diff options
| author | antirez <antirez@gmail.com> | 2017-04-15 10:08:39 +0200 |
|---|---|---|
| committer | antirez <antirez@gmail.com> | 2017-04-15 10:12:08 +0200 |
| commit | 271733f4f83552acc52a8baba4ae3fa7bd6b4ba0 (patch) | |
| tree | 3d9505d645e04a2b7b3384850d7c17d831a8346d | |
| parent | 3f068b92b98c664ccefb457103068583879609e3 (diff) | |
| download | redis-271733f4f83552acc52a8baba4ae3fa7bd6b4ba0.tar.gz | |
Cluster: discard pong times in the future.
However we allow for 500 milliseconds of tolerance, in order to
avoid often discarding semantically valid info (the node is up)
because of natural few milliseconds desync among servers even when
NTP is used.
Note that anyway we should ping the node from time to time regardless and
discover if it's actually down from our point of view, since no update
is accepted while we have an active ping on the node.
Related to #3929.
| -rw-r--r-- | src/cluster.c | 9 |
1 files changed, 8 insertions, 1 deletions
diff --git a/src/cluster.c b/src/cluster.c index b23160b90..d5ad85fe7 100644 --- a/src/cluster.c +++ b/src/cluster.c @@ -1365,7 +1365,14 @@ void clusterProcessGossipSection(clusterMsg *hdr, clusterLink *link) { { mstime_t pongtime = ntohl(g->pong_received); pongtime *= 1000; /* Convert back to milliseconds. */ - if (pongtime > node->pong_received) { + + /* Replace the pong time with the received one only if + * it's greater than our view but is not in the future + * (with 500 milliseconds tolerance) from the POV of our + * clock. */ + if (pongtime <= (server.mstime+500) && + pongtime > node->pong_received) + { node->pong_received = pongtime; } } |