HAProxy stops proxying

[simong]

app0> telnet proxy0 6379
Connection closed

cache0> telnet localhost 6379
Works fine

HAproxy is also no longer sending logs to syslog.

There are some nagios checks in #80 which monitor HAproxy's status

[simong]

For future reference:

HAProxy 1.5 will have a neat feature that should us to do failover and elect a new Redis master when we do. This might allow us to get rid of downinter 365 which means that if both instances go down, HAProxy can elect the new master. (Whereas now, if both of them go down, it needs manual intervention)

See http://blog.exceliance.fr/2014/01/02/haproxy-advanced-redis-health-check/ for more information

[mrvisser]

Does it actually elect a new master, or just check to see who the current master is? I wonder how it works when one goes down, because I'm not too sure a node explicitly becomes master in that scenario (it's still slave, but just with a disconnected master)

[mrvisser]

Might need something like sentinel to go with it which would actually promote a slave to master

[simong]

Durr you're right, I read that too fast and got a bit excited :(

[mrvisser]

Hey, turn that frown upside down. Maybe we can try Sentinel, it seems to at least be actively developed, and I think it's antirez who works on it. It has a bit more credibility than I originally thought with that big "experimental!" warning on it.