#define REPL_MAX_WRITTEN_BEFORE_FSYNC (1024*1024*8) /* 8 MB */
void readSyncBulkPayload(aeEventLoop *el, int fd, void *privdata, int mask) {
char buf[4096];
ssize_t nread, readlen;
+ off_t left;
REDIS_NOTUSED(el);
REDIS_NOTUSED(privdata);
REDIS_NOTUSED(mask);
- /* If repl_transfer_left == -1 we still have to read the bulk length
+ /* If repl_transfer_size == -1 we still have to read the bulk length
* from the master reply. */
- if (server.repl_transfer_left == -1) {
+ if (server.repl_transfer_size == -1) {
if (syncReadLine(fd,buf,1024,server.repl_syncio_timeout*1000) == -1) {
redisLog(REDIS_WARNING,
"I/O error reading bulk count from MASTER: %s",
redisLog(REDIS_WARNING,"Bad protocol from MASTER, the first byte is not '$', are you sure the host and port are right?");
goto error;
}
- server.repl_transfer_left = strtol(buf+1,NULL,10);
+ server.repl_transfer_size = strtol(buf+1,NULL,10);
redisLog(REDIS_NOTICE,
"MASTER <-> SLAVE sync: receiving %ld bytes from master",
- server.repl_transfer_left);
+ server.repl_transfer_size);
return;
}
/* Read bulk data */
- readlen = (server.repl_transfer_left < (signed)sizeof(buf)) ?
- server.repl_transfer_left : (signed)sizeof(buf);
+ left = server.repl_transfer_size - server.repl_transfer_read;
+ readlen = (left < (signed)sizeof(buf)) ? left : (signed)sizeof(buf);
nread = read(fd,buf,readlen);
if (nread <= 0) {
redisLog(REDIS_WARNING,"I/O error trying to sync with MASTER: %s",
redisLog(REDIS_WARNING,"Write error or short write writing to the DB dump file needed for MASTER <-> SLAVE synchronization: %s", strerror(errno));
goto error;
}
- server.repl_transfer_left -= nread;
+ server.repl_transfer_read += nread;
+
+ /* Sync data on disk from time to time, otherwise at the end of the transfer
+ * we may suffer a big delay as the memory buffers are copied into the
+ * actual disk. */
+ if (server.repl_transfer_read >=
+ server.repl_transfer_last_fsync_off + REPL_MAX_WRITTEN_BEFORE_FSYNC)
+ {
+ off_t sync_size = server.repl_transfer_read -
+ server.repl_transfer_last_fsync_off;
+ rdb_fsync_range(server.repl_transfer_fd,
+ server.repl_transfer_last_fsync_off, sync_size);
+ server.repl_transfer_last_fsync_off += sync_size;
+ }
+
/* Check if the transfer is now complete */
- if (server.repl_transfer_left == 0) {
+ if (server.repl_transfer_read == server.repl_transfer_size) {
if (rename(server.repl_transfer_tmpfile,server.rdb_filename) == -1) {
redisLog(REDIS_WARNING,"Failed trying to rename the temp DB into dump.rdb in MASTER <-> SLAVE synchronization: %s", strerror(errno));
replicationAbortSyncTransfer();
return;
}
+/* Send a synchronous command to the master. Used to send AUTH and
+ * REPLCONF commands before starting the replication with SYNC.
+ *
+ * On success NULL is returned.
+ * On error an sds string describing the error is returned.
+ */
+char *sendSynchronousCommand(int fd, ...) {
+ va_list ap;
+ sds cmd = sdsempty();
+ char *arg, buf[256];
+
+ /* Create the command to send to the master, we use simple inline
+ * protocol for simplicity as currently we only send simple strings. */
+ va_start(ap,fd);
+ while(1) {
+ arg = va_arg(ap, char*);
+ if (arg == NULL) break;
+
+ if (sdslen(cmd) != 0) cmd = sdscatlen(cmd," ",1);
+ cmd = sdscat(cmd,arg);
+ }
+ cmd = sdscatlen(cmd,"\r\n",2);
+
+ /* Transfer command to the server. */
+ if (syncWrite(fd,cmd,sdslen(cmd),server.repl_syncio_timeout*1000) == -1) {
+ sdsfree(cmd);
+ return sdscatprintf(sdsempty(),"Writing to master: %s",
+ strerror(errno));
+ }
+ sdsfree(cmd);
+
+ /* Read the reply from the server. */
+ if (syncReadLine(fd,buf,sizeof(buf),server.repl_syncio_timeout*1000) == -1)
+ {
+ return sdscatprintf(sdsempty(),"Reading from master: %s",
+ strerror(errno));
+ }
+
+ /* Check for errors from the server. */
+ if (buf[0] != '+') {
+ return sdscatprintf(sdsempty(),"Error from master: %s", buf);
+ }
+
+ return NULL; /* No errors. */
+}
+
void syncWithMaster(aeEventLoop *el, int fd, void *privdata, int mask) {
- char buf[1024], tmpfile[256];
+ char tmpfile[256], *err;
int dfd, maxtries = 5;
+ int sockerr = 0;
+ socklen_t errlen = sizeof(sockerr);
REDIS_NOTUSED(el);
REDIS_NOTUSED(privdata);
REDIS_NOTUSED(mask);
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_READABLE|AE_WRITABLE);
+ /* Check for errors in the socket. */
+ if (getsockopt(fd, SOL_SOCKET, SO_ERROR, &sockerr, &errlen) == -1)
+ sockerr = errno;
+ if (sockerr) {
+ aeDeleteFileEvent(server.el,fd,AE_READABLE|AE_WRITABLE);
+ redisLog(REDIS_WARNING,"Error condition on socket for SYNC: %s",
+ strerror(sockerr));
+ goto error;
+ }
- /* AUTH with the master if required. */
- if(server.masterauth) {
- char authcmd[1024];
- size_t authlen;
+ /* If we were connecting, it's time to send a non blocking PING, we want to
+ * make sure the master is able to reply before going into the actual
+ * replication process where we have long timeouts in the order of
+ * seconds (in the meantime the slave would block). */
+ if (server.repl_state == REDIS_REPL_CONNECTING) {
+ redisLog(REDIS_NOTICE,"Non blocking connect for SYNC fired the event.");
+ /* Delete the writable event so that the readable event remains
+ * registered and we can wait for the PONG reply. */
+ aeDeleteFileEvent(server.el,fd,AE_WRITABLE);
+ server.repl_state = REDIS_REPL_RECEIVE_PONG;
+ /* Send the PING, don't check for errors at all, we have the timeout
+ * that will take care about this. */
+ syncWrite(fd,"PING\r\n",6,100);
+ return;
+ }
+
+ /* Receive the PONG command. */
+ if (server.repl_state == REDIS_REPL_RECEIVE_PONG) {
+ char buf[1024];
+
+ /* Delete the readable event, we no longer need it now that there is
+ * the PING reply to read. */
+ aeDeleteFileEvent(server.el,fd,AE_READABLE);
- authlen = snprintf(authcmd,sizeof(authcmd),"AUTH %s\r\n",server.masterauth);
- if (syncWrite(fd,authcmd,authlen,server.repl_syncio_timeout*1000) == -1) {
- redisLog(REDIS_WARNING,"Unable to AUTH to MASTER: %s",
+ /* Read the reply with explicit timeout. */
+ buf[0] = '\0';
+ if (syncReadLine(fd,buf,sizeof(buf),
+ server.repl_syncio_timeout*1000) == -1)
+ {
+ redisLog(REDIS_WARNING,
+ "I/O error reading PING reply from master: %s",
strerror(errno));
goto error;
}
- /* Read the AUTH result. */
- if (syncReadLine(fd,buf,1024,server.repl_syncio_timeout*1000) == -1) {
- redisLog(REDIS_WARNING,"I/O error reading auth result from MASTER: %s",
- strerror(errno));
+
+ /* We don't care about the reply, it can be +PONG or an error since
+ * the server requires AUTH. As long as it replies correctly, it's
+ * fine from our point of view. */
+ if (buf[0] != '-' && buf[0] != '+') {
+ redisLog(REDIS_WARNING,"Unexpected reply to PING from master.");
goto error;
+ } else {
+ redisLog(REDIS_NOTICE,
+ "Master replied to PING, replication can continue...");
}
- if (buf[0] != '+') {
- redisLog(REDIS_WARNING,"Cannot AUTH to MASTER, is the masterauth password correct?");
+ }
+
+ /* AUTH with the master if required. */
+ if(server.masterauth) {
+ err = sendSynchronousCommand(fd,"AUTH",server.masterauth,NULL);
+ if (err) {
+ redisLog(REDIS_WARNING,"Unable to AUTH to MASTER: %s",err);
+ sdsfree(err);
goto error;
}
}
+ /* Set the slave port, so that Master's INFO command can list the
+ * slave listening port correctly. */
+ {
+ sds port = sdsfromlonglong(server.port);
+ err = sendSynchronousCommand(fd,"REPLCONF","listening-port",port,
+ NULL);
+ sdsfree(port);
+ /* Ignore the error if any, not all the Redis versions support
+ * REPLCONF listening-port. */
+ if (err) {
+ redisLog(REDIS_NOTICE,"(non critical): Master does not understand REPLCONF listening-port: %s", err);
+ sdsfree(err);
+ }
+ }
+
/* Issue the SYNC command */
if (syncWrite(fd,"SYNC\r\n",6,server.repl_syncio_timeout*1000) == -1) {
redisLog(REDIS_WARNING,"I/O error writing to MASTER: %s",
}
server.repl_state = REDIS_REPL_TRANSFER;
- server.repl_transfer_left = -1;
+ server.repl_transfer_size = -1;
+ server.repl_transfer_read = 0;
+ server.repl_transfer_last_fsync_off = 0;
server.repl_transfer_fd = dfd;
server.repl_transfer_lastio = server.unixtime;
server.repl_transfer_tmpfile = zstrdup(tmpfile);
return;
error:
- server.repl_state = REDIS_REPL_CONNECT;
close(fd);
+ server.repl_transfer_s = -1;
+ server.repl_state = REDIS_REPL_CONNECT;
return;
}
void undoConnectWithMaster(void) {
int fd = server.repl_transfer_s;
- redisAssert(server.repl_state == REDIS_REPL_CONNECTING);
+ redisAssert(server.repl_state == REDIS_REPL_CONNECTING ||
+ server.repl_state == REDIS_REPL_RECEIVE_PONG);
aeDeleteFileEvent(server.el,fd,AE_READABLE|AE_WRITABLE);
close(fd);
server.repl_transfer_s = -1;
if (server.master) freeClient(server.master);
if (server.repl_state == REDIS_REPL_TRANSFER)
replicationAbortSyncTransfer();
- else if (server.repl_state == REDIS_REPL_CONNECTING)
+ else if (server.repl_state == REDIS_REPL_CONNECTING ||
+ server.repl_state == REDIS_REPL_RECEIVE_PONG)
undoConnectWithMaster();
server.repl_state = REDIS_REPL_NONE;
redisLog(REDIS_NOTICE,"MASTER MODE enabled (user request)");
void replicationCron(void) {
/* Non blocking connection timeout? */
- if (server.masterhost && server.repl_state == REDIS_REPL_CONNECTING &&
+ if (server.masterhost &&
+ (server.repl_state == REDIS_REPL_CONNECTING ||
+ server.repl_state == REDIS_REPL_RECEIVE_PONG) &&
(time(NULL)-server.repl_transfer_lastio) > server.repl_timeout)
{
redisLog(REDIS_WARNING,"Timeout connecting to the MASTER...");
if (server.masterhost && server.repl_state == REDIS_REPL_TRANSFER &&
(time(NULL)-server.repl_transfer_lastio) > server.repl_timeout)
{
- redisLog(REDIS_WARNING,"Timeout receiving bulk data from MASTER...");
+ redisLog(REDIS_WARNING,"Timeout receiving bulk data from MASTER... If the problem persists try to set the 'repl-timeout' parameter in redis.conf to a larger value.");
replicationAbortSyncTransfer();
}
* 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 % (server.repl_ping_slave_period*10))) {
+ if (!(server.cronloops % (server.repl_ping_slave_period * REDIS_HZ))) {
listIter li;
listNode *ln;
projects / redis.git / blobdiff