max cluster cron failed connections

Signed-off-by: Sarthak Aggarwal <sarthagg@amazon.com>

Author: sarthakaggarwal97

src/cluster_legacy.c

@@ -1632,6 +1632,7 @@ clusterNode *createClusterNode(char *nodename, int flags) {
     node->replicaof = NULL;
     node->last_in_ping_gossip = 0;
     node->ping_sent = node->pong_received = 0;
+    node->last_connection_attempt = 0;
     node->data_received = 0;
     node->meet_sent = 0;
     node->fail_time = 0;
@@ -5282,10 +5283,12 @@ static int nodeExceedsHandshakeTimeout(clusterNode *node, mstime_t now) {
     return now - node->ctime > getHandshakeTimeout() ? 1 : 0;
 }
 
+#define MAX_CONNECTION_ATTEMPTS_PER_CRON 10
+
 /* Check if the node is disconnected and re-establish the connection.
  * Also update a few stats while we are here, that can be used to make
  * better decisions in other part of the code. */
-static int clusterNodeCronHandleReconnect(clusterNode *node, mstime_t now) {
+static int clusterNodeCronHandleReconnect(clusterNode *node, mstime_t now, int *cluster_conn_attempts) {
     /* Not interested in reconnecting the link with myself or nodes
      * for which we have no address. */
     if (node->flags & (CLUSTER_NODE_MYSELF | CLUSTER_NODE_NOADDR)) return 1;
@@ -5314,6 +5317,11 @@ static int clusterNodeCronHandleReconnect(clusterNode *node, mstime_t now) {
     }
 
     if (node->link == NULL) {
+        if ((now - node->last_connection_attempt) <= 1000 || *cluster_conn_attempts > MAX_CONNECTION_ATTEMPTS_PER_CRON) {
+            return 0;
+        }
+        node->last_connection_attempt = now;
+        (*cluster_conn_attempts)++;
         clusterLink *link = createClusterLink(node);
         link->conn = connCreate(connTypeOfCluster());
         connSetPrivateData(link->conn, link);
@@ -5370,6 +5378,7 @@ void clusterCron(void) {
     mstime_t min_pong = 0, now = mstime();
     clusterNode *min_pong_node = NULL;
     static unsigned long long iteration = 0;
+    int cluster_conn_attempts = 0;
 
     iteration++; /* Number of times this function was called so far. */
 
@@ -5387,7 +5396,7 @@ void clusterCron(void) {
         /* The protocol is that function(s) below return non-zero if the node was
          * terminated.
          */
-        if (clusterNodeCronHandleReconnect(node, now)) continue;
+        if (clusterNodeCronHandleReconnect(node, now, &cluster_conn_attempts)) continue;
     }
     dictReleaseIterator(di);
 

src/cluster_legacy.h

@@ -352,6 +352,7 @@ struct _clusterNode {
     mstime_t fail_time;                     /* Unix time when FAIL flag was set */
     mstime_t repl_offset_time;              /* Unix time we received offset for this node */
     mstime_t orphaned_time;                 /* Starting time of orphaned primary condition */
+    mstime_t last_connection_attempt;       /* Unix time we last tried to establish a link */
     mstime_t inbound_link_freed_time;       /* Last time we freed the inbound link for this node.
                                                If it was never freed, it is the same as ctime */
     long long repl_offset;                  /* Last known repl offset for this node. */