/*
* Copyright (c) 2009-2010, Salvatore Sanfilippo <antirez at gmail dot com>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* * Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* * Neither the name of Redis nor the names of its contributors may be used
* to endorse or promote products derived from this software without
* specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include "redis.h"
#ifdef HAVE_BACKTRACE
#include <execinfo.h>
#include <ucontext.h>
#endif /* HAVE_BACKTRACE */
#include <time.h>
#include <signal.h>
#include <sys/wait.h>
#include <errno.h>
#include <assert.h>
#include <ctype.h>
#include <stdarg.h>
#include <arpa/inet.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/time.h>
#include <sys/resource.h>
#include <sys/uio.h>
#include <limits.h>
#include <float.h>
#include <math.h>
#include <pthread.h>
/* Our shared "common" objects */
struct sharedObjectsStruct shared;
/* Global vars that are actally used as constants. The following double
* values are used for double on-disk serialization, and are initialized
* at runtime to avoid strange compiler optimizations. */
double R_Zero, R_PosInf, R_NegInf, R_Nan;
/*================================= Globals ================================= */
/* Global vars */
struct redisServer server; /* server global state */
struct redisCommand *commandTable;
struct redisCommand readonlyCommandTable[] = {
{"get",getCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"set",setCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,0,0,0},
{"setnx",setnxCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,0,0,0},
{"setex",setexCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,0,0,0},
{"append",appendCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"substr",substrCommand,4,REDIS_CMD_INLINE,NULL,1,1,1},
{"strlen",strlenCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"del",delCommand,-2,REDIS_CMD_INLINE,NULL,0,0,0},
{"exists",existsCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"incr",incrCommand,2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"decr",decrCommand,2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"mget",mgetCommand,-2,REDIS_CMD_INLINE,NULL,1,-1,1},
{"rpush",rpushCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"lpush",lpushCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"rpushx",rpushxCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"lpushx",lpushxCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"linsert",linsertCommand,5,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"rpop",rpopCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"lpop",lpopCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"brpop",brpopCommand,-3,REDIS_CMD_INLINE,NULL,1,1,1},
{"blpop",blpopCommand,-3,REDIS_CMD_INLINE,NULL,1,1,1},
{"llen",llenCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"lindex",lindexCommand,3,REDIS_CMD_INLINE,NULL,1,1,1},
{"lset",lsetCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"lrange",lrangeCommand,4,REDIS_CMD_INLINE,NULL,1,1,1},
{"ltrim",ltrimCommand,4,REDIS_CMD_INLINE,NULL,1,1,1},
{"lrem",lremCommand,4,REDIS_CMD_BULK,NULL,1,1,1},
{"rpoplpush",rpoplpushcommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,2,1},
{"sadd",saddCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"srem",sremCommand,3,REDIS_CMD_BULK,NULL,1,1,1},
{"smove",smoveCommand,4,REDIS_CMD_BULK,NULL,1,2,1},
{"sismember",sismemberCommand,3,REDIS_CMD_BULK,NULL,1,1,1},
{"scard",scardCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"spop",spopCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"srandmember",srandmemberCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"sinter",sinterCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,-1,1},
{"sinterstore",sinterstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,2,-1,1},
{"sunion",sunionCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,-1,1},
{"sunionstore",sunionstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,2,-1,1},
{"sdiff",sdiffCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,-1,1},
{"sdiffstore",sdiffstoreCommand,-3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,2,-1,1},
{"smembers",sinterCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"zadd",zaddCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"zincrby",zincrbyCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"zrem",zremCommand,3,REDIS_CMD_BULK,NULL,1,1,1},
{"zremrangebyscore",zremrangebyscoreCommand,4,REDIS_CMD_INLINE,NULL,1,1,1},
{"zremrangebyrank",zremrangebyrankCommand,4,REDIS_CMD_INLINE,NULL,1,1,1},
{"zunionstore",zunionstoreCommand,-4,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,zunionInterBlockClientOnSwappedKeys,0,0,0},
{"zinterstore",zinterstoreCommand,-4,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,zunionInterBlockClientOnSwappedKeys,0,0,0},
{"zrange",zrangeCommand,-4,REDIS_CMD_INLINE,NULL,1,1,1},
{"zrangebyscore",zrangebyscoreCommand,-4,REDIS_CMD_INLINE,NULL,1,1,1},
{"zcount",zcountCommand,4,REDIS_CMD_INLINE,NULL,1,1,1},
{"zrevrange",zrevrangeCommand,-4,REDIS_CMD_INLINE,NULL,1,1,1},
{"zcard",zcardCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"zscore",zscoreCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"zrank",zrankCommand,3,REDIS_CMD_BULK,NULL,1,1,1},
{"zrevrank",zrevrankCommand,3,REDIS_CMD_BULK,NULL,1,1,1},
{"hset",hsetCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"hsetnx",hsetnxCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"hget",hgetCommand,3,REDIS_CMD_BULK,NULL,1,1,1},
{"hmset",hmsetCommand,-4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"hmget",hmgetCommand,-3,REDIS_CMD_BULK,NULL,1,1,1},
{"hincrby",hincrbyCommand,4,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"hdel",hdelCommand,3,REDIS_CMD_BULK,NULL,1,1,1},
{"hlen",hlenCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"hkeys",hkeysCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"hvals",hvalsCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"hgetall",hgetallCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"hexists",hexistsCommand,3,REDIS_CMD_BULK,NULL,1,1,1},
{"incrby",incrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"decrby",decrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"getset",getsetCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"mset",msetCommand,-3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,-1,2},
{"msetnx",msetnxCommand,-3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,NULL,1,-1,2},
{"randomkey",randomkeyCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"select",selectCommand,2,REDIS_CMD_INLINE,NULL,0,0,0},
{"move",moveCommand,3,REDIS_CMD_INLINE,NULL,1,1,1},
{"rename",renameCommand,3,REDIS_CMD_INLINE,NULL,1,1,1},
{"renamenx",renamenxCommand,3,REDIS_CMD_INLINE,NULL,1,1,1},
{"expire",expireCommand,3,REDIS_CMD_INLINE,NULL,0,0,0},
{"expireat",expireatCommand,3,REDIS_CMD_INLINE,NULL,0,0,0},
{"keys",keysCommand,2,REDIS_CMD_INLINE,NULL,0,0,0},
{"dbsize",dbsizeCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"auth",authCommand,2,REDIS_CMD_INLINE,NULL,0,0,0},
{"ping",pingCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"echo",echoCommand,2,REDIS_CMD_BULK,NULL,0,0,0},
{"save",saveCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"bgsave",bgsaveCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"bgrewriteaof",bgrewriteaofCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"shutdown",shutdownCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"lastsave",lastsaveCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"type",typeCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"multi",multiCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"exec",execCommand,1,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,execBlockClientOnSwappedKeys,0,0,0},
{"discard",discardCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"sync",syncCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"flushdb",flushdbCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"flushall",flushallCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"sort",sortCommand,-2,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,NULL,1,1,1},
{"info",infoCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"monitor",monitorCommand,1,REDIS_CMD_INLINE,NULL,0,0,0},
{"ttl",ttlCommand,2,REDIS_CMD_INLINE,NULL,1,1,1},
{"slaveof",slaveofCommand,3,REDIS_CMD_INLINE,NULL,0,0,0},
{"debug",debugCommand,-2,REDIS_CMD_INLINE,NULL,0,0,0},
{"config",configCommand,-2,REDIS_CMD_BULK,NULL,0,0,0},
{"subscribe",subscribeCommand,-2,REDIS_CMD_INLINE,NULL,0,0,0},
{"unsubscribe",unsubscribeCommand,-1,REDIS_CMD_INLINE,NULL,0,0,0},
{"psubscribe",psubscribeCommand,-2,REDIS_CMD_INLINE,NULL,0,0,0},
{"punsubscribe",punsubscribeCommand,-1,REDIS_CMD_INLINE,NULL,0,0,0},
{"publish",publishCommand,3,REDIS_CMD_BULK|REDIS_CMD_FORCE_REPLICATION,NULL,0,0,0},
{"watch",watchCommand,-2,REDIS_CMD_INLINE,NULL,0,0,0},
{"unwatch",unwatchCommand,1,REDIS_CMD_INLINE,NULL,0,0,0}
};
/*============================ Utility functions ============================ */
void redisLog(int level, const char *fmt, ...) {
va_list ap;
FILE *fp;
char *c = ".-*#";
char buf[64];
time_t now;
if (level < server.verbosity) return;
fp = (server.logfile == NULL) ? stdout : fopen(server.logfile,"a");
if (!fp) return;
va_start(ap, fmt);
now = time(NULL);
strftime(buf,64,"%d %b %H:%M:%S",localtime(&now));
fprintf(fp,"[%d] %s %c ",(int)getpid(),buf,c[level]);
vfprintf(fp, fmt, ap);
fprintf(fp,"\n");
fflush(fp);
va_end(ap);
if (server.logfile) fclose(fp);
}
/* Redis generally does not try to recover from out of memory conditions
* when allocating objects or strings, it is not clear if it will be possible
* to report this condition to the client since the networking layer itself
* is based on heap allocation for send buffers, so we simply abort.
* At least the code will be simpler to read... */
void oom(const char *msg) {
redisLog(REDIS_WARNING, "%s: Out of memory\n",msg);
sleep(1);
abort();
}
/*====================== Hash table type implementation ==================== */
/* This is an hash table type that uses the SDS dynamic strings libary as
* keys and radis objects as values (objects can hold SDS strings,
* lists, sets). */
void dictVanillaFree(void *privdata, void *val)
{
DICT_NOTUSED(privdata);
zfree(val);
}
void dictListDestructor(void *privdata, void *val)
{
DICT_NOTUSED(privdata);
listRelease((list*)val);
}
int dictSdsKeyCompare(void *privdata, const void *key1,
const void *key2)
{
int l1,l2;
DICT_NOTUSED(privdata);
l1 = sdslen((sds)key1);
l2 = sdslen((sds)key2);
if (l1 != l2) return 0;
return memcmp(key1, key2, l1) == 0;
}
void dictRedisObjectDestructor(void *privdata, void *val)
{
DICT_NOTUSED(privdata);
if (val == NULL) return; /* Values of swapped out keys as set to NULL */
decrRefCount(val);
}
void dictSdsDestructor(void *privdata, void *val)
{
DICT_NOTUSED(privdata);
sdsfree(val);
}
int dictObjKeyCompare(void *privdata, const void *key1,
const void *key2)
{
const robj *o1 = key1, *o2 = key2;
return dictSdsKeyCompare(privdata,o1->ptr,o2->ptr);
}
unsigned int dictObjHash(const void *key) {
const robj *o = key;
return dictGenHashFunction(o->ptr, sdslen((sds)o->ptr));
}
unsigned int dictSdsHash(const void *key) {
return dictGenHashFunction((unsigned char*)key, sdslen((char*)key));
}
int dictEncObjKeyCompare(void *privdata, const void *key1,
const void *key2)
{
robj *o1 = (robj*) key1, *o2 = (robj*) key2;
int cmp;
if (o1->encoding == REDIS_ENCODING_INT &&
o2->encoding == REDIS_ENCODING_INT)
return o1->ptr == o2->ptr;
o1 = getDecodedObject(o1);
o2 = getDecodedObject(o2);
cmp = dictSdsKeyCompare(privdata,o1->ptr,o2->ptr);
decrRefCount(o1);
decrRefCount(o2);
return cmp;
}
unsigned int dictEncObjHash(const void *key) {
robj *o = (robj*) key;
if (o->encoding == REDIS_ENCODING_RAW) {
return dictGenHashFunction(o->ptr, sdslen((sds)o->ptr));
} else {
if (o->encoding == REDIS_ENCODING_INT) {
char buf[32];
int len;
len = ll2string(buf,32,(long)o->ptr);
return dictGenHashFunction((unsigned char*)buf, len);
} else {
unsigned int hash;
o = getDecodedObject(o);
hash = dictGenHashFunction(o->ptr, sdslen((sds)o->ptr));
decrRefCount(o);
return hash;
}
}
}
/* Sets type */
dictType setDictType = {
dictEncObjHash, /* hash function */
NULL, /* key dup */
NULL, /* val dup */
dictEncObjKeyCompare, /* key compare */
dictRedisObjectDestructor, /* key destructor */
NULL /* val destructor */
};
/* Sorted sets hash (note: a skiplist is used in addition to the hash table) */
dictType zsetDictType = {
dictEncObjHash, /* hash function */
NULL, /* key dup */
NULL, /* val dup */
dictEncObjKeyCompare, /* key compare */
dictRedisObjectDestructor, /* key destructor */
dictVanillaFree /* val destructor of malloc(sizeof(double)) */
};
/* Db->dict, keys are sds strings, vals are Redis objects. */
dictType dbDictType = {
dictSdsHash, /* hash function */
NULL, /* key dup */
NULL, /* val dup */
dictSdsKeyCompare, /* key compare */
dictSdsDestructor, /* key destructor */
dictRedisObjectDestructor /* val destructor */
};
/* Db->expires */
dictType keyptrDictType = {
dictSdsHash, /* hash function */
NULL, /* key dup */
NULL, /* val dup */
dictSdsKeyCompare, /* key compare */
NULL, /* key destructor */
NULL /* val destructor */
};
/* Hash type hash table (note that small hashes are represented with zimpaps) */
dictType hashDictType = {
dictEncObjHash, /* hash function */
NULL, /* key dup */
NULL, /* val dup */
dictEncObjKeyCompare, /* key compare */
dictRedisObjectDestructor, /* key destructor */
dictRedisObjectDestructor /* val destructor */
};
/* Keylist hash table type has unencoded redis objects as keys and
* lists as values. It's used for blocking operations (BLPOP) and to
* map swapped keys to a list of clients waiting for this keys to be loaded. */
dictType keylistDictType = {
dictObjHash, /* hash function */
NULL, /* key dup */
NULL, /* val dup */
dictObjKeyCompare, /* key compare */
dictRedisObjectDestructor, /* key destructor */
dictListDestructor /* val destructor */
};
int htNeedsResize(dict *dict) {
long long size, used;
size = dictSlots(dict);
used = dictSize(dict);
return (size && used && size > DICT_HT_INITIAL_SIZE &&
(used*100/size < REDIS_HT_MINFILL));
}
/* If the percentage of used slots in the HT reaches REDIS_HT_MINFILL
* we resize the hash table to save memory */
void tryResizeHashTables(void) {
int j;
for (j = 0; j < server.dbnum; j++) {
if (htNeedsResize(server.db[j].dict))
dictResize(server.db[j].dict);
if (htNeedsResize(server.db[j].expires))
dictResize(server.db[j].expires);
}
}
/* Our hash table implementation performs rehashing incrementally while
* we write/read from the hash table. Still if the server is idle, the hash
* table will use two tables for a long time. So we try to use 1 millisecond
* of CPU time at every serverCron() loop in order to rehash some key. */
void incrementallyRehash(void) {
int j;
for (j = 0; j < server.dbnum; j++) {
if (dictIsRehashing(server.db[j].dict)) {
dictRehashMilliseconds(server.db[j].dict,1);
break; /* already used our millisecond for this loop... */
}
}
}
/* This function is called once a background process of some kind terminates,
* as we want to avoid resizing the hash tables when there is a child in order
* to play well with copy-on-write (otherwise when a resize happens lots of
* memory pages are copied). The goal of this function is to update the ability
* for dict.c to resize the hash tables accordingly to the fact we have o not
* running childs. */
void updateDictResizePolicy(void) {
if (server.bgsavechildpid == -1 && server.bgrewritechildpid == -1)
dictEnableResize();
else
dictDisableResize();
}
/* ======================= Cron: called every 100 ms ======================== */
int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) {
int j, loops = server.cronloops++;
REDIS_NOTUSED(eventLoop);
REDIS_NOTUSED(id);
REDIS_NOTUSED(clientData);
/* We take a cached value of the unix time in the global state because
* with virtual memory and aging there is to store the current time
* in objects at every object access, and accuracy is not needed.
* To access a global var is faster than calling time(NULL) */
server.unixtime = time(NULL);
/* We have just 21 bits per object for LRU information.
* So we use an (eventually wrapping) LRU clock with minutes resolution.
*
* When we need to select what object to swap, we compute the minimum
* time distance between the current lruclock and the object last access
* lruclock info. Even if clocks will wrap on overflow, there is
* the interesting property that we are sure that at least
* ABS(A-B) minutes passed between current time and timestamp B.
*
* This is not precise but we don't need at all precision, but just
* something statistically reasonable.
*/
server.lruclock = (time(NULL)/60)&((1<<21)-1);
/* We received a SIGTERM, shutting down here in a safe way, as it is
* not ok doing so inside the signal handler. */
if (server.shutdown_asap) {
if (prepareForShutdown() == REDIS_OK) exit(0);
redisLog(REDIS_WARNING,"SIGTERM received but errors trying to shut down the server, check the logs for more information");
}
/* Show some info about non-empty databases */
for (j = 0; j < server.dbnum; j++) {
long long size, used, vkeys;
size = dictSlots(server.db[j].dict);
used = dictSize(server.db[j].dict);
vkeys = dictSize(server.db[j].expires);
if (!(loops % 50) && (used || vkeys)) {
redisLog(REDIS_VERBOSE,"DB %d: %lld keys (%lld volatile) in %lld slots HT.",j,used,vkeys,size);
/* dictPrintStats(server.dict); */
}
}
/* We don't want to resize the hash tables while a bacground saving
* is in progress: the saving child is created using fork() that is
* implemented with a copy-on-write semantic in most modern systems, so
* if we resize the HT while there is the saving child at work actually
* a lot of memory movements in the parent will cause a lot of pages
* copied. */
if (server.bgsavechildpid == -1 && server.bgrewritechildpid == -1) {
if (!(loops % 10)) tryResizeHashTables();
if (server.activerehashing) incrementallyRehash();
}
/* Show information about connected clients */
if (!(loops % 50)) {
redisLog(REDIS_VERBOSE,"%d clients connected (%d slaves), %zu bytes in use",
listLength(server.clients)-listLength(server.slaves),
listLength(server.slaves),
zmalloc_used_memory());
}
/* Close connections of timedout clients */
if ((server.maxidletime && !(loops % 100)) || server.blpop_blocked_clients)
closeTimedoutClients();
/* Check if a background saving or AOF rewrite in progress terminated */
if (server.bgsavechildpid != -1 || server.bgrewritechildpid != -1) {
int statloc;
pid_t pid;
if ((pid = wait3(&statloc,WNOHANG,NULL)) != 0) {
if (pid == server.bgsavechildpid) {
backgroundSaveDoneHandler(statloc);
} else {
backgroundRewriteDoneHandler(statloc);
}
updateDictResizePolicy();
}
} else {
/* If there is not a background saving in progress check if
* we have to save now */
time_t now = time(NULL);
for (j = 0; j < server.saveparamslen; j++) {
struct saveparam *sp = server.saveparams+j;
if (server.dirty >= sp->changes &&
now-server.lastsave > sp->seconds) {
redisLog(REDIS_NOTICE,"%d changes in %d seconds. Saving...",
sp->changes, sp->seconds);
rdbSaveBackground(server.dbfilename);
break;
}
}
}
/* Try to expire a few timed out keys. The algorithm used is adaptive and
* will use few CPU cycles if there are few expiring keys, otherwise
* it will get more aggressive to avoid that too much memory is used by
* keys that can be removed from the keyspace. */
for (j = 0; j < server.dbnum; j++) {
int expired;
redisDb *db = server.db+j;
/* Continue to expire if at the end of the cycle more than 25%
* of the keys were expired. */
do {
long num = dictSize(db->expires);
time_t now = time(NULL);
expired = 0;
if (num > REDIS_EXPIRELOOKUPS_PER_CRON)
num = REDIS_EXPIRELOOKUPS_PER_CRON;
while (num--) {
dictEntry *de;
time_t t;
if ((de = dictGetRandomKey(db->expires)) == NULL) break;
t = (time_t) dictGetEntryVal(de);
if (now > t) {
sds key = dictGetEntryKey(de);
robj *keyobj = createStringObject(key,sdslen(key));
dbDelete(db,keyobj);
decrRefCount(keyobj);
expired++;
server.stat_expiredkeys++;
}
}
} while (expired > REDIS_EXPIRELOOKUPS_PER_CRON/4);
}
/* Swap a few keys on disk if we are over the memory limit and VM
* is enbled. Try to free objects from the free list first. */
if (vmCanSwapOut()) {
while (server.vm_enabled && zmalloc_used_memory() >
server.vm_max_memory)
{
int retval;
if (tryFreeOneObjectFromFreelist() == REDIS_OK) continue;
retval = (server.vm_max_threads == 0) ?
vmSwapOneObjectBlocking() :
vmSwapOneObjectThreaded();
if (retval == REDIS_ERR && !(loops % 300) &&
zmalloc_used_memory() >
(server.vm_max_memory+server.vm_max_memory/10))
{
redisLog(REDIS_WARNING,"WARNING: vm-max-memory limit exceeded by more than 10%% but unable to swap more objects out!");
}
/* Note that when using threade I/O we free just one object,
* because anyway when the I/O thread in charge to swap this
* object out will finish, the handler of completed jobs
* will try to swap more objects if we are still out of memory. */
if (retval == REDIS_ERR || server.vm_max_threads > 0) break;
}
}
/* Check if we should connect to a MASTER */
if (server.replstate == REDIS_REPL_CONNECT && !(loops % 10)) {
redisLog(REDIS_NOTICE,"Connecting to MASTER...");
if (syncWithMaster() == REDIS_OK) {
redisLog(REDIS_NOTICE,"MASTER <-> SLAVE sync succeeded");
if (server.appendonly) rewriteAppendOnlyFileBackground();
}
}
return 100;
}
/* This function gets called every time Redis is entering the
* main loop of the event driven library, that is, before to sleep
* for ready file descriptors. */
void beforeSleep(struct aeEventLoop *eventLoop) {
REDIS_NOTUSED(eventLoop);
/* Awake clients that got all the swapped keys they requested */
if (server.vm_enabled && listLength(server.io_ready_clients)) {
listIter li;
listNode *ln;
listRewind(server.io_ready_clients,&li);
while((ln = listNext(&li))) {
redisClient *c = ln->value;
struct redisCommand *cmd;
/* Resume the client. */
listDelNode(server.io_ready_clients,ln);
c->flags &= (~REDIS_IO_WAIT);
server.vm_blocked_clients--;
aeCreateFileEvent(server.el, c->fd, AE_READABLE,
readQueryFromClient, c);
cmd = lookupCommand(c->argv[0]->ptr);
redisAssert(cmd != NULL);
call(c,cmd);
resetClient(c);
/* There may be more data to process in the input buffer. */
if (c->querybuf && sdslen(c->querybuf) > 0)
processInputBuffer(c);
}
}
/* Write the AOF buffer on disk */
flushAppendOnlyFile();
}
/* =========================== Server initialization ======================== */
void createSharedObjects(void) {
int j;
shared.crlf = createObject(REDIS_STRING,sdsnew("\r\n"));
shared.ok = createObject(REDIS_STRING,sdsnew("+OK\r\n"));
shared.err = createObject(REDIS_STRING,sdsnew("-ERR\r\n"));
shared.emptybulk = createObject(REDIS_STRING,sdsnew("$0\r\n\r\n"));
shared.czero = createObject(REDIS_STRING,sdsnew(":0\r\n"));
shared.cone = createObject(REDIS_STRING,sdsnew(":1\r\n"));
shared.cnegone = createObject(REDIS_STRING,sdsnew(":-1\r\n"));
shared.nullbulk = createObject(REDIS_STRING,sdsnew("$-1\r\n"));
shared.nullmultibulk = createObject(REDIS_STRING,sdsnew("*-1\r\n"));
shared.emptymultibulk = createObject(REDIS_STRING,sdsnew("*0\r\n"));
shared.pong = createObject(REDIS_STRING,sdsnew("+PONG\r\n"));
shared.queued = createObject(REDIS_STRING,sdsnew("+QUEUED\r\n"));
shared.wrongtypeerr = createObject(REDIS_STRING,sdsnew(
"-ERR Operation against a key holding the wrong kind of value\r\n"));
shared.nokeyerr = createObject(REDIS_STRING,sdsnew(
"-ERR no such key\r\n"));
shared.syntaxerr = createObject(REDIS_STRING,sdsnew(
"-ERR syntax error\r\n"));
shared.sameobjecterr = createObject(REDIS_STRING,sdsnew(
"-ERR source and destination objects are the same\r\n"));
shared.outofrangeerr = createObject(REDIS_STRING,sdsnew(
"-ERR index out of range\r\n"));
shared.space = createObject(REDIS_STRING,sdsnew(" "));
shared.colon = createObject(REDIS_STRING,sdsnew(":"));
shared.plus = createObject(REDIS_STRING,sdsnew("+"));
shared.select0 = createStringObject("select 0\r\n",10);
shared.select1 = createStringObject("select 1\r\n",10);
shared.select2 = createStringObject("select 2\r\n",10);
shared.select3 = createStringObject("select 3\r\n",10);
shared.select4 = createStringObject("select 4\r\n",10);
shared.select5 = createStringObject("select 5\r\n",10);
shared.select6 = createStringObject("select 6\r\n",10);
shared.select7 = createStringObject("select 7\r\n",10);
shared.select8 = createStringObject("select 8\r\n",10);
shared.select9 = createStringObject("select 9\r\n",10);
shared.messagebulk = createStringObject("$7\r\nmessage\r\n",13);
shared.pmessagebulk = createStringObject("$8\r\npmessage\r\n",14);
shared.subscribebulk = createStringObject("$9\r\nsubscribe\r\n",15);
shared.unsubscribebulk = createStringObject("$11\r\nunsubscribe\r\n",18);
shared.psubscribebulk = createStringObject("$10\r\npsubscribe\r\n",17);
shared.punsubscribebulk = createStringObject("$12\r\npunsubscribe\r\n",19);
shared.mbulk3 = createStringObject("*3\r\n",4);
shared.mbulk4 = createStringObject("*4\r\n",4);
for (j = 0; j < REDIS_SHARED_INTEGERS; j++) {
shared.integers[j] = createObject(REDIS_STRING,(void*)(long)j);
shared.integers[j]->encoding = REDIS_ENCODING_INT;
}
}
void initServerConfig() {
server.dbnum = REDIS_DEFAULT_DBNUM;
server.port = REDIS_SERVERPORT;
server.verbosity = REDIS_VERBOSE;
server.maxidletime = REDIS_MAXIDLETIME;
server.saveparams = NULL;
server.logfile = NULL; /* NULL = log on standard output */
server.bindaddr = NULL;
server.glueoutputbuf = 1;
server.daemonize = 0;
server.appendonly = 0;
server.appendfsync = APPENDFSYNC_EVERYSEC;
server.no_appendfsync_on_rewrite = 0;
server.lastfsync = time(NULL);
server.appendfd = -1;
server.appendseldb = -1; /* Make sure the first time will not match */
server.pidfile = zstrdup("/var/run/redis.pid");
server.dbfilename = zstrdup("dump.rdb");
server.appendfilename = zstrdup("appendonly.aof");
server.requirepass = NULL;
server.rdbcompression = 1;
server.activerehashing = 1;
server.maxclients = 0;
server.blpop_blocked_clients = 0;
server.maxmemory = 0;
server.vm_enabled = 0;
server.vm_swap_file = zstrdup("/tmp/redis-%p.vm");
server.vm_page_size = 256; /* 256 bytes per page */
server.vm_pages = 1024*1024*100; /* 104 millions of pages */
server.vm_max_memory = 1024LL*1024*1024*1; /* 1 GB of RAM */
server.vm_max_threads = 4;
server.vm_blocked_clients = 0;
server.hash_max_zipmap_entries = REDIS_HASH_MAX_ZIPMAP_ENTRIES;
server.hash_max_zipmap_value = REDIS_HASH_MAX_ZIPMAP_VALUE;
server.list_max_ziplist_entries = REDIS_LIST_MAX_ZIPLIST_ENTRIES;
server.list_max_ziplist_value = REDIS_LIST_MAX_ZIPLIST_VALUE;
server.shutdown_asap = 0;
resetServerSaveParams();
appendServerSaveParams(60*60,1); /* save after 1 hour and 1 change */
appendServerSaveParams(300,100); /* save after 5 minutes and 100 changes */
appendServerSaveParams(60,10000); /* save after 1 minute and 10000 changes */
/* Replication related */
server.isslave = 0;
server.masterauth = NULL;
server.masterhost = NULL;
server.masterport = 6379;
server.master = NULL;
server.replstate = REDIS_REPL_NONE;
/* Double constants initialization */
R_Zero = 0.0;
R_PosInf = 1.0/R_Zero;
R_NegInf = -1.0/R_Zero;
R_Nan = R_Zero/R_Zero;
}
void initServer() {
int j;
signal(SIGHUP, SIG_IGN);
signal(SIGPIPE, SIG_IGN);
setupSigSegvAction();
server.mainthread = pthread_self();
server.devnull = fopen("/dev/null","w");
if (server.devnull == NULL) {
redisLog(REDIS_WARNING, "Can't open /dev/null: %s", server.neterr);
exit(1);
}
server.clients = listCreate();
server.slaves = listCreate();
server.monitors = listCreate();
server.objfreelist = listCreate();
createSharedObjects();
server.el = aeCreateEventLoop();
server.db = zmalloc(sizeof(redisDb)*server.dbnum);
server.fd = anetTcpServer(server.neterr, server.port, server.bindaddr);
if (server.fd == -1) {
redisLog(REDIS_WARNING, "Opening TCP port: %s", server.neterr);
exit(1);
}
for (j = 0; j < server.dbnum; j++) {
server.db[j].dict = dictCreate(&dbDictType,NULL);
server.db[j].expires = dictCreate(&keyptrDictType,NULL);
server.db[j].blocking_keys = dictCreate(&keylistDictType,NULL);
server.db[j].watched_keys = dictCreate(&keylistDictType,NULL);
if (server.vm_enabled)
server.db[j].io_keys = dictCreate(&keylistDictType,NULL);
server.db[j].id = j;
}
server.pubsub_channels = dictCreate(&keylistDictType,NULL);
server.pubsub_patterns = listCreate();
listSetFreeMethod(server.pubsub_patterns,freePubsubPattern);
listSetMatchMethod(server.pubsub_patterns,listMatchPubsubPattern);
server.cronloops = 0;
server.bgsavechildpid = -1;
server.bgrewritechildpid = -1;
server.bgrewritebuf = sdsempty();
server.aofbuf = sdsempty();
server.lastsave = time(NULL);
server.dirty = 0;
server.stat_numcommands = 0;
server.stat_numconnections = 0;
server.stat_expiredkeys = 0;
server.stat_starttime = time(NULL);
server.unixtime = time(NULL);
aeCreateTimeEvent(server.el, 1, serverCron, NULL, NULL);
if (aeCreateFileEvent(server.el, server.fd, AE_READABLE,
acceptHandler, NULL) == AE_ERR) oom("creating file event");
if (server.appendonly) {
server.appendfd = open(server.appendfilename,O_WRONLY|O_APPEND|O_CREAT,0644);
if (server.appendfd == -1) {
redisLog(REDIS_WARNING, "Can't open the append-only file: %s",
strerror(errno));
exit(1);
}
}
if (server.vm_enabled) vmInit();
}
int qsortRedisCommands(const void *r1, const void *r2) {
return strcasecmp(
((struct redisCommand*)r1)->name,
((struct redisCommand*)r2)->name);
}
void sortCommandTable() {
/* Copy and sort the read-only version of the command table */
commandTable = (struct redisCommand*)zmalloc(sizeof(readonlyCommandTable));
memcpy(commandTable,readonlyCommandTable,sizeof(readonlyCommandTable));
qsort(commandTable,
sizeof(readonlyCommandTable)/sizeof(struct redisCommand),
sizeof(struct redisCommand),qsortRedisCommands);
}
/* ====================== Commands lookup and execution ===================== */
struct redisCommand *lookupCommand(char *name) {
struct redisCommand tmp = {name,NULL,0,0,NULL,0,0,0};
return bsearch(
&tmp,
commandTable,
sizeof(readonlyCommandTable)/sizeof(struct redisCommand),
sizeof(struct redisCommand),
qsortRedisCommands);
}
/* Call() is the core of Redis execution of a command */
void call(redisClient *c, struct redisCommand *cmd) {
long long dirty;
dirty = server.dirty;
cmd->proc(c);
dirty = server.dirty-dirty;
if (server.appendonly && dirty)
feedAppendOnlyFile(cmd,c->db->id,c->argv,c->argc);
if ((dirty || cmd->flags & REDIS_CMD_FORCE_REPLICATION) &&
listLength(server.slaves))
replicationFeedSlaves(server.slaves,c->db->id,c->argv,c->argc);
if (listLength(server.monitors))
replicationFeedMonitors(server.monitors,c->db->id,c->argv,c->argc);
server.stat_numcommands++;
}
/* If this function gets called we already read a whole
* command, argments are in the client argv/argc fields.
* processCommand() execute the command or prepare the
* server for a bulk read from the client.
*
* If 1 is returned the client is still alive and valid and
* and other operations can be performed by the caller. Otherwise
* if 0 is returned the client was destroied (i.e. after QUIT). */
int processCommand(redisClient *c) {
struct redisCommand *cmd;
/* Free some memory if needed (maxmemory setting) */
if (server.maxmemory) freeMemoryIfNeeded();
/* Handle the multi bulk command type. This is an alternative protocol
* supported by Redis in order to receive commands that are composed of
* multiple binary-safe "bulk" arguments. The latency of processing is
* a bit higher but this allows things like multi-sets, so if this
* protocol is used only for MSET and similar commands this is a big win. */
if (c->multibulk == 0 && c->argc == 1 && ((char*)(c->argv[0]->ptr))[0] == '*') {
c->multibulk = atoi(((char*)c->argv[0]->ptr)+1);
if (c->multibulk <= 0) {
resetClient(c);
return 1;
} else {
decrRefCount(c->argv[c->argc-1]);
c->argc--;
return 1;
}
} else if (c->multibulk) {
if (c->bulklen == -1) {
if (((char*)c->argv[0]->ptr)[0] != '$') {
addReplySds(c,sdsnew("-ERR multi bulk protocol error\r\n"));
resetClient(c);
return 1;
} else {
int bulklen = atoi(((char*)c->argv[0]->ptr)+1);
decrRefCount(c->argv[0]);
if (bulklen < 0 || bulklen > 1024*1024*1024) {
c->argc--;
addReplySds(c,sdsnew("-ERR invalid bulk write count\r\n"));
resetClient(c);
return 1;
}
c->argc--;
c->bulklen = bulklen+2; /* add two bytes for CR+LF */
return 1;
}
} else {
c->mbargv = zrealloc(c->mbargv,(sizeof(robj*))*(c->mbargc+1));
c->mbargv[c->mbargc] = c->argv[0];
c->mbargc++;
c->argc--;
c->multibulk--;
if (c->multibulk == 0) {
robj **auxargv;
int auxargc;
/* Here we need to swap the multi-bulk argc/argv with the
* normal argc/argv of the client structure. */
auxargv = c->argv;
c->argv = c->mbargv;
c->mbargv = auxargv;
auxargc = c->argc;
c->argc = c->mbargc;
c->mbargc = auxargc;
/* We need to set bulklen to something different than -1
* in order for the code below to process the command without
* to try to read the last argument of a bulk command as
* a special argument. */
c->bulklen = 0;
/* continue below and process the command */
} else {
c->bulklen = -1;
return 1;
}
}
}
/* -- end of multi bulk commands processing -- */
/* The QUIT command is handled as a special case. Normal command
* procs are unable to close the client connection safely */
if (!strcasecmp(c->argv[0]->ptr,"quit")) {
freeClient(c);
return 0;
}
/* Now lookup the command and check ASAP about trivial error conditions
* such wrong arity, bad command name and so forth. */
cmd = lookupCommand(c->argv[0]->ptr);
if (!cmd) {
addReplySds(c,
sdscatprintf(sdsempty(), "-ERR unknown command '%s'\r\n",
(char*)c->argv[0]->ptr));
resetClient(c);
return 1;
} else if ((cmd->arity > 0 && cmd->arity != c->argc) ||
(c->argc < -cmd->arity)) {
addReplySds(c,
sdscatprintf(sdsempty(),
"-ERR wrong number of arguments for '%s' command\r\n",
cmd->name));
resetClient(c);
return 1;
} else if (cmd->flags & REDIS_CMD_BULK && c->bulklen == -1) {
/* This is a bulk command, we have to read the last argument yet. */
int bulklen = atoi(c->argv[c->argc-1]->ptr);
decrRefCount(c->argv[c->argc-1]);
if (bulklen < 0 || bulklen > 1024*1024*1024) {
c->argc--;
addReplySds(c,sdsnew("-ERR invalid bulk write count\r\n"));
resetClient(c);
return 1;
}
c->argc--;
c->bulklen = bulklen+2; /* add two bytes for CR+LF */
/* It is possible that the bulk read is already in the
* buffer. Check this condition and handle it accordingly.
* This is just a fast path, alternative to call processInputBuffer().
* It's a good idea since the code is small and this condition
* happens most of the times. */
if ((signed)sdslen(c->querybuf) >= c->bulklen) {
c->argv[c->argc] = createStringObject(c->querybuf,c->bulklen-2);
c->argc++;
c->querybuf = sdsrange(c->querybuf,c->bulklen,-1);
} else {
/* Otherwise return... there is to read the last argument
* from the socket. */
return 1;
}
}
/* Let's try to encode the bulk object to save space. */
if (cmd->flags & REDIS_CMD_BULK)
c->argv[c->argc-1] = tryObjectEncoding(c->argv[c->argc-1]);
/* Check if the user is authenticated */
if (server.requirepass && !c->authenticated && cmd->proc != authCommand) {
addReplySds(c,sdsnew("-ERR operation not permitted\r\n"));
resetClient(c);
return 1;
}
/* Handle the maxmemory directive */
if (server.maxmemory && (cmd->flags & REDIS_CMD_DENYOOM) &&
zmalloc_used_memory() > server.maxmemory)
{
addReplySds(c,sdsnew("-ERR command not allowed when used memory > 'maxmemory'\r\n"));
resetClient(c);
return 1;
}
/* Only allow SUBSCRIBE and UNSUBSCRIBE in the context of Pub/Sub */
if ((dictSize(c->pubsub_channels) > 0 || listLength(c->pubsub_patterns) > 0)
&&
cmd->proc != subscribeCommand && cmd->proc != unsubscribeCommand &&
cmd->proc != psubscribeCommand && cmd->proc != punsubscribeCommand) {
addReplySds(c,sdsnew("-ERR only (P)SUBSCRIBE / (P)UNSUBSCRIBE / QUIT allowed in this context\r\n"));
resetClient(c);
return 1;
}
/* Exec the command */
if (c->flags & REDIS_MULTI &&
cmd->proc != execCommand && cmd->proc != discardCommand &&
cmd->proc != multiCommand && cmd->proc != watchCommand)
{
queueMultiCommand(c,cmd);
addReply(c,shared.queued);
} else {
if (server.vm_enabled && server.vm_max_threads > 0 &&
blockClientOnSwappedKeys(c,cmd)) return 1;
call(c,cmd);
}
/* Prepare the client for the next command */
resetClient(c);
return 1;
}
/*================================== Shutdown =============================== */
int prepareForShutdown() {
redisLog(REDIS_WARNING,"User requested shutdown, saving DB...");
/* Kill the saving child if there is a background saving in progress.
We want to avoid race conditions, for instance our saving child may
overwrite the synchronous saving did by SHUTDOWN. */
if (server.bgsavechildpid != -1) {
redisLog(REDIS_WARNING,"There is a live saving child. Killing it!");
kill(server.bgsavechildpid,SIGKILL);
rdbRemoveTempFile(server.bgsavechildpid);
}
if (server.appendonly) {
/* Append only file: fsync() the AOF and exit */
aof_fsync(server.appendfd);
if (server.vm_enabled) unlink(server.vm_swap_file);
} else {
/* Snapshotting. Perform a SYNC SAVE and exit */
if (rdbSave(server.dbfilename) == REDIS_OK) {
if (server.daemonize)
unlink(server.pidfile);
redisLog(REDIS_WARNING,"%zu bytes used at exit",zmalloc_used_memory());
} else {
/* Ooops.. error saving! The best we can do is to continue
* operating. Note that if there was a background saving process,
* in the next cron() Redis will be notified that the background
* saving aborted, handling special stuff like slaves pending for
* synchronization... */
redisLog(REDIS_WARNING,"Error trying to save the DB, can't exit");
return REDIS_ERR;
}
}
redisLog(REDIS_WARNING,"Server exit now, bye bye...");
return REDIS_OK;
}
/*================================== Commands =============================== */
void authCommand(redisClient *c) {
if (!server.requirepass || !strcmp(c->argv[1]->ptr, server.requirepass)) {
c->authenticated = 1;
addReply(c,shared.ok);
} else {
c->authenticated = 0;
addReplySds(c,sdscatprintf(sdsempty(),"-ERR invalid password\r\n"));
}
}
void pingCommand(redisClient *c) {
addReply(c,shared.pong);
}
void echoCommand(redisClient *c) {
addReplyBulk(c,c->argv[1]);
}
/* Convert an amount of bytes into a human readable string in the form
* of 100B, 2G, 100M, 4K, and so forth. */
void bytesToHuman(char *s, unsigned long long n) {
double d;
if (n < 1024) {
/* Bytes */
sprintf(s,"%lluB",n);
return;
} else if (n < (1024*1024)) {
d = (double)n/(1024);
sprintf(s,"%.2fK",d);
} else if (n < (1024LL*1024*1024)) {
d = (double)n/(1024*1024);
sprintf(s,"%.2fM",d);
} else if (n < (1024LL*1024*1024*1024)) {
d = (double)n/(1024LL*1024*1024);
sprintf(s,"%.2fG",d);
}
}
/* Create the string returned by the INFO command. This is decoupled
* by the INFO command itself as we need to report the same information
* on memory corruption problems. */
sds genRedisInfoString(void) {
sds info;
time_t uptime = time(NULL)-server.stat_starttime;
int j;
char hmem[64];
bytesToHuman(hmem,zmalloc_used_memory());
info = sdscatprintf(sdsempty(),
"redis_version:%s\r\n"
"redis_git_sha1:%s\r\n"
"redis_git_dirty:%d\r\n"
"arch_bits:%s\r\n"
"multiplexing_api:%s\r\n"
"process_id:%ld\r\n"
"uptime_in_seconds:%ld\r\n"
"uptime_in_days:%ld\r\n"
"connected_clients:%d\r\n"
"connected_slaves:%d\r\n"
"blocked_clients:%d\r\n"
"used_memory:%zu\r\n"
"used_memory_human:%s\r\n"
"changes_since_last_save:%lld\r\n"
"bgsave_in_progress:%d\r\n"
"last_save_time:%ld\r\n"
"bgrewriteaof_in_progress:%d\r\n"
"total_connections_received:%lld\r\n"
"total_commands_processed:%lld\r\n"
"expired_keys:%lld\r\n"
"hash_max_zipmap_entries:%zu\r\n"
"hash_max_zipmap_value:%zu\r\n"
"pubsub_channels:%ld\r\n"
"pubsub_patterns:%u\r\n"
"vm_enabled:%d\r\n"
"role:%s\r\n"
,REDIS_VERSION,
redisGitSHA1(),
strtol(redisGitDirty(),NULL,10) > 0,
(sizeof(long) == 8) ? "64" : "32",
aeGetApiName(),
(long) getpid(),
uptime,
uptime/(3600*24),
listLength(server.clients)-listLength(server.slaves),
listLength(server.slaves),
server.blpop_blocked_clients,
zmalloc_used_memory(),
hmem,
server.dirty,
server.bgsavechildpid != -1,
server.lastsave,
server.bgrewritechildpid != -1,
server.stat_numconnections,
server.stat_numcommands,
server.stat_expiredkeys,
server.hash_max_zipmap_entries,
server.hash_max_zipmap_value,
dictSize(server.pubsub_channels),
listLength(server.pubsub_patterns),
server.vm_enabled != 0,
server.masterhost == NULL ? "master" : "slave"
);
if (server.masterhost) {
info = sdscatprintf(info,
"master_host:%s\r\n"
"master_port:%d\r\n"
"master_link_status:%s\r\n"
"master_last_io_seconds_ago:%d\r\n"
,server.masterhost,
server.masterport,
(server.replstate == REDIS_REPL_CONNECTED) ?
"up" : "down",
server.master ? ((int)(time(NULL)-server.master->lastinteraction)) : -1
);
}
if (server.vm_enabled) {
lockThreadedIO();
info = sdscatprintf(info,
"vm_conf_max_memory:%llu\r\n"
"vm_conf_page_size:%llu\r\n"
"vm_conf_pages:%llu\r\n"
"vm_stats_used_pages:%llu\r\n"
"vm_stats_swapped_objects:%llu\r\n"
"vm_stats_swappin_count:%llu\r\n"
"vm_stats_swappout_count:%llu\r\n"
"vm_stats_io_newjobs_len:%lu\r\n"
"vm_stats_io_processing_len:%lu\r\n"
"vm_stats_io_processed_len:%lu\r\n"
"vm_stats_io_active_threads:%lu\r\n"
"vm_stats_blocked_clients:%lu\r\n"
,(unsigned long long) server.vm_max_memory,
(unsigned long long) server.vm_page_size,
(unsigned long long) server.vm_pages,
(unsigned long long) server.vm_stats_used_pages,
(unsigned long long) server.vm_stats_swapped_objects,
(unsigned long long) server.vm_stats_swapins,
(unsigned long long) server.vm_stats_swapouts,
(unsigned long) listLength(server.io_newjobs),
(unsigned long) listLength(server.io_processing),
(unsigned long) listLength(server.io_processed),
(unsigned long) server.io_active_threads,
(unsigned long) server.vm_blocked_clients
);
unlockThreadedIO();
}
for (j = 0; j < server.dbnum; j++) {
long long keys, vkeys;
keys = dictSize(server.db[j].dict);
vkeys = dictSize(server.db[j].expires);
if (keys || vkeys) {
info = sdscatprintf(info, "db%d:keys=%lld,expires=%lld\r\n",
j, keys, vkeys);
}
}
return info;
}
void infoCommand(redisClient *c) {
sds info = genRedisInfoString();
addReplySds(c,sdscatprintf(sdsempty(),"$%lu\r\n",
(unsigned long)sdslen(info)));
addReplySds(c,info);
addReply(c,shared.crlf);
}
void monitorCommand(redisClient *c) {
/* ignore MONITOR if aleady slave or in monitor mode */
if (c->flags & REDIS_SLAVE) return;
c->flags |= (REDIS_SLAVE|REDIS_MONITOR);
c->slaveseldb = 0;
listAddNodeTail(server.monitors,c);
addReply(c,shared.ok);
}
/* ============================ Maxmemory directive ======================== */
/* Try to free one object form the pre-allocated objects free list.
* This is useful under low mem conditions as by default we take 1 million
* free objects allocated. On success REDIS_OK is returned, otherwise
* REDIS_ERR. */
int tryFreeOneObjectFromFreelist(void) {
robj *o;
if (server.vm_enabled) pthread_mutex_lock(&server.obj_freelist_mutex);
if (listLength(server.objfreelist)) {
listNode *head = listFirst(server.objfreelist);
o = listNodeValue(head);
listDelNode(server.objfreelist,head);
if (server.vm_enabled) pthread_mutex_unlock(&server.obj_freelist_mutex);
zfree(o);
return REDIS_OK;
} else {
if (server.vm_enabled) pthread_mutex_unlock(&server.obj_freelist_mutex);
return REDIS_ERR;
}
}
/* This function gets called when 'maxmemory' is set on the config file to limit
* the max memory used by the server, and we are out of memory.
* This function will try to, in order:
*
* - Free objects from the free list
* - Try to remove keys with an EXPIRE set
*
* It is not possible to free enough memory to reach used-memory < maxmemory
* the server will start refusing commands that will enlarge even more the
* memory usage.
*/
void freeMemoryIfNeeded(void) {
while (server.maxmemory && zmalloc_used_memory() > server.maxmemory) {
int j, k, freed = 0;
if (tryFreeOneObjectFromFreelist() == REDIS_OK) continue;
for (j = 0; j < server.dbnum; j++) {
int minttl = -1;
robj *minkey = NULL;
struct dictEntry *de;
if (dictSize(server.db[j].expires)) {
freed = 1;
/* From a sample of three keys drop the one nearest to
* the natural expire */
for (k = 0; k < 3; k++) {
time_t t;
de = dictGetRandomKey(server.db[j].expires);
t = (time_t) dictGetEntryVal(de);
if (minttl == -1 || t < minttl) {
minkey = dictGetEntryKey(de);
minttl = t;
}
}
dbDelete(server.db+j,minkey);
}
}
if (!freed) return; /* nothing to free... */
}
}
/* =================================== Main! ================================ */
#ifdef __linux__
int linuxOvercommitMemoryValue(void) {
FILE *fp = fopen("/proc/sys/vm/overcommit_memory","r");
char buf[64];
if (!fp) return -1;
if (fgets(buf,64,fp) == NULL) {
fclose(fp);
return -1;
}
fclose(fp);
return atoi(buf);
}
void linuxOvercommitMemoryWarning(void) {
if (linuxOvercommitMemoryValue() == 0) {
redisLog(REDIS_WARNING,"WARNING overcommit_memory is set to 0! Background save may fail under low memory condition. To fix this issue add 'vm.overcommit_memory = 1' to /etc/sysctl.conf and then reboot or run the command 'sysctl vm.overcommit_memory=1' for this to take effect.");
}
}
#endif /* __linux__ */
void daemonize(void) {
int fd;
FILE *fp;
if (fork() != 0) exit(0); /* parent exits */
setsid(); /* create a new session */
/* Every output goes to /dev/null. If Redis is daemonized but
* the 'logfile' is set to 'stdout' in the configuration file
* it will not log at all. */
if ((fd = open("/dev/null", O_RDWR, 0)) != -1) {
dup2(fd, STDIN_FILENO);
dup2(fd, STDOUT_FILENO);
dup2(fd, STDERR_FILENO);
if (fd > STDERR_FILENO) close(fd);
}
/* Try to write the pid file */
fp = fopen(server.pidfile,"w");
if (fp) {
fprintf(fp,"%d\n",getpid());
fclose(fp);
}
}
void version() {
printf("Redis server version %s (%s:%d)\n", REDIS_VERSION,
redisGitSHA1(), atoi(redisGitDirty()) > 0);
exit(0);
}
void usage() {
fprintf(stderr,"Usage: ./redis-server [/path/to/redis.conf]\n");
fprintf(stderr," ./redis-server - (read config from stdin)\n");
exit(1);
}
int main(int argc, char **argv) {
time_t start;
initServerConfig();
sortCommandTable();
if (argc == 2) {
if (strcmp(argv[1], "-v") == 0 ||
strcmp(argv[1], "--version") == 0) version();
if (strcmp(argv[1], "--help") == 0) usage();
resetServerSaveParams();
loadServerConfig(argv[1]);
} else if ((argc > 2)) {
usage();
} else {
redisLog(REDIS_WARNING,"Warning: no config file specified, using the default config. In order to specify a config file use 'redis-server /path/to/redis.conf'");
}
if (server.daemonize) daemonize();
initServer();
redisLog(REDIS_NOTICE,"Server started, Redis version " REDIS_VERSION);
#ifdef __linux__
linuxOvercommitMemoryWarning();
#endif
start = time(NULL);
if (server.appendonly) {
if (loadAppendOnlyFile(server.appendfilename) == REDIS_OK)
redisLog(REDIS_NOTICE,"DB loaded from append only file: %ld seconds",time(NULL)-start);
} else {
if (rdbLoad(server.dbfilename) == REDIS_OK)
redisLog(REDIS_NOTICE,"DB loaded from disk: %ld seconds",time(NULL)-start);
}
redisLog(REDIS_NOTICE,"The server is now ready to accept connections on port %d", server.port);
aeSetBeforeSleepProc(server.el,beforeSleep);
aeMain(server.el);
aeDeleteEventLoop(server.el);
return 0;
}
/* ============================= Backtrace support ========================= */
#ifdef HAVE_BACKTRACE
void *getMcontextEip(ucontext_t *uc) {
#if defined(__FreeBSD__)
return (void*) uc->uc_mcontext.mc_eip;
#elif defined(__dietlibc__)
return (void*) uc->uc_mcontext.eip;
#elif defined(__APPLE__) && !defined(MAC_OS_X_VERSION_10_6)
#if __x86_64__
return (void*) uc->uc_mcontext->__ss.__rip;
#else
return (void*) uc->uc_mcontext->__ss.__eip;
#endif
#elif defined(__APPLE__) && defined(MAC_OS_X_VERSION_10_6)
#if defined(_STRUCT_X86_THREAD_STATE64) && !defined(__i386__)
return (void*) uc->uc_mcontext->__ss.__rip;
#else
return (void*) uc->uc_mcontext->__ss.__eip;
#endif
#elif defined(__i386__)
return (void*) uc->uc_mcontext.gregs[14]; /* Linux 32 */
#elif defined(__X86_64__) || defined(__x86_64__)
return (void*) uc->uc_mcontext.gregs[16]; /* Linux 64 */
#elif defined(__ia64__) /* Linux IA64 */
return (void*) uc->uc_mcontext.sc_ip;
#else
return NULL;
#endif
}
void segvHandler(int sig, siginfo_t *info, void *secret) {
void *trace[100];
char **messages = NULL;
int i, trace_size = 0;
ucontext_t *uc = (ucontext_t*) secret;
sds infostring;
REDIS_NOTUSED(info);
redisLog(REDIS_WARNING,
"======= Ooops! Redis %s got signal: -%d- =======", REDIS_VERSION, sig);
infostring = genRedisInfoString();
redisLog(REDIS_WARNING, "%s",infostring);
/* It's not safe to sdsfree() the returned string under memory
* corruption conditions. Let it leak as we are going to abort */
trace_size = backtrace(trace, 100);
/* overwrite sigaction with caller's address */
if (getMcontextEip(uc) != NULL) {
trace[1] = getMcontextEip(uc);
}
messages = backtrace_symbols(trace, trace_size);
for (i=1; i<trace_size; ++i)
redisLog(REDIS_WARNING,"%s", messages[i]);
/* free(messages); Don't call free() with possibly corrupted memory. */
_exit(0);
}
void sigtermHandler(int sig) {
REDIS_NOTUSED(sig);
redisLog(REDIS_WARNING,"SIGTERM received, scheduling shutting down...");
server.shutdown_asap = 1;
}
void setupSigSegvAction(void) {
struct sigaction act;
sigemptyset (&act.sa_mask);
/* When the SA_SIGINFO flag is set in sa_flags then sa_sigaction
* is used. Otherwise, sa_handler is used */
act.sa_flags = SA_NODEFER | SA_ONSTACK | SA_RESETHAND | SA_SIGINFO;
act.sa_sigaction = segvHandler;
sigaction (SIGSEGV, &act, NULL);
sigaction (SIGBUS, &act, NULL);
sigaction (SIGFPE, &act, NULL);
sigaction (SIGILL, &act, NULL);
sigaction (SIGBUS, &act, NULL);
act.sa_flags = SA_NODEFER | SA_ONSTACK | SA_RESETHAND;
act.sa_handler = sigtermHandler;
sigaction (SIGTERM, &act, NULL);
return;
}
#else /* HAVE_BACKTRACE */
void setupSigSegvAction(void) {
}
#endif /* HAVE_BACKTRACE */
/* The End */