serverCron() frequency is now a runtime parameter (was REDIS_HZ).

REDIS_HZ is the frequency our serverCron() function is called with.
A more frequent call to this function results into less latency when the
server is trying to handle very expansive background operations like
mass expires of a lot of keys at the same time.

Redis 2.4 used to have an HZ of 10. This was good enough with almost
every setup, but the incremental key expiration algorithm was working a
bit better under *extreme* pressure when HZ was set to 100 for Redis
2.6.

However for most users a latency spike of 30 milliseconds when million
of keys are expiring at the same time is acceptable, on the other hand a
default HZ of 100 in Redis 2.6 was causing idle instances to use some
CPU time compared to Redis 2.4. The CPU usage was in the order of 0.3%
for an idle instance, however this is a shame as more energy is consumed
by the server, if not important resources.

This commit introduces HZ as a runtime parameter, that can be queried by
INFO or CONFIG GET, and can be modified with CONFIG SET. At the same
time the default frequency is set back to 10.

In this way we default to a sane value of 10, but allows users to
easily switch to values up to 500 for near real-time applications if
needed and if they are willing to pay this small CPU usage penalty.

redis.conf

@@ -529,6 +529,23 @@ client-output-buffer-limit normal 0 0 0
 client-output-buffer-limit slave 256mb 64mb 60
 client-output-buffer-limit pubsub 32mb 8mb 60
 
+# Redis calls an internal function to perform many background tasks, like
+# closing connections of clients in timeot, purging expired keys that are
+# never requested, and so forth.
+#
+# Not all tasks are perforemd with the same frequency, but Redis checks for
+# tasks to perform accordingly to the specified "hz" value.
+#
+# By default "hz" is set to 10. Raising the value will use more CPU when
+# Redis is idle, but at the same time will make Redis more responsive when
+# there are many keys expiring at the same time, and timeouts may be
+# handled with more precision.
+#
+# The range is between 1 and 500, however a value over 100 is usually not
+# a good idea. Most users should use the default of 10 and raise this up to
+# 100 only in environments where very low latency is required.
+hz 10
+
 ################################## INCLUDES ###################################
 
 # Include one or more other config files here.  This is useful if you

src/config.c

@@ -256,6 +256,10 @@ void loadServerConfigFromString(char *config) {
             if ((server.daemonize = yesnotoi(argv[1])) == -1) {
                 err = "argument must be 'yes' or 'no'"; goto loaderr;
             }
+        } else if (!strcasecmp(argv[0],"hz") && argc == 2) {
+            server.hz = atoi(argv[1]);
+            if (server.hz < REDIS_MIN_HZ) server.hz = REDIS_MIN_HZ;
+            if (server.hz > REDIS_MAX_HZ) server.hz = REDIS_MAX_HZ;
         } else if (!strcasecmp(argv[0],"appendonly") && argc == 2) {
             int yes;
 
@@ -475,6 +479,12 @@ void configSetCommand(redisClient *c) {
             }
             freeMemoryIfNeeded();
         }
+    } else if (!strcasecmp(c->argv[2]->ptr,"hz")) {
+        if (getLongLongFromObject(o,&ll) == REDIS_ERR ||
+            ll < 0) goto badfmt;
+        server.hz = (int) ll;
+        if (server.hz < REDIS_MIN_HZ) server.hz = REDIS_MIN_HZ;
+        if (server.hz > REDIS_MAX_HZ) server.hz = REDIS_MAX_HZ;
     } else if (!strcasecmp(c->argv[2]->ptr,"maxmemory-policy")) {
         if (!strcasecmp(o->ptr,"volatile-lru")) {
             server.maxmemory_policy = REDIS_MAXMEMORY_VOLATILE_LRU;
@@ -791,6 +801,7 @@ void configGetCommand(redisClient *c) {
     config_get_numerical_field("maxclients",server.maxclients);
     config_get_numerical_field("watchdog-period",server.watchdog_period);
     config_get_numerical_field("slave-priority",server.slave_priority);
+    config_get_numerical_field("hz",server.hz);
 
     /* Bool (yes/no) values */
     config_get_bool_field("no-appendfsync-on-rewrite",

src/debug.c

@@ -891,7 +891,7 @@ void enableWatchdog(int period) {
     /* If the configured period is smaller than twice the timer period, it is
      * too short for the software watchdog to work reliably. Fix it now
      * if needed. */
-    min_period = (1000/REDIS_HZ)*2;
+    min_period = (1000/server.hz)*2;
     if (period < min_period) period = min_period;
     watchdogScheduleSignal(period); /* Adjust the current timer. */
     server.watchdog_period = period;

src/redis.c

@@ -627,9 +627,9 @@ void activeExpireCycle(void) {
 
     /* We can use at max REDIS_EXPIRELOOKUPS_TIME_PERC percentage of CPU time
      * per iteration. Since this function gets called with a frequency of
-     * REDIS_HZ times per second, the following is the max amount of
+     * server.hz times per second, the following is the max amount of
      * microseconds we can spend in this function. */
-    timelimit = 1000000*REDIS_EXPIRELOOKUPS_TIME_PERC/REDIS_HZ/100;
+    timelimit = 1000000*REDIS_EXPIRELOOKUPS_TIME_PERC/server.hz/100;
     if (timelimit <= 0) timelimit = 1;
 
     for (j = 0; j < server.dbnum; j++) {
@@ -762,13 +762,13 @@ int clientsCronResizeQueryBuffer(redisClient *c) {
 }
 
 void clientsCron(void) {
-    /* Make sure to process at least 1/(REDIS_HZ*10) of clients per call.
-     * Since this function is called REDIS_HZ times per second we are sure that
+    /* Make sure to process at least 1/(server.hz*10) of clients per call.
+     * Since this function is called server.hz times per second we are sure that
      * in the worst case we process all the clients in 10 seconds.
      * In normal conditions (a reasonable number of clients) we process
      * all the clients in a shorter time. */
     int numclients = listLength(server.clients);
-    int iterations = numclients/(REDIS_HZ*10);
+    int iterations = numclients/(server.hz*10);
 
     if (iterations < 50)
         iterations = (numclients < 50) ? numclients : 50;
@@ -790,7 +790,7 @@ void clientsCron(void) {
     }
 }
 
-/* This is our timer interrupt, called REDIS_HZ times per second.
+/* This is our timer interrupt, called server.hz times per second.
  * Here is where we do a number of things that need to be done asynchronously.
  * For instance:
  *
@@ -804,7 +804,7 @@ void clientsCron(void) {
  * - Replication reconnection.
  * - Many more...
  *
- * Everything directly called here will be called REDIS_HZ times per second,
+ * Everything directly called here will be called server.hz times per second,
  * so in order to throttle execution of things we want to do less frequently
  * a macro is used: run_with_period(milliseconds) { .... }
  */
@@ -976,7 +976,7 @@ int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) {
     }
 
     server.cronloops++;
-    return 1000/REDIS_HZ;
+    return 1000/server.hz;
 }
 
 /* This function gets called every time Redis is entering the
@@ -1082,6 +1082,7 @@ void createSharedObjects(void) {
 
 void initServerConfig() {
     getRandomHexChars(server.runid,REDIS_RUN_ID_SIZE);
+    server.hz = REDIS_DEFAULT_HZ;
     server.runid[REDIS_RUN_ID_SIZE] = '\0';
     server.arch_bits = (sizeof(long) == 8) ? 64 : 32;
     server.port = REDIS_SERVERPORT;
@@ -1879,6 +1880,7 @@ sds genRedisInfoString(char *section) {
             "tcp_port:%d\r\n"
             "uptime_in_seconds:%ld\r\n"
             "uptime_in_days:%ld\r\n"
+            "hz:%d\r\n"
             "lru_clock:%ld\r\n",
             REDIS_VERSION,
             redisGitSHA1(),
@@ -1898,6 +1900,7 @@ sds genRedisInfoString(char *section) {
             server.port,
             uptime,
             uptime/(3600*24),
+            server.hz,
             (unsigned long) server.lruclock);
     }
 

src/redis.h

@@ -67,7 +67,9 @@
 #define REDIS_ERR               -1
 
 /* Static server configuration */
-#define REDIS_HZ                100     /* Time interrupt calls/sec. */
+#define REDIS_DEFAULT_HZ        10      /* Time interrupt calls/sec. */
+#define REDIS_MIN_HZ            1
+#define REDIS_MAX_HZ            500 
 #define REDIS_SERVERPORT        6379    /* TCP port */
 #define REDIS_MAXIDLETIME       0       /* default client timeout: infinite */
 #define REDIS_DEFAULT_DBNUM     16
@@ -291,8 +293,8 @@
 
 /* Using the following macro you can run code inside serverCron() with the
  * specified period, specified in milliseconds.
- * The actual resolution depends on REDIS_HZ. */
-#define run_with_period(_ms_) if (!(server.cronloops%((_ms_)/(1000/REDIS_HZ))))
+ * The actual resolution depends on server.hz. */
+#define run_with_period(_ms_) if ((_ms_ <= 1000/server.hz) || !(server.cronloops%((_ms_)/(1000/server.hz))))
 
 /* We can print the stacktrace, so our assert is defined this way: */
 #define redisAssertWithInfo(_c,_o,_e) ((_e)?(void)0 : (_redisAssertWithInfo(_c,_o,#_e,__FILE__,__LINE__),_exit(1)))
@@ -493,6 +495,7 @@ typedef struct redisOpArray {
 
 struct redisServer {
     /* General */
+    int hz;                     /* serverCron() calls frequency in hertz */
     redisDb *db;
     dict *commands;             /* Command table hash table */
     aeEventLoop *el;

src/replication.c

@@ -778,7 +778,7 @@ void replicationCron(void) {
      * So slaves can implement an explicit timeout to masters, and will
      * be able to detect a link disconnection even if the TCP connection
      * will not actually go down. */
-    if (!(server.cronloops % (server.repl_ping_slave_period * REDIS_HZ))) {
+    if (!(server.cronloops % (server.repl_ping_slave_period * server.hz))) {
         listIter li;
         listNode *ln;