bio.o added to redis-server objects

[redis.git] / src / aof.c

diff --git a/src/aof.c b/src/aof.c
index 78387cd366c8112d309e27c85cbde1b40e34e5ee..45dbc0db456dd81e6ee964f94559b04d12adb775 100644 (file)
--- a/src/aof.c
+++ b/src/aof.c
@@ -60,7 +60,6 @@ int startAppendOnly(void) {
  * buffer and write it on disk using this function just before entering
  * the event loop again. */
 void flushAppendOnlyFile(void) {
  * buffer and write it on disk using this function just before entering
  * the event loop again. */
 void flushAppendOnlyFile(void) {

-    time_t now;

     ssize_t nwritten;
 
     if (sdslen(server.aofbuf) == 0) return;
     ssize_t nwritten;
 
     if (sdslen(server.aofbuf) == 0) return;


@@ -70,51 +69,70 @@ void flushAppendOnlyFile(void) {
      * While this will save us against the server being killed I don't think
      * there is much to do about the whole server stopping for power problems
      * or alike */
      * While this will save us against the server being killed I don't think
      * there is much to do about the whole server stopping for power problems
      * or alike */

-     nwritten = write(server.appendfd,server.aofbuf,sdslen(server.aofbuf));
-     if (nwritten != (signed)sdslen(server.aofbuf)) {
+    nwritten = write(server.appendfd,server.aofbuf,sdslen(server.aofbuf));
+    if (nwritten != (signed)sdslen(server.aofbuf)) {

         /* Ooops, we are in troubles. The best thing to do for now is
          * aborting instead of giving the illusion that everything is
          * working as expected. */
         /* Ooops, we are in troubles. The best thing to do for now is
          * aborting instead of giving the illusion that everything is
          * working as expected. */

-         if (nwritten == -1) {
+        if (nwritten == -1) {

             redisLog(REDIS_WARNING,"Exiting on error writing to the append-only file: %s",strerror(errno));
             redisLog(REDIS_WARNING,"Exiting on error writing to the append-only file: %s",strerror(errno));

-         } else {
+        } else {

             redisLog(REDIS_WARNING,"Exiting on short write while writing to the append-only file: %s",strerror(errno));
             redisLog(REDIS_WARNING,"Exiting on short write while writing to the append-only file: %s",strerror(errno));

-         }
-         exit(1);
+        }
+        exit(1);

     }
     }

-    sdsfree(server.aofbuf);
-    server.aofbuf = sdsempty();

     server.appendonly_current_size += nwritten;
 
     server.appendonly_current_size += nwritten;
 

-    /* Don't Fsync if no-appendfsync-on-rewrite is set to yes and we have
-     * childs performing heavy I/O on disk. */
+    /* Re-use AOF buffer when it is small enough. The maximum comes from the
+     * arena size of 4k minus some overhead (but is otherwise arbitrary). */
+    if ((sdslen(server.aofbuf)+sdsavail(server.aofbuf)) < 4000) {
+        sdsclear(server.aofbuf);
+    } else {
+        sdsfree(server.aofbuf);
+        server.aofbuf = sdsempty();
+    }
+
+    /* Don't fsync if no-appendfsync-on-rewrite is set to yes and there are
+     * children doing I/O in the background. */

     if (server.no_appendfsync_on_rewrite &&
         (server.bgrewritechildpid != -1 || server.bgsavechildpid != -1))
             return;
     if (server.no_appendfsync_on_rewrite &&
         (server.bgrewritechildpid != -1 || server.bgsavechildpid != -1))
             return;

-    /* Fsync if needed */
-    now = time(NULL);
+
+    /* Perform the fsync if needed. */

     if (server.appendfsync == APPENDFSYNC_ALWAYS ||
         (server.appendfsync == APPENDFSYNC_EVERYSEC &&
     if (server.appendfsync == APPENDFSYNC_ALWAYS ||
         (server.appendfsync == APPENDFSYNC_EVERYSEC &&

-         now-server.lastfsync > 1))
+         server.unixtime > server.lastfsync))

     {
         /* aof_fsync is defined as fdatasync() for Linux in order to avoid
          * flushing metadata. */
         aof_fsync(server.appendfd); /* Let's try to get this data on the disk */
     {
         /* aof_fsync is defined as fdatasync() for Linux in order to avoid
          * flushing metadata. */
         aof_fsync(server.appendfd); /* Let's try to get this data on the disk */

-        server.lastfsync = now;
+        server.lastfsync = server.unixtime;

     }
 }
 
     }
 }
 

-sds catAppendOnlyGenericCommand(sds buf, int argc, robj **argv) {
-    int j;
-    buf = sdscatprintf(buf,"*%d\r\n",argc);
+sds catAppendOnlyGenericCommand(sds dst, int argc, robj **argv) {
+    char buf[32];
+    int len, j;
+    robj *o;
+
+    buf[0] = '*';
+    len = 1+ll2string(buf+1,sizeof(buf)-1,argc);
+    buf[len++] = '\r';
+    buf[len++] = '\n';
+    dst = sdscatlen(dst,buf,len);
+

     for (j = 0; j < argc; j++) {
     for (j = 0; j < argc; j++) {

-        robj *o = getDecodedObject(argv[j]);
-        buf = sdscatprintf(buf,"$%lu\r\n",(unsigned long)sdslen(o->ptr));
-        buf = sdscatlen(buf,o->ptr,sdslen(o->ptr));
-        buf = sdscatlen(buf,"\r\n",2);
+        o = getDecodedObject(argv[j]);
+        buf[0] = '$';
+        len = 1+ll2string(buf+1,sizeof(buf)-1,sdslen(o->ptr));
+        buf[len++] = '\r';
+        buf[len++] = '\n';
+        dst = sdscatlen(dst,buf,len);
+        dst = sdscatlen(dst,o->ptr,sdslen(o->ptr));
+        dst = sdscatlen(dst,"\r\n",2);

         decrRefCount(o);
     }
         decrRefCount(o);
     }

-    return buf;
+    return dst;

 }
 
 sds catAppendOnlyExpireAtCommand(sds buf, robj *key, robj *seconds) {
 }
 
 sds catAppendOnlyExpireAtCommand(sds buf, robj *key, robj *seconds) {


@@ -287,6 +305,8 @@ int loadAppendOnlyFile(char *filename) {
 
         /* The fake client should not have a reply */
         redisAssert(fakeClient->bufpos == 0 && listLength(fakeClient->reply) == 0);
 
         /* 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. */
 
         /* Clean up. Command code may have changed argv/argc so we use the
          * argv/argc of the client instead of the local variables. */


@@ -304,6 +324,7 @@ int loadAppendOnlyFile(char *filename) {
     server.appendonly = appendonly;
     stopLoading();
     aofUpdateCurrentSize();
     server.appendonly = appendonly;
     stopLoading();
     aofUpdateCurrentSize();

+    server.auto_aofrewrite_base_size = server.appendonly_current_size;

     return REDIS_OK;
 
 readerr:
     return REDIS_OK;
 
 readerr:


@@ -341,7 +362,7 @@ int rewriteAppendOnlyFile(char *filename) {
         redisDb *db = server.db+j;
         dict *d = db->dict;
         if (dictSize(d) == 0) continue;
         redisDb *db = server.db+j;
         dict *d = db->dict;
         if (dictSize(d) == 0) continue;

-        di = dictGetIterator(d);
+        di = dictGetSafeIterator(d);

         if (!di) {
             fclose(fp);
             return REDIS_ERR;
         if (!di) {
             fclose(fp);
             return REDIS_ERR;


@@ -573,10 +594,6 @@ int rewriteAppendOnlyFileBackground(void) {
     long long start;
 
     if (server.bgrewritechildpid != -1) return REDIS_ERR;
     long long start;
 
     if (server.bgrewritechildpid != -1) return REDIS_ERR;

-    if (server.ds_enabled != 0) {
-        redisLog(REDIS_WARNING,"BGREWRITEAOF called with diskstore enabled: AOF is not supported when diskstore is enabled. Operation not performed.");
-        return REDIS_ERR;
-    }

     start = ustime();
     if ((childpid = fork()) == 0) {
         char tmpfile[256];
     start = ustime();
     if ((childpid = fork()) == 0) {
         char tmpfile[256];


@@ -618,7 +635,7 @@ void bgrewriteaofCommand(redisClient *c) {
         addReplyError(c,"Background append only file rewriting already in progress");
     } else if (server.bgsavechildpid != -1) {
         server.aofrewrite_scheduled = 1;
         addReplyError(c,"Background append only file rewriting already in progress");
     } else if (server.bgsavechildpid != -1) {
         server.aofrewrite_scheduled = 1;

-        addReplyStatus(c,"Background append only file rewriting started");
+        addReplyStatus(c,"Background append only file rewriting scheduled");

     } else if (rewriteAppendOnlyFileBackground() == REDIS_OK) {
         addReplyStatus(c,"Background append only file rewriting started");
     } else {
     } else if (rewriteAppendOnlyFileBackground() == REDIS_OK) {
         addReplyStatus(c,"Background append only file rewriting started");
     } else {


@@ -689,6 +706,7 @@ void backgroundRewriteDoneHandler(int exitcode, int bysignal) {
             server.appendseldb = -1; /* Make sure it will issue SELECT */
             redisLog(REDIS_NOTICE,"The new append only file was selected for future appends.");
             aofUpdateCurrentSize();
             server.appendseldb = -1; /* Make sure it will issue SELECT */
             redisLog(REDIS_NOTICE,"The new append only file was selected for future appends.");
             aofUpdateCurrentSize();

+            server.auto_aofrewrite_base_size = server.appendonly_current_size;

         } else {
             /* If append only is disabled we just generate a dump in this
              * format. Why not? */
         } else {
             /* If append only is disabled we just generate a dump in this
              * format. Why not? */