size_t zmalloc_get_rss(void) {
/* If we can't get the RSS in an OS-specific way for this system just
* * return the memory usage we estimated in zmalloc()..
* *
* * Fragmentation will appear to be always 1 (no fragmentation)
* * of course... */
return zmalloc_used_memory();
}
/* This gets called when the function "client)" gets called in Lua */
int luaCall_client(lua_State *L) {
LuaFlag = PIPE_NONE_FLAG;
int argc = lua_gettop(L);
const char *arg1 = lua_tostring(L, 1);
if (!arg1) {
return redisLuaArityErr(L, NULL);
}
redisCommand *cmd = lookupCommand((char *)arg1);
if (!cmd) {
char buf[64];
snprintf(buf, 63, "-ERR Unknown command '%s'\r\n", arg1);
buf[63] = '\0';
lua_pushstring(L, buf);
LuaFlag = PIPE_ONE_LINER_FLAG;
return 1;
} else if ((cmd->arity > 0 && cmd->arity != argc) || (argc < -cmd->arity)) {
return redisLuaArityErr(L, cmd->name);
}
if (server.maxmemory && (cmd->flags & REDIS_CMD_DENYOOM) &&
(zmalloc_used_memory() > server.maxmemory)) {
LuaFlag = PIPE_ONE_LINER_FLAG;
lua_pushstring(L,
"-ERR command not allowed when used memory > 'maxmemory'\r\n");
return 1;
}
if (server.vm_enabled && server.vm_max_threads > 0 &&
blockClientOnSwappedKeys(LuaClient, cmd)) return 1;
long ok = 0; /* must come before first GOTO */
redisClient *rfc = rsql_createFakeClient();
robj **rargv = zmalloc(sizeof(robj *) * argc);
for (int i = 0; i < argc; i++) {
char *arg = (char *)lua_tostring(L, i + 1);
if (!arg) {
char *lbuf = "args must be strings";
luaL_argerror (L, i, lbuf);
LuaFlag = PIPE_ONE_LINER_FLAG;
ok = 1;
goto redis_lua_end;
}
rargv[i] = _createStringObject(arg);
}
rfc->argv = rargv;
rfc->argc = argc;
ok = fakeClientPipe(LuaClient, rfc, L, 0, &LuaFlag, luaLine, emptyNoop);
redis_lua_end:
for (int i = 0; i < argc; i++) decrRefCount(rargv[i]);
zfree(rargv);
rsql_freeFakeClient(rfc);
return (ok > 0) ? 1 : 0;
}
/* This is how rewriting of the append only file in background works:
*
* 1) The user calls BGREWRITEAOF
* 2) Redis calls this function, that forks():
* 2a) the child rewrite the append only file in a temp file.
* 2b) the parent accumulates differences in server.aof_rewrite_buf.
* 3) When the child finished '2a' exists.
* 4) The parent will trap the exit code, if it's OK, will append the
* data accumulated into server.aof_rewrite_buf into the temp file, and
* finally will rename(2) the temp file in the actual file name.
* The the new file is reopened as the new append only file. Profit!
*/
int rewriteAppendOnlyFileBackground(void) {
pid_t childpid;
long long start;
if (server.aof_child_pid != -1) return REDIS_ERR;
start = ustime();
if ((childpid = fork()) == 0) {
char tmpfile[256];
/* Child */
closeListeningSockets(0);
redisSetProcTitle("redis-aof-rewrite");
snprintf(tmpfile,256,"temp-rewriteaof-bg-%d.aof", (int) getpid());
if (rewriteAppendOnlyFile(tmpfile) == REDIS_OK) {
size_t private_dirty = zmalloc_get_private_dirty();
if (private_dirty) {
redisLog(REDIS_NOTICE,
"AOF rewrite: %zu MB of memory used by copy-on-write",
private_dirty/(1024*1024));
}
exitFromChild(0);
} else {
exitFromChild(1);
}
} else {
/* Parent */
server.stat_fork_time = ustime()-start;
server.stat_fork_rate = (double) zmalloc_used_memory() * 1000000 / server.stat_fork_time / (1024*1024*1024); /* GB per second. */
latencyAddSampleIfNeeded("fork",server.stat_fork_time/1000);
if (childpid == -1) {
redisLog(REDIS_WARNING,
"Can't rewrite append only file in background: fork: %s",
strerror(errno));
return REDIS_ERR;
}
redisLog(REDIS_NOTICE,
"Background append only file rewriting started by pid %d",childpid);
server.aof_rewrite_scheduled = 0;
server.aof_rewrite_time_start = time(NULL);
server.aof_child_pid = childpid;
updateDictResizePolicy();
/* We set appendseldb to -1 in order to force the next call to the
* feedAppendOnlyFile() to issue a SELECT command, so the differences
* accumulated by the parent into server.aof_rewrite_buf will start
* with a SELECT statement and it will be safe to merge. */
server.aof_selected_db = -1;
replicationScriptCacheFlush();
return REDIS_OK;
}
return REDIS_OK; /* unreached */
}
/* Fragmentation = RSS / allocated-bytes */
float zmalloc_get_fragmentation_ratio(void) {
return (float)zmalloc_get_rss()/zmalloc_used_memory();
}
/* Fragmentation = RSS / allocated-bytes */
float zmalloc_get_fragmentation_ratio(size_t rss) {
return (float)rss/zmalloc_used_memory();
}
/* Replay the append log file. On error REDIS_OK is returned. On non fatal
* error (the append only file is zero-length) REDIS_ERR is returned. On
* fatal error an error message is logged and the program exists. */
int loadAppendOnlyFile(char *filename) {
struct redisClient *fakeClient;
FILE *fp = fopen(filename,"r");
struct redis_stat sb;
int appendonly = server.appendonly;
long loops = 0;
if (fp && redis_fstat(fileno(fp),&sb) != -1 && sb.st_size == 0) {
server.appendonly_current_size = 0;
fclose(fp);
return REDIS_ERR;
}
if (fp == NULL) {
redisLog(REDIS_WARNING,"Fatal error: can't open the append log file for reading: %s",strerror(errno));
exit(1);
}
/* Temporarily disable AOF, to prevent EXEC from feeding a MULTI
* to the same file we're about to read. */
server.appendonly = 0;
fakeClient = createFakeClient();
startLoading(fp);
while(1) {
int argc, j;
unsigned long len;
robj **argv;
char buf[128];
sds argsds;
struct redisCommand *cmd;
int force_swapout;
/* Serve the clients from time to time */
if (!(loops++ % 1000)) {
loadingProgress(ftello(fp));
aeProcessEvents(server.el, AE_FILE_EVENTS|AE_DONT_WAIT);
}
if (fgets(buf,sizeof(buf),fp) == NULL) {
if (feof(fp))
break;
else
goto readerr;
}
if (buf[0] != '*') goto fmterr;
argc = atoi(buf+1);
if (argc < 1) goto fmterr;
argv = zmalloc(sizeof(robj*)*argc);
for (j = 0; j < argc; j++) {
if (fgets(buf,sizeof(buf),fp) == NULL) goto readerr;
if (buf[0] != '$') goto fmterr;
len = strtol(buf+1,NULL,10);
argsds = sdsnewlen(NULL,len);
if (len && fread(argsds,len,1,fp) == 0) goto fmterr;
argv[j] = createObject(REDIS_STRING,argsds);
if (fread(buf,2,1,fp) == 0) goto fmterr; /* discard CRLF */
}
/* Command lookup */
cmd = lookupCommand(argv[0]->ptr);
if (!cmd) {
redisLog(REDIS_WARNING,"Unknown command '%s' reading the append only file", argv[0]->ptr);
exit(1);
}
/* Run the command in the context of a fake client */
fakeClient->argc = argc;
fakeClient->argv = argv;
cmd->proc(fakeClient);
/* The fake client should not have a reply */
redisAssert(fakeClient->bufpos == 0 && listLength(fakeClient->reply) == 0);
/* The fake client should never get blocked */
redisAssert((fakeClient->flags & REDIS_BLOCKED) == 0);
/* Clean up. Command code may have changed argv/argc so we use the
* argv/argc of the client instead of the local variables. */
for (j = 0; j < fakeClient->argc; j++)
decrRefCount(fakeClient->argv[j]);
zfree(fakeClient->argv);
/* Handle swapping while loading big datasets when VM is on */
force_swapout = 0;
if ((zmalloc_used_memory() - server.vm_max_memory) > 1024*1024*32)
force_swapout = 1;
if (server.vm_enabled && force_swapout) {
while (zmalloc_used_memory() > server.vm_max_memory) {
if (vmSwapOneObjectBlocking() == REDIS_ERR) break;
}
}
}
/* This point can only be reached when EOF is reached without errors.
* If the client is in the middle of a MULTI/EXEC, log error and quit. */
if (fakeClient->flags & REDIS_MULTI) goto readerr;
fclose(fp);
freeFakeClient(fakeClient);
server.appendonly = appendonly;
stopLoading();
aofUpdateCurrentSize();
server.auto_aofrewrite_base_size = server.appendonly_current_size;
return REDIS_OK;
readerr:
if (feof(fp)) {
redisLog(REDIS_WARNING,"Unexpected end of file reading the append only file");
} else {
redisLog(REDIS_WARNING,"Unrecoverable error reading the append only file: %s", strerror(errno));
}
exit(1);
fmterr:
redisLog(REDIS_WARNING,"Bad file format reading the append only file: make a backup of your AOF file, then use ./redis-check-aof --fix <filename>");
exit(1);
}
int freeMemoryIfNeeded(void) {
size_t mem_reported, mem_used, mem_tofree, mem_freed;
int slaves = listLength(server.slaves);
mstime_t latency, eviction_latency;
long long delta;
/* Check if we are over the memory usage limit. If we are not, no need
* to subtract the slaves output buffers. We can just return ASAP. */
mem_reported = zmalloc_used_memory();
if (mem_reported <= server.maxmemory) return C_OK;
/* Remove the size of slaves output buffers and AOF buffer from the
* count of used memory. */
mem_used = mem_reported;
if (slaves) {
listIter li;
listNode *ln;
listRewind(server.slaves,&li);
while((ln = listNext(&li))) {
client *slave = listNodeValue(ln);
unsigned long obuf_bytes = getClientOutputBufferMemoryUsage(slave);
if (obuf_bytes > mem_used)
mem_used = 0;
else
mem_used -= obuf_bytes;
}
}
if (server.aof_state != AOF_OFF) {
mem_used -= sdslen(server.aof_buf);
mem_used -= aofRewriteBufferSize();
}
/* Check if we are still over the memory limit. */
if (mem_used <= server.maxmemory) return C_OK;
/* Compute how much memory we need to free. */
mem_tofree = mem_used - server.maxmemory;
mem_freed = 0;
if (server.maxmemory_policy == MAXMEMORY_NO_EVICTION)
goto cant_free; /* We need to free memory, but policy forbids. */
latencyStartMonitor(latency);
while (mem_freed < mem_tofree) {
int j, k, i, keys_freed = 0;
static int next_db = 0;
sds bestkey = NULL;
int bestdbid;
redisDb *db;
dict *dict;
dictEntry *de;
if (server.maxmemory_policy == MAXMEMORY_ALLKEYS_LRU ||
server.maxmemory_policy == MAXMEMORY_VOLATILE_LRU)
{
struct evictionPoolEntry *pool = EvictionPoolLRU;
while(bestkey == NULL) {
unsigned long total_keys = 0, keys;
/* We don't want to make local-db choices when expiring keys,
* so to start populate the eviction pool sampling keys from
* every DB. */
for (i = 0; i < server.dbnum; i++) {
db = server.db+i;
dict = (server.maxmemory_policy == MAXMEMORY_ALLKEYS_LRU) ?
db->dict : db->expires;
if ((keys = dictSize(dict)) != 0) {
evictionPoolPopulate(i, dict, db->dict, pool);
total_keys += keys;
}
}
if (!total_keys) break; /* No keys to evict. */
/* Go backward from best to worst element to evict. */
for (k = EVPOOL_SIZE-1; k >= 0; k--) {
if (pool[k].key == NULL) continue;
bestdbid = pool[k].dbid;
if (server.maxmemory_policy == MAXMEMORY_ALLKEYS_LRU) {
de = dictFind(server.db[pool[k].dbid].dict,
pool[k].key);
} else {
de = dictFind(server.db[pool[k].dbid].expires,
pool[k].key);
}
/* Remove the entry from the pool. */
if (pool[k].key != pool[k].cached)
sdsfree(pool[k].key);
pool[k].key = NULL;
pool[k].idle = 0;
/* If the key exists, is our pick. Otherwise it is
* a ghost and we need to try the next element. */
if (de) {
bestkey = dictGetKey(de);
break;
} else {
/* Ghost... Iterate again. */
}
}
}
}
/* volatile-random and allkeys-random policy */
else if (server.maxmemory_policy == MAXMEMORY_ALLKEYS_RANDOM ||
server.maxmemory_policy == MAXMEMORY_VOLATILE_RANDOM)
{
/* When evicting a random key, we try to evict a key for
* each DB, so we use the static 'next_db' variable to
* incrementally visit all DBs. */
for (i = 0; i < server.dbnum; i++) {
j = (++next_db) % server.dbnum;
db = server.db+j;
dict = (server.maxmemory_policy == MAXMEMORY_ALLKEYS_RANDOM) ?
db->dict : db->expires;
if (dictSize(dict) != 0) {
de = dictGetRandomKey(dict);
bestkey = dictGetKey(de);
bestdbid = j;
break;
}
}
}
/* volatile-ttl */
else if (server.maxmemory_policy == MAXMEMORY_VOLATILE_TTL) {
long bestttl = 0; /* Initialized to avoid warning. */
/* In this policy we scan a single DB per iteration (visiting
* a different DB per call), expiring the key with the smallest
* TTL among the few sampled.
*
* Note that this algorithm makes local-DB choices, and should
* use a pool and code more similr to the one used in the
* LRU eviction policies in the future. */
for (i = 0; i < server.dbnum; i++) {
j = (++next_db) % server.dbnum;
db = server.db+j;
dict = db->expires;
if (dictSize(dict) != 0) {
for (k = 0; k < server.maxmemory_samples; k++) {
sds thiskey;
long thisttl;
de = dictGetRandomKey(dict);
thiskey = dictGetKey(de);
thisttl = (long) dictGetVal(de);
/* Keys expiring sooner (smaller unix timestamp) are
* better candidates for deletion */
if (bestkey == NULL || thisttl < bestttl) {
bestkey = thiskey;
bestttl = thisttl;
bestdbid = j;
}
}
}
}
}
/* Finally remove the selected key. */
if (bestkey) {
db = server.db+bestdbid;
robj *keyobj = createStringObject(bestkey,sdslen(bestkey));
propagateExpire(db,keyobj,server.lazyfree_lazy_eviction);
/* We compute the amount of memory freed by db*Delete() alone.
* It is possible that actually the memory needed to propagate
* the DEL in AOF and replication link is greater than the one
* we are freeing removing the key, but we can't account for
* that otherwise we would never exit the loop.
*
* AOF and Output buffer memory will be freed eventually so
* we only care about memory used by the key space. */
delta = (long long) zmalloc_used_memory();
latencyStartMonitor(eviction_latency);
if (server.lazyfree_lazy_eviction)
dbAsyncDelete(db,keyobj);
else
dbSyncDelete(db,keyobj);
latencyEndMonitor(eviction_latency);
latencyAddSampleIfNeeded("eviction-del",eviction_latency);
latencyRemoveNestedEvent(latency,eviction_latency);
delta -= (long long) zmalloc_used_memory();
mem_freed += delta;
server.stat_evictedkeys++;
notifyKeyspaceEvent(NOTIFY_EVICTED, "evicted",
keyobj, db->id);
decrRefCount(keyobj);
keys_freed++;
/* When the memory to free starts to be big enough, we may
* start spending so much time here that is impossible to
* deliver data to the slaves fast enough, so we force the
* transmission here inside the loop. */
if (slaves) flushSlavesOutputBuffers();
}
if (!keys_freed) {
latencyEndMonitor(latency);
latencyAddSampleIfNeeded("eviction-cycle",latency);
goto cant_free; /* nothing to free... */
}
}
latencyEndMonitor(latency);
latencyAddSampleIfNeeded("eviction-cycle",latency);
return C_OK;
cant_free:
/* We are here if we are not able to reclaim memory. There is only one
* last thing we can try: check if the lazyfree thread has jobs in queue
* and wait... */
while(bioPendingJobsOfType(BIO_LAZY_FREE)) {
if (((mem_reported - zmalloc_used_memory()) + mem_freed) >= mem_tofree)
break;
usleep(1000);
}
return C_ERR;
}
/* Every time a thread finished a Job, it writes a byte into the write side
* of an unix pipe in order to "awake" the main thread, and this function
* is called.
*
* Note that this is called both by the event loop, when a I/O thread
* sends a byte in the notification pipe, and is also directly called from
* waitEmptyIOJobsQueue().
*
* In the latter case we don't want to swap more, so we use the
* "privdata" argument setting it to a not NULL value to signal this
* condition. */
void vmThreadedIOCompletedJob(aeEventLoop *el, int fd, void *privdata,
int mask)
{
char buf[1];
int retval, processed = 0, toprocess = -1, trytoswap = 1;
REDIS_NOTUSED(el);
REDIS_NOTUSED(mask);
REDIS_NOTUSED(privdata);
if (privdata != NULL) trytoswap = 0; /* check the comments above... */
/* For every byte we read in the read side of the pipe, there is one
* I/O job completed to process. */
#ifndef _WIN32
while((retval = read(fd,buf,1)) == 1) {
#else
DWORD pipe_is_on = 0;
while (1) {
retval = 0;
/*Windows fix: We need to peek pipe, since read would block. */
if (!PeekNamedPipe((HANDLE) _get_osfhandle(fd), NULL, 0, NULL, &pipe_is_on, NULL)) {
redisLog(REDIS_DEBUG,"PeekReadPipe failed %s", strerror(GetLastError()));
break;
}
/* No data on pipe */
if (!pipe_is_on)
break;
if ((retval = read(fd,buf,1)) != 1)
break;
#endif
iojob *j;
listNode *ln;
struct dictEntry *de;
/* Get the processed element (the oldest one) */
lockThreadedIO();
redisLog(REDIS_DEBUG,"Processing I/O completed job");
redisAssert(listLength(server.io_processed) != 0);
if (toprocess == -1) {
toprocess = (listLength(server.io_processed)*REDIS_MAX_COMPLETED_JOBS_PROCESSED)/100;
if (toprocess <= 0) toprocess = 1;
}
ln = listFirst(server.io_processed);
j = ln->value;
listDelNode(server.io_processed,ln);
unlockThreadedIO();
/* If this job is marked as canceled, just ignore it */
if (j->canceled) {
freeIOJob(j);
continue;
}
/* Post process it in the main thread, as there are things we
* can do just here to avoid race conditions and/or invasive locks */
redisLog(REDIS_DEBUG,"COMPLETED Job type: %d, ID %p, key: %s", j->type, (void*)j->id, (unsigned char*)j->key->ptr);
de = dictFind(j->db->dict,j->key->ptr);
redisAssert(de != NULL);
if (j->type == REDIS_IOJOB_LOAD) {
redisDb *db;
vmpointer *vp = dictGetEntryVal(de);
/* Key loaded, bring it at home */
vmMarkPagesFree(vp->page,vp->usedpages);
redisLog(REDIS_DEBUG, "VM: object %s loaded from disk (threaded)",
(unsigned char*) j->key->ptr);
server.vm_stats_swapped_objects--;
server.vm_stats_swapins++;
dictGetEntryVal(de) = j->val;
incrRefCount(j->val);
db = j->db;
/* Handle clients waiting for this key to be loaded. */
handleClientsBlockedOnSwappedKey(db,j->key);
freeIOJob(j);
zfree(vp);
} else if (j->type == REDIS_IOJOB_PREPARE_SWAP) {
/* Now we know the amount of pages required to swap this object.
* Let's find some space for it, and queue this task again
* rebranded as REDIS_IOJOB_DO_SWAP. */
if (!vmCanSwapOut() ||
vmFindContiguousPages(&j->page,j->pages) == REDIS_ERR)
{
/* Ooops... no space or we can't swap as there is
* a fork()ed Redis trying to save stuff on disk. */
j->val->storage = REDIS_VM_MEMORY; /* undo operation */
freeIOJob(j);
} else {
/* Note that we need to mark this pages as used now,
* if the job will be canceled, we'll mark them as freed
* again. */
vmMarkPagesUsed(j->page,j->pages);
j->type = REDIS_IOJOB_DO_SWAP;
lockThreadedIO();
queueIOJob(j);
unlockThreadedIO();
}
} else if (j->type == REDIS_IOJOB_DO_SWAP) {
vmpointer *vp;
/* Key swapped. We can finally free some memory. */
if (j->val->storage != REDIS_VM_SWAPPING) {
vmpointer *vp = (vmpointer*) j->id;
printf("storage: %d\n",vp->storage);
printf("key->name: %s\n",(char*)j->key->ptr);
printf("val: %p\n",(void*)j->val);
printf("val->type: %d\n",j->val->type);
printf("val->ptr: %s\n",(char*)j->val->ptr);
}
redisAssert(j->val->storage == REDIS_VM_SWAPPING);
vp = createVmPointer(j->val);
vp->page = j->page;
vp->usedpages = j->pages;
dictGetEntryVal(de) = vp;
/* Fix the storage otherwise decrRefCount will attempt to
* remove the associated I/O job */
j->val->storage = REDIS_VM_MEMORY;
decrRefCount(j->val);
redisLog(REDIS_DEBUG,
"VM: object %s swapped out at %lld (%lld pages) (threaded)",
(unsigned char*) j->key->ptr,
(unsigned long long) j->page, (unsigned long long) j->pages);
server.vm_stats_swapped_objects++;
server.vm_stats_swapouts++;
freeIOJob(j);
/* Put a few more swap requests in queue if we are still
* out of memory */
if (trytoswap && vmCanSwapOut() &&
zmalloc_used_memory() > server.vm_max_memory)
{
int more = 1;
while(more) {
lockThreadedIO();
more = listLength(server.io_newjobs) <
(unsigned) server.vm_max_threads;
unlockThreadedIO();
/* Don't waste CPU time if swappable objects are rare. */
if (vmSwapOneObjectThreaded() == REDIS_ERR) {
trytoswap = 0;
break;
}
}
}
}
processed++;
if (processed == toprocess) return;
}
if (retval < 0 && errno != EAGAIN) {
redisLog(REDIS_WARNING,
"WARNING: read(2) error in vmThreadedIOCompletedJob() %s",
strerror(errno));
}
}
void lockThreadedIO(void) {
pthread_mutex_lock(&server.io_mutex);
}
int freeMemoryIfNeeded(void) {
size_t mem_reported, mem_used, mem_tofree, mem_freed;
mstime_t latency, eviction_latency;
long long delta;
int slaves = listLength(server.slaves);
/* When clients are paused the dataset should be static not just from the
* POV of clients not being able to write, but also from the POV of
* expires and evictions of keys not being performed. */
if (clientsArePaused()) return C_OK;
/* Check if we are over the memory usage limit. If we are not, no need
* to subtract the slaves output buffers. We can just return ASAP. */
mem_reported = zmalloc_used_memory();
if (mem_reported <= server.maxmemory) return C_OK;
/* Remove the size of slaves output buffers and AOF buffer from the
* count of used memory. */
mem_used = mem_reported;
size_t overhead = freeMemoryGetNotCountedMemory();
mem_used = (mem_used > overhead) ? mem_used-overhead : 0;
/* Check if we are still over the memory limit. */
if (mem_used <= server.maxmemory) return C_OK;
/* Compute how much memory we need to free. */
mem_tofree = mem_used - server.maxmemory;
mem_freed = 0;
if (server.maxmemory_policy == MAXMEMORY_NO_EVICTION)
goto cant_free; /* We need to free memory, but policy forbids. */
latencyStartMonitor(latency);
while (mem_freed < mem_tofree) {
int j, k, i, keys_freed = 0;
static int next_db = 0;
sds bestkey = NULL;
int bestdbid;
redisDb *db;
dict *dict;
dictEntry *de;
if (server.maxmemory_policy & (MAXMEMORY_FLAG_LRU|MAXMEMORY_FLAG_LFU) ||
server.maxmemory_policy == MAXMEMORY_VOLATILE_TTL)
{
struct evictionPoolEntry *pool = EvictionPoolLRU;
while(bestkey == NULL) {
unsigned long total_keys = 0, keys;
/* We don't want to make local-db choices when expiring keys,
* so to start populate the eviction pool sampling keys from
* every DB. */
for (i = 0; i < server.dbnum; i++) {
db = server.db+i;
dict = (server.maxmemory_policy & MAXMEMORY_FLAG_ALLKEYS) ?
db->dict : db->expires;
if ((keys = dictSize(dict)) != 0) {
evictionPoolPopulate(i, dict, db->dict, pool);
total_keys += keys;
}
}
if (!total_keys) break; /* No keys to evict. */
/* Go backward from best to worst element to evict. */
for (k = EVPOOL_SIZE-1; k >= 0; k--) {
if (pool[k].key == NULL) continue;
bestdbid = pool[k].dbid;
if (server.maxmemory_policy & MAXMEMORY_FLAG_ALLKEYS) {
de = dictFind(server.db[pool[k].dbid].dict,
pool[k].key);
} else {
de = dictFind(server.db[pool[k].dbid].expires,
pool[k].key);
}
/* Remove the entry from the pool. */
if (pool[k].key != pool[k].cached)
sdsfree(pool[k].key);
pool[k].key = NULL;
pool[k].idle = 0;
/* If the key exists, is our pick. Otherwise it is
* a ghost and we need to try the next element. */
if (de) {
bestkey = dictGetKey(de);
break;
} else {
/* Ghost... Iterate again. */
}
}
}
}
/* volatile-random and allkeys-random policy */
else if (server.maxmemory_policy == MAXMEMORY_ALLKEYS_RANDOM ||
server.maxmemory_policy == MAXMEMORY_VOLATILE_RANDOM)
{
/* When evicting a random key, we try to evict a key for
* each DB, so we use the static 'next_db' variable to
* incrementally visit all DBs. */
for (i = 0; i < server.dbnum; i++) {
j = (++next_db) % server.dbnum;
db = server.db+j;
dict = (server.maxmemory_policy == MAXMEMORY_ALLKEYS_RANDOM) ?
db->dict : db->expires;
if (dictSize(dict) != 0) {
de = dictGetRandomKey(dict);
bestkey = dictGetKey(de);
bestdbid = j;
break;
}
}
}
/* Finally remove the selected key. */
if (bestkey) {
db = server.db+bestdbid;
robj *keyobj = createStringObject(bestkey,sdslen(bestkey));
propagateExpire(db,keyobj,server.lazyfree_lazy_eviction);
/* We compute the amount of memory freed by db*Delete() alone.
* It is possible that actually the memory needed to propagate
* the DEL in AOF and replication link is greater than the one
* we are freeing removing the key, but we can't account for
* that otherwise we would never exit the loop.
*
* AOF and Output buffer memory will be freed eventually so
* we only care about memory used by the key space. */
delta = (long long) zmalloc_used_memory();
latencyStartMonitor(eviction_latency);
if (server.lazyfree_lazy_eviction)
dbAsyncDelete(db,keyobj);
else
dbSyncDelete(db,keyobj);
latencyEndMonitor(eviction_latency);
latencyAddSampleIfNeeded("eviction-del",eviction_latency);
latencyRemoveNestedEvent(latency,eviction_latency);
delta -= (long long) zmalloc_used_memory();
mem_freed += delta;
server.stat_evictedkeys++;
notifyKeyspaceEvent(NOTIFY_EVICTED, "evicted",
keyobj, db->id);
decrRefCount(keyobj);
keys_freed++;
/* When the memory to free starts to be big enough, we may
* start spending so much time here that is impossible to
* deliver data to the slaves fast enough, so we force the
* transmission here inside the loop. */
if (slaves) flushSlavesOutputBuffers();
/* Normally our stop condition is the ability to release
* a fixed, pre-computed amount of memory. However when we
* are deleting objects in another thread, it's better to
* check, from time to time, if we already reached our target
* memory, since the "mem_freed" amount is computed only
* across the dbAsyncDelete() call, while the thread can
* release the memory all the time. */
if (server.lazyfree_lazy_eviction && !(keys_freed % 16)) {
overhead = freeMemoryGetNotCountedMemory();
mem_used = zmalloc_used_memory();
mem_used = (mem_used > overhead) ? mem_used-overhead : 0;
if (mem_used <= server.maxmemory) {
mem_freed = mem_tofree;
}
}
}
if (!keys_freed) {
latencyEndMonitor(latency);
latencyAddSampleIfNeeded("eviction-cycle",latency);
goto cant_free; /* nothing to free... */
}
}
latencyEndMonitor(latency);
latencyAddSampleIfNeeded("eviction-cycle",latency);
return C_OK;
cant_free:
/* We are here if we are not able to reclaim memory. There is only one
* last thing we can try: check if the lazyfree thread has jobs in queue
* and wait... */
while(bioPendingJobsOfType(BIO_LAZY_FREE)) {
if (((mem_reported - zmalloc_used_memory()) + mem_freed) >= mem_tofree)
break;
usleep(1000);
}
return C_ERR;
}
static bool sendRestAPIReply(cli *c, sds file) { //printf("sendRestAPIReply\n");
int argc; bool ret = 0;
sds pb = c->http.post_body; //TODO cat [file,pb] is too much copying
sds url = pb ? sdscatprintf(sdsempty(), "%s%s", file, pb) :
sdsdup(file); //FREE 156
sds *argv = sdssplitlen(url, sdslen(url), "/", 1, &argc); //FREE 157
rcommand *cmd = lookupCommand(argv[0]);
if (!cmd) goto send_rest_end;
ret = 1;
//printf("sendRestAPIReply: found cmd: %s\n", cmd->name);
if ((cmd->arity > 0 && cmd->arity != argc) || (argc < -cmd->arity)) {
addReplyErrorFormat(c,"wrong number of arguments for '%s' command",
cmd->name);
goto send_rest_end;
}
//NOTE: rest is copy of redis.c processCommand()
if (server.maxmemory) freeMemoryIfNeeded();
if (server.maxmemory && (cmd->flags & REDIS_CMD_DENYOOM) &&
zmalloc_used_memory() > server.maxmemory) {
addReplyError(c, "command not allowed when used memory > 'maxmemory'");
goto send_rest_end;
}
if ((dictSize(c->pubsub_channels) > 0 || listLength(c->pubsub_patterns) > 0)
&&
cmd->proc != subscribeCommand && cmd->proc != unsubscribeCommand &&
cmd->proc != psubscribeCommand && cmd->proc != punsubscribeCommand) {
addReplyError(c,
"only (P)SUBSCRIBE / (P)UNSUBSCRIBE / QUIT allowed in this context");
goto send_rest_end;
}
if (server.masterhost && server.replstate != REDIS_REPL_CONNECTED &&
server.repl_serve_stale_data == 0 &&
cmd->proc != infoCommand && cmd->proc != slaveofCommand) {
addReplyError(c,
"link with MASTER is down and slave-serve-stale-data is set to no");
goto send_rest_end;
}
if (server.loading && cmd->proc != infoCommand) {
addReply(c, shared.loadingerr);
goto send_rest_end;
}
if (c->flags & REDIS_MULTI &&
cmd->proc != execCommand && cmd->proc != discardCommand &&
cmd->proc != multiCommand && cmd->proc != watchCommand) {
queueMultiCommand(c, cmd); addReply(c, shared.queued);
goto send_rest_end;
}
listNode *ln; listIter *li; cli *restc = server.alc.RestClient;
// 1.) call() cmd in RestClient
{ // REST CLIENT call
robj **rargv = zmalloc(sizeof(robj *) * argc);
for (int i = 0; i < argc; i++) {
rargv[i] = createStringObject(argv[i], sdslen(argv[i]));
}
restc->argc = argc; restc->argv = rargv; call(restc, cmd);
for (int i = 0; i < argc; i++) decrRefCount(rargv[i]);
zfree(rargv);
}
// 2.) calculate Content-Length from RestClient's response
ulong brlen = restc->bufpos; ulong trlen = brlen;
if (restc->reply->len) {
li = listGetIterator(restc->reply, AL_START_HEAD);
while((ln = listNext(li))) {
robj *r = ln->value; trlen += sdslen(r->ptr);
} listReleaseIterator(li);
}
bool err = brlen && (*restc->buf == '-');
//TODO check for "+OK" and return 201 w/ no body
// 3.) create header w/ Content-Length
sds s = err ? send_http404_reponse_header(c, trlen) :
send_http200_reponse_header(c, trlen);
robj *ho = createObject(REDIS_STRING, s);
addReply(c, ho); decrRefCount(ho);
// 4.) tack on RestClient's response as HTTP Body
if (brlen) { addReplyString(c, restc->buf, brlen); }
if (restc->reply->len) {
li = listGetIterator(restc->reply, AL_START_HEAD);
while((ln = listNext(li))) {
robj *r = ln->value; addReply(c, r);
} listReleaseIterator(li);
}
// 5.) reset RestClient
restc->bufpos = 0;
while (restc->reply->len) {
listDelNode(restc->reply, listFirst(restc->reply));
}
send_rest_end:
sdsfree(url); // FREE 156
for (int i = 0; i < argc; i++) sdsfree(argv[i]); zfree(argv); // FREE 157
return ret;
}
/* Replay the append log file. On error REDIS_OK is returned. On non fatal
* error (the append only file is zero-length) REDIS_ERR is returned. On
* fatal error an error message is logged and the program exists. */
int loadAppendOnlyFile(char *filename) {
struct redisClient *fakeClient;
FILE *fp = fopen(filename,"r");
struct redis_stat sb;
unsigned long long loadedkeys = 0;
if (redis_fstat(fileno(fp),&sb) != -1 && sb.st_size == 0)
return REDIS_ERR;
fakeClient = createFakeClient();
while(1) {
int argc, j;
unsigned long len;
robj **argv;
char buf[128];
sds argsds;
struct redisCommand *cmd;
if (fgets(buf,sizeof(buf),fp) == NULL) {
if (feof(fp))
break;
else
goto readerr;
}
if (buf[0] != '*') goto fmterr;
argc = atoi(buf+1);
argv = zmalloc(sizeof(robj*)*argc);
for (j = 0; j < argc; j++) {
if (fgets(buf,sizeof(buf),fp) == NULL) goto readerr;
if (buf[0] != '$') goto fmterr;
len = strtol(buf+1,NULL,10);
argsds = sdsnewlen(NULL,len);
if (len && fread(argsds,len,1,fp) == 0) goto fmterr;
argv[j] = createObject(REDIS_STRING,argsds);
if (fread(buf,2,1,fp) == 0) goto fmterr; /* discard CRLF */
}
/* Command lookup */
cmd = lookupCommand(argv[0]->ptr);
/* Try object sharing and encoding */
if (server.shareobjects) {
int j;
for(j = 1; j < argc; j++)
argv[j] = tryObjectSharing(argv[j]);
}
if (cmd->flags & REDIS_CMD_BULK)
tryObjectEncoding(argv[argc-1]);
/* Run the command in the context of a fake client */
fakeClient->argc = argc;
fakeClient->argv = argv;
cmd->proc(fakeClient);
/* Discard the reply objects list from the fake client */
while(listLength(fakeClient->reply))
listDelNode(fakeClient->reply,listFirst(fakeClient->reply));
/* Clean up, ready for the next command */
for (j = 0; j < argc; j++) decrRefCount(argv[j]);
zfree(argv);
/* Handle swapping while loading big datasets when VM is on */
loadedkeys++;
if (server.vm_enabled && (loadedkeys % 5000) == 0) {
while (zmalloc_used_memory() > server.vm_max_memory) {
if (vmSwapOneObjectBlocking() == REDIS_ERR) break;
}
}
}
fclose(fp);
freeFakeClient(fakeClient);
return REDIS_OK;
}