fluencelabs/redis
1
0
Fork 0
mirror of https://github.com/fluencelabs/redis synced 2025-03-20 09:30:55 +00:00
Code Issues Projects Releases Wiki Activity

Cluster: a few random fixes to the new failure detection.

Browse Source
This commit is contained in:
antirez 2013-02-26 15:15:44 +01:00
parent f288b07563
commit d0992d6e8b
1 changed files with 14 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
src/cluster.c
View File

@ -567,11 +567,12 @@ void markNodeAsFailingIfNeeded(clusterNode *node) {
* node, so that no election was possible. */
void clearNodeFailureIfNeeded(clusterNode *node) {
if (node->flags & REDIS_NODE_FAIL && !node->numslaves) {
node->flags &= ~REDIS_NODE_FAIL;
clusterUpdateState();
redisLog(REDIS_NOTICE,
"Clear FAIL state for node %.40s: is reachable and has no slaves.",
node->name);
node->flags &= ~REDIS_NODE_FAIL;
clusterUpdateState();
clusterSaveConfigOrDie();
}
}
@ -618,7 +619,9 @@ void clusterProcessGossipSection(clusterMsg *hdr, clusterLink *link) {
node->pong_received = ntohl(g->pong_received);
}
/* Handle failure reports, only when the sender is a master. */
if (sender && sender->flags & REDIS_NODE_MASTER) {
if (sender && sender->flags & REDIS_NODE_MASTER &&
node != server.cluster->myself)
{
if (flags & (REDIS_NODE_FAIL|REDIS_NODE_PFAIL)) {
if (clusterNodeAddFailureReport(node,sender)) {
redisLog(REDIS_NOTICE,
@ -1239,7 +1242,14 @@ void clusterCron(void) {
if (node->ping_sent == 0 ||
node->ping_sent <= node->pong_received) continue;
delay = time(NULL) - node->pong_received;
/* If we never received a pong, use the ping time to compute
* the delay. */
if (node->pong_received) {
delay = time(NULL) - node->pong_received;
} else {
delay = time(NULL) - node->ping_sent;
}
if (delay < server.cluster->node_timeout) {
/* The PFAIL condition can be reversed without external
* help if it is not transitive (that is, if it does not