10:45:03 GMT <ENDEH^> hello. I've an issue I can't seem to figure out (I've read the docs plus numerous stackoverflow posts etc). I have a 3 node redis + sentinel setup, in replication. When one of the redis instances goes down everything continues to operate smoothly. For instance, if the master redis instance goes down, there's a quick failover and slave 1 becomes the new master. However, if both the master node and the
10:45:09 GMT <ENDEH^> sentinel watching that node go down at the same time, the two other sentinels reach a quorum about the master being down, but the failover never happens. Is this by design, and if so, why do the Redis docs mention this as a valid setup that should work (or am I reading the docs wrong)?
10:55:23 GMT <minus> ENDEH^: failover should still happen with 2 sentinels if the quorum is 2
10:55:52 GMT <ENDEH^> quorum is 2, I'll paste some logs in a sec
11:08:35 GMT <ENDEH^> minus: https://pastebin.com/LRNy45z1
11:09:42 GMT <ENDEH^> also, redis-cli SENTINEL ckquorom says there are 2 sentinels and therefore no quorum, while quorum is set at 2
14:21:27 GMT <jezeniel> what will happen if EXEC failed in the middle of executing operations?
14:21:45 GMT <jezeniel> will the previous still be applied? or not?