X-Git-Url: https://git.saurik.com/redis.git/blobdiff_plain/f03fe802b79d1d41377016fd9755523dfae102e9..f797c7dc176c833e6aa412c557d7fedd59dc1124:/src/redis.h?ds=sidebyside diff --git a/src/redis.h b/src/redis.h index 6ff62916..91a64ecf 100644 --- a/src/redis.h +++ b/src/redis.h @@ -18,6 +18,7 @@ #include #include #include +#include #include "ae.h" /* Event driven programming library */ #include "sds.h" /* Dynamic safe strings */ @@ -70,6 +71,10 @@ #define REDIS_ZSET 3 #define REDIS_HASH 4 #define REDIS_VMPOINTER 8 +/* Object types only used for persistence in .rdb files */ +#define REDIS_HASH_ZIPMAP 9 +#define REDIS_LIST_ZIPLIST 10 +#define REDIS_SET_INTSET 11 /* Objects encoding. Some kind of objects like Strings and Hashes can be * internally represented in multiple ways. The 'encoding' field of the object @@ -136,6 +141,8 @@ #define REDIS_IO_WAIT 32 /* The client is waiting for Virtual Memory I/O */ #define REDIS_DIRTY_CAS 64 /* Watched keys modified. EXEC will fail. */ #define REDIS_CLOSE_AFTER_REPLY 128 /* Close after writing entire reply. */ +#define REDIS_UNBLOCKED 256 /* This client was unblocked and is stored in + server.unblocked_clients */ /* Client request types */ #define REDIS_REQ_INLINE 1 @@ -184,8 +191,8 @@ #define APPENDFSYNC_EVERYSEC 2 /* Zip structure related defaults */ -#define REDIS_HASH_MAX_ZIPMAP_ENTRIES 64 -#define REDIS_HASH_MAX_ZIPMAP_VALUE 512 +#define REDIS_HASH_MAX_ZIPMAP_ENTRIES 512 +#define REDIS_HASH_MAX_ZIPMAP_VALUE 64 #define REDIS_LIST_MAX_ZIPLIST_ENTRIES 512 #define REDIS_LIST_MAX_ZIPLIST_VALUE 64 #define REDIS_SET_MAX_INTSET_ENTRIES 512 @@ -354,19 +361,140 @@ struct sharedObjectsStruct { *integers[REDIS_SHARED_INTEGERS]; }; -/* Global server state structure */ +/*----------------------------------------------------------------------------- + * Redis cluster data structures + *----------------------------------------------------------------------------*/ + +#define REDIS_CLUSTER_SLOTS 4096 +#define REDIS_CLUSTER_OK 0 /* Everything looks ok */ +#define REDIS_CLUSTER_FAIL 1 /* The cluster can't work */ +#define REDIS_CLUSTER_NEEDHELP 2 /* The cluster works, but needs some help */ +#define REDIS_CLUSTER_NAMELEN 40 /* sha1 hex length */ +#define REDIS_CLUSTER_PORT_INCR 10000 /* Cluster port = baseport + PORT_INCR */ + +struct clusterNode; + +/* clusterLink encapsulates everything needed to talk with a remote node. */ +typedef struct clusterLink { + int fd; /* TCP socket file descriptor */ + sds sndbuf; /* Packet send buffer */ + sds rcvbuf; /* Packet reception buffer */ + struct clusterNode *node; /* Node related to this link if any, or NULL */ +} clusterLink; + +/* Node flags */ +#define REDIS_NODE_MASTER 1 /* The node is a master */ +#define REDIS_NODE_SLAVE 2 /* The node is a slave */ +#define REDIS_NODE_PFAIL 4 /* Failure? Need acknowledge */ +#define REDIS_NODE_FAIL 8 /* The node is believed to be malfunctioning */ +#define REDIS_NODE_MYSELF 16 /* This node is myself */ +#define REDIS_NODE_HANDSHAKE 32 /* We have still to exchange the first ping */ +#define REDIS_NODE_NOADDR 64 /* We don't know the address of this node */ +#define REDIS_NODE_MEET 128 /* Send a MEET message to this node */ +#define REDIS_NODE_NULL_NAME "\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000\000" + +struct clusterNode { + char name[REDIS_CLUSTER_NAMELEN]; /* Node name, hex string, sha1-size */ + int flags; /* REDIS_NODE_... */ + unsigned char slots[REDIS_CLUSTER_SLOTS/8]; /* slots handled by this node */ + int numslaves; /* Number of slave nodes, if this is a master */ + struct clusterNode **slaves; /* pointers to slave nodes */ + struct clusterNode *slaveof; /* pointer to the master node */ + time_t ping_sent; /* Unix time we sent latest ping */ + time_t pong_received; /* Unix time we received the pong */ + char *configdigest; /* Configuration digest of this node */ + time_t configdigest_ts; /* Configuration digest timestamp */ + char ip[16]; /* Latest known IP address of this node */ + int port; /* Latest known port of this node */ + clusterLink *link; /* TCP/IP link with this node */ +}; +typedef struct clusterNode clusterNode; + +typedef struct { + char *configfile; + clusterNode *myself; /* This node */ + int state; /* REDIS_CLUSTER_OK, REDIS_CLUSTER_FAIL, ... */ + int node_timeout; + dict *nodes; /* Hash table of name -> clusterNode structures */ + clusterNode *migrating_slots_to[REDIS_CLUSTER_SLOTS]; + clusterNode *importing_slots_from[REDIS_CLUSTER_SLOTS]; + clusterNode *slots[REDIS_CLUSTER_SLOTS]; +} clusterState; + +/* Redis cluster messages header */ + +/* Note that the PING, PONG and MEET messages are actually the same exact + * kind of packet. PONG is the reply to ping, in the extact format as a PING, + * while MEET is a special PING that forces the receiver to add the sender + * as a node (if it is not already in the list). */ +#define CLUSTERMSG_TYPE_PING 0 /* Ping */ +#define CLUSTERMSG_TYPE_PONG 1 /* Pong (reply to Ping) */ +#define CLUSTERMSG_TYPE_MEET 2 /* Meet "let's join" message */ +#define CLUSTERMSG_TYPE_FAIL 3 /* Mark node xxx as failing */ + +/* Initially we don't know our "name", but we'll find it once we connect + * to the first node, using the getsockname() function. Then we'll use this + * address for all the next messages. */ +typedef struct { + char nodename[REDIS_CLUSTER_NAMELEN]; + uint32_t ping_sent; + uint32_t pong_received; + char ip[16]; /* IP address last time it was seen */ + uint16_t port; /* port last time it was seen */ + uint16_t flags; + uint32_t notused; /* for 64 bit alignment */ +} clusterMsgDataGossip; + +typedef struct { + char nodename[REDIS_CLUSTER_NAMELEN]; +} clusterMsgDataFail; + +union clusterMsgData { + /* PING, MEET and PONG */ + struct { + /* Array of N clusterMsgDataGossip structures */ + clusterMsgDataGossip gossip[1]; + } ping; + /* FAIL */ + struct { + clusterMsgDataFail about; + } fail; +}; + +typedef struct { + uint32_t totlen; /* Total length of this message */ + uint16_t type; /* Message type */ + uint16_t count; /* Only used for some kind of messages. */ + char sender[REDIS_CLUSTER_NAMELEN]; /* Name of the sender node */ + unsigned char myslots[REDIS_CLUSTER_SLOTS/8]; + char slaveof[REDIS_CLUSTER_NAMELEN]; + char configdigest[32]; + uint16_t port; /* Sender TCP base port */ + unsigned char state; /* Cluster state from the POV of the sender */ + unsigned char notused[5]; /* Reserved for future use. For alignment. */ + union clusterMsgData data; +} clusterMsg; + +/*----------------------------------------------------------------------------- + * Global server state + *----------------------------------------------------------------------------*/ + struct redisServer { + /* General */ pthread_t mainthread; + redisDb *db; + dict *commands; /* Command table hahs table */ + aeEventLoop *el; + /* Networking */ int port; char *bindaddr; char *unixsocket; int ipfd; int sofd; - redisDb *db; - long long dirty; /* changes to DB from the last save */ - long long dirty_before_bgsave; /* used to restore dirty on failed BGSAVE */ + int cfd; list *clients; - dict *commands; /* Command table hahs table */ + list *slaves, *monitors; + char neterr[ANET_ERR_LEN]; /* RDB / AOF loading information */ int loading; off_t loading_total_bytes; @@ -374,9 +502,6 @@ struct redisServer { time_t loading_start_time; /* Fast pointers to often looked up command */ struct redisCommand *delCommand, *multiCommand; - list *slaves, *monitors; - char neterr[ANET_ERR_LEN]; - aeEventLoop *el; int cronloops; /* number of times the cron function run */ time_t lastsave; /* Unix time of last save succeeede */ /* Fields used only for stats */ @@ -399,6 +524,8 @@ struct redisServer { int activerehashing; char *requirepass; /* Persistence */ + long long dirty; /* changes to DB from the last save */ + long long dirty_before_bgsave; /* used to restore dirty on failed BGSAVE */ time_t lastfsync; int appendfd; int appendseldb; @@ -491,6 +618,8 @@ struct redisServer { /* Misc */ unsigned lruclock:22; /* clock incrementing every minute, for LRU */ unsigned lruclock_padding:10; + int cluster_enabled; + clusterState cluster; }; typedef struct pubsubPattern { @@ -499,20 +628,20 @@ typedef struct pubsubPattern { } pubsubPattern; typedef void redisCommandProc(redisClient *c); -typedef void redisVmPreloadProc(redisClient *c, struct redisCommand *cmd, int argc, robj **argv); +typedef int *redisGetKeysProc(struct redisCommand *cmd, robj **argv, int argc, int *numkeys, int flags); struct redisCommand { char *name; redisCommandProc *proc; int arity; int flags; - /* Use a function to determine which keys need to be loaded - * in the background prior to executing this command. Takes precedence - * over vm_firstkey and others, ignored when NULL */ - redisVmPreloadProc *vm_preload_proc; + /* Use a function to determine keys arguments in a command line. + * Used both for diskstore preloading and Redis Cluster. */ + redisGetKeysProc *getkeys_proc; /* What keys should be loaded in background when calling this command? */ - int vm_firstkey; /* The first argument that's a key (0 = no keys) */ - int vm_lastkey; /* THe last argument that's a key */ - int vm_keystep; /* The step between first and last key */ + int firstkey; /* The first argument that's a key (0 = no keys) */ + int lastkey; /* THe last argument that's a key */ + int keystep; /* The step between first and last key */ + long long microseconds, calls; }; struct redisFunctionSym { @@ -625,6 +754,7 @@ extern struct redisServer server; extern struct sharedObjectsStruct shared; extern dictType setDictType; extern dictType zsetDictType; +extern dictType clusterNodesDictType; extern double R_Zero, R_PosInf, R_NegInf, R_Nan; dictType hashDictType; @@ -632,6 +762,9 @@ dictType hashDictType; * Functions prototypes *----------------------------------------------------------------------------*/ +/* Utils */ +long long ustime(void); + /* networking.c -- Networking and Client related operations */ redisClient *createClient(int fd); void closeTimedoutClients(void); @@ -659,6 +792,8 @@ void addReplyDouble(redisClient *c, double d); void addReplyLongLong(redisClient *c, long long ll); void addReplyMultiBulkLen(redisClient *c, long length); void *dupClientReplyValue(void *o); +void getClientsMaxBuffers(unsigned long *longest_output_list, + unsigned long *biggest_input_buffer); #ifdef __GNUC__ void addReplyErrorFormat(redisClient *c, const char *fmt, ...) @@ -734,6 +869,7 @@ int fwriteBulkString(FILE *fp, char *s, unsigned long len); int fwriteBulkDouble(FILE *fp, double d); int fwriteBulkLongLong(FILE *fp, long long l); int fwriteBulkObject(FILE *fp, robj *obj); +int fwriteBulkCount(FILE *fp, char prefix, int count); /* Replication */ void replicationFeedSlaves(list *slaves, int dictid, robj **argv, int argc); @@ -757,7 +893,7 @@ off_t rdbSavedObjectLen(robj *o); off_t rdbSavedObjectPages(robj *o); robj *rdbLoadObject(int type, FILE *fp); void backgroundSaveDoneHandler(int exitcode, int bysignal); -int rdbSaveKeyValuePair(FILE *fp, redisDb *db, robj *key, robj *val, time_t now); +int rdbSaveKeyValuePair(FILE *fp, robj *key, robj *val, time_t expireitme, time_t now); int rdbLoadType(FILE *fp); time_t rdbLoadTime(FILE *fp); robj *rdbLoadStringObject(FILE *fp); @@ -782,7 +918,7 @@ zskiplistNode *zslInsert(zskiplist *zsl, double score, robj *obj); /* Core functions */ void freeMemoryIfNeeded(void); int processCommand(redisClient *c); -void setupSigSegvAction(void); +void setupSignalHandlers(void); struct redisCommand *lookupCommand(sds name); struct redisCommand *lookupCommandByCString(char *s); void call(redisClient *c, struct redisCommand *cmd); @@ -793,15 +929,17 @@ void updateDictResizePolicy(void); int htNeedsResize(dict *dict); void oom(const char *msg); void populateCommandTable(void); +void resetCommandTableStats(void); /* Disk store */ int dsOpen(void); int dsClose(void); -int dsSet(redisDb *db, robj *key, robj *val); +int dsSet(redisDb *db, robj *key, robj *val, time_t expire); robj *dsGet(redisDb *db, robj *key, time_t *expire); int dsDel(redisDb *db, robj *key); int dsExists(redisDb *db, robj *key); void dsFlushDb(int dbid); +int dsRdbSaveBackground(char *filename); int dsRdbSave(char *filename); /* Disk Store Cache */ @@ -813,8 +951,6 @@ void freeIOJob(iojob *j); void queueIOJob(iojob *j); void waitEmptyIOJobsQueue(void); void processAllPendingIOJobs(void); -void zunionInterBlockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd, int argc, robj **argv); -void execBlockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd, int argc, robj **argv); int blockClientOnSwappedKeys(redisClient *c, struct redisCommand *cmd); int dontWaitForSwappedKey(redisClient *c, robj *key); void handleClientsBlockedOnSwappedKey(redisDb *db, robj *key); @@ -901,6 +1037,24 @@ int selectDb(redisClient *c, int id); void signalModifiedKey(redisDb *db, robj *key); void signalFlushedDb(int dbid); +/* API to get key arguments from commands */ +#define REDIS_GETKEYS_ALL 0 +#define REDIS_GETKEYS_PRELOAD 1 +int *getKeysFromCommand(struct redisCommand *cmd, robj **argv, int argc, int *numkeys, int flags); +void getKeysFreeResult(int *result); +int *noPreloadGetKeys(struct redisCommand *cmd,robj **argv, int argc, int *numkeys, int flags); +int *renameGetKeys(struct redisCommand *cmd,robj **argv, int argc, int *numkeys, int flags); +int *zunionInterGetKeys(struct redisCommand *cmd,robj **argv, int argc, int *numkeys, int flags); + +/* Cluster */ +void clusterInit(void); +unsigned short crc16(const char *buf, int len); +unsigned int keyHashSlot(char *key, int keylen); +clusterNode *createClusterNode(char *nodename, int flags); +int clusterAddNode(clusterNode *node); +void clusterCron(void); +clusterNode *getNodeByQuery(redisClient *c, struct redisCommand *cmd, robj **argv, int argc, int *hashslot); + /* Git SHA1 */ char *redisGitSHA1(void); char *redisGitDirty(void); @@ -1023,6 +1177,10 @@ void punsubscribeCommand(redisClient *c); void publishCommand(redisClient *c); void watchCommand(redisClient *c); void unwatchCommand(redisClient *c); +void clusterCommand(redisClient *c); +void restoreCommand(redisClient *c); +void migrateCommand(redisClient *c); +void dumpCommand(redisClient *c); #if defined(__GNUC__) void *calloc(size_t count, size_t size) __attribute__ ((deprecated));