#include <inttypes.h>
#include <pthread.h>
#include <syslog.h>
+#include <netinet/in.h>
#include "ae.h" /* Event driven programming library */
#include "sds.h" /* Dynamic safe strings */
#include "ziplist.h" /* Compact list data structure */
#include "intset.h" /* Compact integer set structure */
#include "version.h"
+#include "util.h"
/* Error codes */
#define REDIS_OK 0
#define REDIS_MAXIDLETIME (60*5) /* default client timeout */
#define REDIS_IOBUF_LEN 1024
#define REDIS_LOADBUF_LEN 1024
-#define REDIS_STATIC_ARGS 8
#define REDIS_DEFAULT_DBNUM 16
#define REDIS_CONFIGLINE_MAX 1024
#define REDIS_MAX_SYNC_TIME 60 /* Slave can't take more to sync */
#define REDIS_SHARED_INTEGERS 10000
#define REDIS_REPLY_CHUNK_BYTES (5*1500) /* 5 TCP packets with default MTU */
#define REDIS_MAX_LOGMSG_LEN 1024 /* Default maximum length of syslog messages */
+#define REDIS_AUTO_AOFREWRITE_PERC 100
+#define REDIS_AUTO_AOFREWRITE_MIN_SIZE (1024*1024)
/* Hash table parameters */
#define REDIS_HT_MINFILL 10 /* Minimal hash table fill 10% */
#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
+#define REDIS_ZSET_ZIPLIST 12
+
/* Objects encoding. Some kind of objects like Strings and Hashes can be
* internally represented in multiple ways. The 'encoding' field of the object
* is set to one of this fields for this object. */
#define REDIS_ENCODING_LINKEDLIST 4 /* Encoded as regular linked list */
#define REDIS_ENCODING_ZIPLIST 5 /* Encoded as ziplist */
#define REDIS_ENCODING_INTSET 6 /* Encoded as intset */
+#define REDIS_ENCODING_SKIPLIST 7 /* Encoded as skiplist */
/* Object types only used for dumping to disk */
#define REDIS_EXPIRETIME 253
#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
#define REDIS_REQ_MULTIBULK 2
/* Slave replication state - slave side */
-#define REDIS_REPL_NONE 0 /* No active replication */
-#define REDIS_REPL_CONNECT 1 /* Must connect to master */
-#define REDIS_REPL_TRANSFER 2 /* Receiving .rdb from master */
-#define REDIS_REPL_CONNECTED 3 /* Connected to master */
+#define REDIS_REPL_NONE 0 /* No active replication */
+#define REDIS_REPL_CONNECT 1 /* Must connect to master */
+#define REDIS_REPL_CONNECTING 2 /* Connecting to master */
+#define REDIS_REPL_TRANSFER 3 /* Receiving .rdb from master */
+#define REDIS_REPL_CONNECTED 4 /* Connected to master */
+
+/* Synchronous read timeout - slave side */
+#define REDIS_REPL_SYNCIO_TIMEOUT 5
/* Slave replication state - from the point of view of master
* Note that in SEND_BULK and ONLINE state the slave receives new updates
#define REDIS_VERBOSE 1
#define REDIS_NOTICE 2
#define REDIS_WARNING 3
+#define REDIS_LOG_RAW (1<<10) /* Modifier to log without timestamp */
/* Anti-warning macro... */
#define REDIS_NOTUSED(V) ((void) V)
#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
+#define REDIS_ZSET_MAX_ZIPLIST_ENTRIES 128
+#define REDIS_ZSET_MAX_ZIPLIST_VALUE 64
/* Sets operations codes */
#define REDIS_OP_UNION 0
*integers[REDIS_SHARED_INTEGERS];
};
-/* Global server state structure */
+/* ZSETs use a specialized version of Skiplists */
+typedef struct zskiplistNode {
+ robj *obj;
+ double score;
+ struct zskiplistNode *backward;
+ struct zskiplistLevel {
+ struct zskiplistNode *forward;
+ unsigned int span;
+ } level[];
+} zskiplistNode;
+
+typedef struct zskiplist {
+ struct zskiplistNode *header, *tail;
+ unsigned long length;
+ int level;
+} zskiplist;
+
+typedef struct zset {
+ dict *dict;
+ zskiplist *zsl;
+} zset;
+
+/*-----------------------------------------------------------------------------
+ * 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];
+ zskiplist *slots_to_keys;
+} 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;
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 */
long long stat_evictedkeys; /* number of evicted keys (maxmemory) */
long long stat_keyspace_hits; /* number of successful lookups of keys */
long long stat_keyspace_misses; /* number of failed lookups of keys */
+ size_t stat_peak_memory; /* max used memory record */
+ long long stat_fork_time; /* time needed to perform latets fork() */
/* Configuration */
int verbosity;
int maxidletime;
int appendonly;
int appendfsync;
int no_appendfsync_on_rewrite;
+ int auto_aofrewrite_perc; /* Rewrite AOF if % growth is > M and... */
+ off_t auto_aofrewrite_min_size; /* the AOF file is at least N bytes. */
+ off_t auto_aofrewrite_base_size;/* AOF size on latest startup or rewrite. */
+ off_t appendonly_current_size; /* AOF current size. */
+ int aofrewrite_scheduled; /* Rewrite once BGSAVE terminates. */
int shutdown_asap;
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;
char *masterhost;
int masterport;
redisClient *master; /* client that is master for this slave */
+ int repl_syncio_timeout; /* timeout for synchronous I/O calls */
int replstate; /* replication status if the instance is a slave */
off_t repl_transfer_left; /* bytes left reading .rdb */
int repl_transfer_s; /* slave -> master SYNC socket */
size_t list_max_ziplist_entries;
size_t list_max_ziplist_value;
size_t set_max_intset_entries;
+ size_t zset_max_ziplist_entries;
+ size_t zset_max_ziplist_value;
time_t unixtime; /* Unix time sampled every second. */
/* Virtual memory I/O threads stuff */
/* An I/O thread process an element taken from the io_jobs queue and
/* Misc */
unsigned lruclock:22; /* clock incrementing every minute, for LRU */
unsigned lruclock_padding:10;
+ /* Cluster */
+ int cluster_enabled;
+ clusterState cluster;
};
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 {
robj *pattern;
} redisSortOperation;
-/* ZSETs use a specialized version of Skiplists */
-typedef struct zskiplistNode {
- robj *obj;
- double score;
- struct zskiplistNode *backward;
- struct zskiplistLevel {
- struct zskiplistNode *forward;
- unsigned int span;
- } level[];
-} zskiplistNode;
-
-typedef struct zskiplist {
- struct zskiplistNode *header, *tail;
- unsigned long length;
- int level;
-} zskiplist;
-
-typedef struct zset {
- dict *dict;
- zskiplist *zsl;
-} zset;
-
/* DIsk store threaded I/O request message */
#define REDIS_IOJOB_LOAD 0
#define REDIS_IOJOB_SAVE 1
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;
* Functions prototypes
*----------------------------------------------------------------------------*/
+/* Utils */
+long long ustime(void);
+
/* networking.c -- Networking and Client related operations */
redisClient *createClient(int fd);
void closeTimedoutClients(void);
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, ...)
robj *createObject(int type, void *ptr);
robj *createStringObject(char *ptr, size_t len);
robj *dupStringObject(robj *o);
+int isObjectRepresentableAsLongLong(robj *o, long long *llongval);
robj *tryObjectEncoding(robj *o);
robj *getDecodedObject(robj *o);
size_t stringObjectLen(robj *o);
robj *createIntsetObject(void);
robj *createHashObject(void);
robj *createZsetObject(void);
+robj *createZsetZiplistObject(void);
int getLongFromObjectOrReply(redisClient *c, robj *o, long *target, const char *msg);
int checkType(redisClient *c, robj *o, int type);
int getLongLongFromObjectOrReply(redisClient *c, robj *o, long long *target, const char *msg);
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);
void replicationFeedMonitors(list *monitors, int dictid, robj **argv, int argc);
-int syncWithMaster(void);
void updateSlavesWaitingBgsave(int bgsaveerr);
void replicationCron(void);
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);
void backgroundRewriteDoneHandler(int exitcode, int bysignal);
/* Sorted sets data type */
+
+/* Struct to hold a inclusive/exclusive range spec. */
+typedef struct {
+ double min, max;
+ int minex, maxex; /* are min or max exclusive? */
+} zrangespec;
+
zskiplist *zslCreate(void);
void zslFree(zskiplist *zsl);
zskiplistNode *zslInsert(zskiplist *zsl, double score, robj *obj);
+unsigned char *zzlInsert(unsigned char *zl, robj *ele, double score);
+int zslDelete(zskiplist *zsl, double score, robj *obj);
+zskiplistNode *zslFirstInRange(zskiplist *zsl, zrangespec range);
+double zzlGetScore(unsigned char *sptr);
+void zzlNext(unsigned char *zl, unsigned char **eptr, unsigned char **sptr);
+void zzlPrev(unsigned char *zl, unsigned char **eptr, unsigned char **sptr);
+unsigned int zsetLength(robj *zobj);
+void zsetConvert(robj *zobj, int encoding);
/* 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);
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 */
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);
void freePubsubPattern(void *p);
int listMatchPubsubPattern(void *a, void *b);
-/* Utility functions */
-int stringmatchlen(const char *pattern, int patternLen,
- const char *string, int stringLen, int nocase);
-int stringmatch(const char *pattern, const char *string, int nocase);
-long long memtoll(const char *p, int *err);
-int ll2string(char *s, size_t len, long long value);
-int isStringRepresentableAsLong(sds s, long *longval);
-int isStringRepresentableAsLongLong(sds s, long long *longval);
-int isObjectRepresentableAsLongLong(robj *o, long long *llongval);
-
/* Configuration */
void loadServerConfig(char *filename);
void appendServerSaveParams(time_t seconds, int changes);
int selectDb(redisClient *c, int id);
void signalModifiedKey(redisDb *db, robj *key);
void signalFlushedDb(int dbid);
+unsigned int GetKeysInSlot(unsigned int hashslot, robj **keys, unsigned int count);
+
+/* 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, int *ask);
/* Git SHA1 */
char *redisGitSHA1(void);
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);
+void objectCommand(redisClient *c);
+void clientCommand(redisClient *c);
#if defined(__GNUC__)
void *calloc(size_t count, size_t size) __attribute__ ((deprecated));
projects / redis.git / blobdiff