Fix for slaves chains. Force resync of slaves (simply disconnecting them) when SLAVE...

[redis.git] / src / rdb.c

diff --git a/src/rdb.c b/src/rdb.c
index a401fe89aa0c7c456d7e244a7ed7b72ccb9a3168..1e23fa70cf392b399806ce9a7086af65114dd7c2 100644 (file)
--- a/src/rdb.c
+++ b/src/rdb.c
@@ -1,5 +1,6 @@
 #include "redis.h"
 #include "lzf.h"    /* LZF compression library */
 #include "redis.h"
 #include "lzf.h"    /* LZF compression library */

+#include "zipmap.h"

 
 #include <math.h>
 #include <sys/types.h>
 
 #include <math.h>
 #include <sys/types.h>


@@ -253,7 +254,7 @@ int rdbSaveRawString(rio *rdb, unsigned char *s, size_t len) {
 
     /* Try LZF compression - under 20 bytes it's unable to compress even
      * aaaaaaaaaaaaaaaaaa so skip it */
 
     /* Try LZF compression - under 20 bytes it's unable to compress even
      * aaaaaaaaaaaaaaaaaa so skip it */

-    if (server.rdbcompression && len > 20) {
+    if (server.rdb_compression && len > 20) {

         n = rdbSaveLzfStringObject(rdb,s,len);
         if (n == -1) return -1;
         if (n > 0) return n;
         n = rdbSaveLzfStringObject(rdb,s,len);
         if (n == -1) return -1;
         if (n > 0) return n;


@@ -424,8 +425,8 @@ int rdbSaveObjectType(rio *rdb, robj *o) {
         else
             redisPanic("Unknown sorted set encoding");
     case REDIS_HASH:
         else
             redisPanic("Unknown sorted set encoding");
     case REDIS_HASH:

-        if (o->encoding == REDIS_ENCODING_ZIPMAP)
-            return rdbSaveType(rdb,REDIS_RDB_TYPE_HASH_ZIPMAP);
+        if (o->encoding == REDIS_ENCODING_ZIPLIST)
+            return rdbSaveType(rdb,REDIS_RDB_TYPE_HASH_ZIPLIST);

         else if (o->encoding == REDIS_ENCODING_HT)
             return rdbSaveType(rdb,REDIS_RDB_TYPE_HASH);
         else
         else if (o->encoding == REDIS_ENCODING_HT)
             return rdbSaveType(rdb,REDIS_RDB_TYPE_HASH);
         else


@@ -530,12 +531,13 @@ int rdbSaveObject(rio *rdb, robj *o) {
         }
     } else if (o->type == REDIS_HASH) {
         /* Save a hash value */
         }
     } else if (o->type == REDIS_HASH) {
         /* Save a hash value */

-        if (o->encoding == REDIS_ENCODING_ZIPMAP) {
-            size_t l = zipmapBlobLen((unsigned char*)o->ptr);
+        if (o->encoding == REDIS_ENCODING_ZIPLIST) {
+            size_t l = ziplistBlobLen((unsigned char*)o->ptr);

 
             if ((n = rdbSaveRawString(rdb,o->ptr,l)) == -1) return -1;
             nwritten += n;
 
             if ((n = rdbSaveRawString(rdb,o->ptr,l)) == -1) return -1;
             nwritten += n;

-        } else {
+
+        } else if (o->encoding == REDIS_ENCODING_HT) {

             dictIterator *di = dictGetIterator(o->ptr);
             dictEntry *de;
 
             dictIterator *di = dictGetIterator(o->ptr);
             dictEntry *de;
 


@@ -552,7 +554,11 @@ int rdbSaveObject(rio *rdb, robj *o) {
                 nwritten += n;
             }
             dictReleaseIterator(di);
                 nwritten += n;
             }
             dictReleaseIterator(di);

+
+        } else {
+            redisPanic("Unknown hash encoding");

         }
         }

+

     } else {
         redisPanic("Unknown object type");
     }
     } else {
         redisPanic("Unknown object type");
     }


@@ -610,7 +616,7 @@ int rdbSave(char *filename) {
     }
 
     rioInitWithFile(&rdb,fp);
     }
 
     rioInitWithFile(&rdb,fp);

-    if (rdbWriteRaw(&rdb,"REDIS0003",9) == -1) goto werr;
+    if (rdbWriteRaw(&rdb,"REDIS0004",9) == -1) goto werr;

 
     for (j = 0; j < server.dbnum; j++) {
         redisDb *db = server.db+j;
 
     for (j = 0; j < server.dbnum; j++) {
         redisDb *db = server.db+j;


@@ -656,6 +662,7 @@ int rdbSave(char *filename) {
     redisLog(REDIS_NOTICE,"DB saved on disk");
     server.dirty = 0;
     server.lastsave = time(NULL);
     redisLog(REDIS_NOTICE,"DB saved on disk");
     server.dirty = 0;
     server.lastsave = time(NULL);

+    server.lastbgsave_status = REDIS_OK;

     return REDIS_OK;
 
 werr:
     return REDIS_OK;
 
 werr:


@@ -670,7 +677,7 @@ int rdbSaveBackground(char *filename) {
     pid_t childpid;
     long long start;
 
     pid_t childpid;
     long long start;
 

-    if (server.bgsavechildpid != -1) return REDIS_ERR;
+    if (server.rdb_child_pid != -1) return REDIS_ERR;

 
     server.dirty_before_bgsave = server.dirty;
 
 
     server.dirty_before_bgsave = server.dirty;
 


@@ -692,7 +699,7 @@ int rdbSaveBackground(char *filename) {
             return REDIS_ERR;
         }
         redisLog(REDIS_NOTICE,"Background saving started by pid %d",childpid);
             return REDIS_ERR;
         }
         redisLog(REDIS_NOTICE,"Background saving started by pid %d",childpid);

-        server.bgsavechildpid = childpid;
+        server.rdb_child_pid = childpid;

         updateDictResizePolicy();
         return REDIS_OK;
     }
         updateDictResizePolicy();
         return REDIS_OK;
     }


@@ -824,55 +831,74 @@ robj *rdbLoadObject(int rdbtype, rio *rdb) {
             maxelelen <= server.zset_max_ziplist_value)
                 zsetConvert(o,REDIS_ENCODING_ZIPLIST);
     } else if (rdbtype == REDIS_RDB_TYPE_HASH) {
             maxelelen <= server.zset_max_ziplist_value)
                 zsetConvert(o,REDIS_ENCODING_ZIPLIST);
     } else if (rdbtype == REDIS_RDB_TYPE_HASH) {

-        size_t hashlen;
+        size_t len;
+        int ret;
+
+        len = rdbLoadLen(rdb, NULL);
+        if (len == REDIS_RDB_LENERR) return NULL;

 
 

-        if ((hashlen = rdbLoadLen(rdb,NULL)) == REDIS_RDB_LENERR) return NULL;

         o = createHashObject();
         o = createHashObject();

+

         /* Too many entries? Use an hash table. */
         /* Too many entries? Use an hash table. */

-        if (hashlen > server.hash_max_zipmap_entries)
-            convertToRealHash(o);
-        /* Load every key/value, then set it into the zipmap or hash
-         * table, as needed. */
-        while(hashlen--) {
-            robj *key, *val;
-
-            if ((key = rdbLoadEncodedStringObject(rdb)) == NULL) return NULL;
-            if ((val = rdbLoadEncodedStringObject(rdb)) == NULL) return NULL;
-            /* If we are using a zipmap and there are too big values
-             * the object is converted to real hash table encoding. */
-            if (o->encoding != REDIS_ENCODING_HT &&
-               ((key->encoding == REDIS_ENCODING_RAW &&
-                sdslen(key->ptr) > server.hash_max_zipmap_value) ||
-                (val->encoding == REDIS_ENCODING_RAW &&
-                sdslen(val->ptr) > server.hash_max_zipmap_value)))
+        if (len > server.hash_max_ziplist_entries)
+            hashTypeConvert(o, REDIS_ENCODING_HT);
+
+        /* Load every field and value into the ziplist */
+        while (o->encoding == REDIS_ENCODING_ZIPLIST && len > 0) {
+            robj *field, *value;
+
+            len--;
+            /* Load raw strings */
+            field = rdbLoadStringObject(rdb);
+            if (field == NULL) return NULL;
+            redisAssert(field->encoding == REDIS_ENCODING_RAW);
+            value = rdbLoadStringObject(rdb);
+            if (value == NULL) return NULL;
+            redisAssert(field->encoding == REDIS_ENCODING_RAW);
+
+            /* Add pair to ziplist */
+            o->ptr = ziplistPush(o->ptr, field->ptr, sdslen(field->ptr), ZIPLIST_TAIL);
+            o->ptr = ziplistPush(o->ptr, value->ptr, sdslen(value->ptr), ZIPLIST_TAIL);
+            /* Convert to hash table if size threshold is exceeded */
+            if (sdslen(field->ptr) > server.hash_max_ziplist_value ||
+                sdslen(value->ptr) > server.hash_max_ziplist_value)

             {
             {

-                    convertToRealHash(o);
+                decrRefCount(field);
+                decrRefCount(value);
+                hashTypeConvert(o, REDIS_ENCODING_HT);
+                break;

             }
             }

+            decrRefCount(field);
+            decrRefCount(value);
+        }

 
 

-            if (o->encoding == REDIS_ENCODING_ZIPMAP) {
-                unsigned char *zm = o->ptr;
-                robj *deckey, *decval;
-
-                /* We need raw string objects to add them to the zipmap */
-                deckey = getDecodedObject(key);
-                decval = getDecodedObject(val);
-                zm = zipmapSet(zm,deckey->ptr,sdslen(deckey->ptr),
-                                  decval->ptr,sdslen(decval->ptr),NULL);
-                o->ptr = zm;
-                decrRefCount(deckey);
-                decrRefCount(decval);
-                decrRefCount(key);
-                decrRefCount(val);
-            } else {
-                key = tryObjectEncoding(key);
-                val = tryObjectEncoding(val);
-                dictAdd((dict*)o->ptr,key,val);
-            }
+        /* Load remaining fields and values into the hash table */
+        while (o->encoding == REDIS_ENCODING_HT && len > 0) {
+            robj *field, *value;
+
+            len--;
+            /* Load encoded strings */
+            field = rdbLoadEncodedStringObject(rdb);
+            if (field == NULL) return NULL;
+            value = rdbLoadEncodedStringObject(rdb);
+            if (value == NULL) return NULL;
+
+            field = tryObjectEncoding(field);
+            value = tryObjectEncoding(value);
+
+            /* Add pair to hash table */
+            ret = dictAdd((dict*)o->ptr, field, value);
+            redisAssert(ret == REDIS_OK);

         }
         }

+
+        /* All pairs should be read by now */
+        redisAssert(len == 0);
+

     } else if (rdbtype == REDIS_RDB_TYPE_HASH_ZIPMAP  ||
                rdbtype == REDIS_RDB_TYPE_LIST_ZIPLIST ||
                rdbtype == REDIS_RDB_TYPE_SET_INTSET   ||
     } else if (rdbtype == REDIS_RDB_TYPE_HASH_ZIPMAP  ||
                rdbtype == REDIS_RDB_TYPE_LIST_ZIPLIST ||
                rdbtype == REDIS_RDB_TYPE_SET_INTSET   ||

-               rdbtype == REDIS_RDB_TYPE_ZSET_ZIPLIST)
+               rdbtype == REDIS_RDB_TYPE_ZSET_ZIPLIST ||
+               rdbtype == REDIS_RDB_TYPE_HASH_ZIPLIST)

     {
         robj *aux = rdbLoadStringObject(rdb);
 
     {
         robj *aux = rdbLoadStringObject(rdb);
 


@@ -890,10 +916,33 @@ robj *rdbLoadObject(int rdbtype, rio *rdb) {
          * converted. */
         switch(rdbtype) {
             case REDIS_RDB_TYPE_HASH_ZIPMAP:
          * converted. */
         switch(rdbtype) {
             case REDIS_RDB_TYPE_HASH_ZIPMAP:

-                o->type = REDIS_HASH;
-                o->encoding = REDIS_ENCODING_ZIPMAP;
-                if (zipmapLen(o->ptr) > server.hash_max_zipmap_entries)
-                    convertToRealHash(o);
+                /* Convert to ziplist encoded hash. This must be deprecated
+                 * when loading dumps created by Redis 2.4 gets deprecated. */
+                {
+                    unsigned char *zl = ziplistNew();
+                    unsigned char *zi = zipmapRewind(o->ptr);
+                    unsigned char *fstr, *vstr;
+                    unsigned int flen, vlen;
+                    unsigned int maxlen = 0;
+
+                    while ((zi = zipmapNext(zi, &fstr, &flen, &vstr, &vlen)) != NULL) {
+                        if (flen > maxlen) maxlen = flen;
+                        if (vlen > maxlen) maxlen = vlen;
+                        zl = ziplistPush(zl, fstr, flen, ZIPLIST_TAIL);
+                        zl = ziplistPush(zl, vstr, vlen, ZIPLIST_TAIL);
+                    }
+
+                    zfree(o->ptr);
+                    o->ptr = zl;
+                    o->type = REDIS_HASH;
+                    o->encoding = REDIS_ENCODING_ZIPLIST;
+
+                    if (hashTypeLength(o) > server.hash_max_ziplist_entries ||
+                        maxlen > server.hash_max_ziplist_value)
+                    {
+                        hashTypeConvert(o, REDIS_ENCODING_HT);
+                    }
+                }

                 break;
             case REDIS_RDB_TYPE_LIST_ZIPLIST:
                 o->type = REDIS_LIST;
                 break;
             case REDIS_RDB_TYPE_LIST_ZIPLIST:
                 o->type = REDIS_LIST;


@@ -913,6 +962,12 @@ robj *rdbLoadObject(int rdbtype, rio *rdb) {
                 if (zsetLength(o) > server.zset_max_ziplist_entries)
                     zsetConvert(o,REDIS_ENCODING_SKIPLIST);
                 break;
                 if (zsetLength(o) > server.zset_max_ziplist_entries)
                     zsetConvert(o,REDIS_ENCODING_SKIPLIST);
                 break;

+            case REDIS_RDB_TYPE_HASH_ZIPLIST:
+                o->type = REDIS_HASH;
+                o->encoding = REDIS_ENCODING_ZIPLIST;
+                if (hashTypeLength(o) > server.hash_max_ziplist_entries)
+                    hashTypeConvert(o, REDIS_ENCODING_HT);
+                break;

             default:
                 redisPanic("Unknown encoding");
                 break;
             default:
                 redisPanic("Unknown encoding");
                 break;


@@ -973,7 +1028,7 @@ int rdbLoad(char *filename) {
         return REDIS_ERR;
     }
     rdbver = atoi(buf+5);
         return REDIS_ERR;
     }
     rdbver = atoi(buf+5);

-    if (rdbver < 1 || rdbver > 3) {
+    if (rdbver < 1 || rdbver > 4) {

         fclose(fp);
         redisLog(REDIS_WARNING,"Can't handle RDB format version %d",rdbver);
         errno = EINVAL;
         fclose(fp);
         redisLog(REDIS_WARNING,"Can't handle RDB format version %d",rdbver);
         errno = EINVAL;


@@ -1026,8 +1081,12 @@ int rdbLoad(char *filename) {
         if ((key = rdbLoadStringObject(&rdb)) == NULL) goto eoferr;
         /* Read value */
         if ((val = rdbLoadObject(type,&rdb)) == NULL) goto eoferr;
         if ((key = rdbLoadStringObject(&rdb)) == NULL) goto eoferr;
         /* Read value */
         if ((val = rdbLoadObject(type,&rdb)) == NULL) goto eoferr;

-        /* Check if the key already expired */
-        if (expiretime != -1 && expiretime < now) {
+        /* Check if the key already expired. This function is used when loading
+         * an RDB file from disk, either at startup, or when an RDB was
+         * received from the master. In the latter case, the master is
+         * responsible for key expiry. If we would expire keys here, the
+         * snapshot taken by the master may not be reflected on the slave. */
+        if (server.masterhost == NULL && expiretime != -1 && expiretime < now) {

             decrRefCount(key);
             decrRefCount(val);
             continue;
             decrRefCount(key);
             decrRefCount(val);
             continue;


@@ -1057,25 +1116,28 @@ void backgroundSaveDoneHandler(int exitcode, int bysignal) {
             "Background saving terminated with success");
         server.dirty = server.dirty - server.dirty_before_bgsave;
         server.lastsave = time(NULL);
             "Background saving terminated with success");
         server.dirty = server.dirty - server.dirty_before_bgsave;
         server.lastsave = time(NULL);

+        server.lastbgsave_status = REDIS_OK;

     } else if (!bysignal && exitcode != 0) {
         redisLog(REDIS_WARNING, "Background saving error");
     } else if (!bysignal && exitcode != 0) {
         redisLog(REDIS_WARNING, "Background saving error");

+        server.lastbgsave_status = REDIS_ERR;

     } else {
         redisLog(REDIS_WARNING,
             "Background saving terminated by signal %d", bysignal);
     } else {
         redisLog(REDIS_WARNING,
             "Background saving terminated by signal %d", bysignal);

-        rdbRemoveTempFile(server.bgsavechildpid);
+        rdbRemoveTempFile(server.rdb_child_pid);
+        server.lastbgsave_status = REDIS_ERR;

     }
     }

-    server.bgsavechildpid = -1;
+    server.rdb_child_pid = -1;

     /* Possibly there are slaves waiting for a BGSAVE in order to be served
      * (the first stage of SYNC is a bulk transfer of dump.rdb) */
     updateSlavesWaitingBgsave(exitcode == 0 ? REDIS_OK : REDIS_ERR);
 }
 
 void saveCommand(redisClient *c) {
     /* Possibly there are slaves waiting for a BGSAVE in order to be served
      * (the first stage of SYNC is a bulk transfer of dump.rdb) */
     updateSlavesWaitingBgsave(exitcode == 0 ? REDIS_OK : REDIS_ERR);
 }
 
 void saveCommand(redisClient *c) {

-    if (server.bgsavechildpid != -1) {
+    if (server.rdb_child_pid != -1) {

         addReplyError(c,"Background save already in progress");
         return;
     }
         addReplyError(c,"Background save already in progress");
         return;
     }

-    if (rdbSave(server.dbfilename) == REDIS_OK) {
+    if (rdbSave(server.rdb_filename) == REDIS_OK) {

         addReply(c,shared.ok);
     } else {
         addReply(c,shared.err);
         addReply(c,shared.ok);
     } else {
         addReply(c,shared.err);


@@ -1083,11 +1145,11 @@ void saveCommand(redisClient *c) {
 }
 
 void bgsaveCommand(redisClient *c) {
 }
 
 void bgsaveCommand(redisClient *c) {

-    if (server.bgsavechildpid != -1) {
+    if (server.rdb_child_pid != -1) {

         addReplyError(c,"Background save already in progress");
     } else if (server.aof_child_pid != -1) {
         addReplyError(c,"Can't BGSAVE while AOF log rewriting is in progress");
         addReplyError(c,"Background save already in progress");
     } else if (server.aof_child_pid != -1) {
         addReplyError(c,"Can't BGSAVE while AOF log rewriting is in progress");

-    } else if (rdbSaveBackground(server.dbfilename) == REDIS_OK) {
+    } else if (rdbSaveBackground(server.rdb_filename) == REDIS_OK) {

         addReplyStatus(c,"Background saving started");
     } else {
         addReply(c,shared.err);
         addReplyStatus(c,"Background saving started");
     } else {
         addReply(c,shared.err);