Notification Script not triggered on Failover

[Dominic Hopf]

Greetings, 

I have a setup of two Redis-Servers and two Sentinels watching over the Redis Servers. The Sentinel configuration basically is the following:

127.0.0.1:26379> sentinel masters

1)  1) "name"

    2) "mymaster"

    3) "ip"

    4) "172.16.12.96"

    5) "port"

    6) "6379"

    7) "runid"

    8) "4ff31341e69e9bc05cd993b393eea7809aa6f3ee"

    9) "flags"

   10) "master"

   11) "pending-commands"

   12) "0"

   13) "last-ping-sent"

   14) "0"

   15) "last-ok-ping-reply"

   16) "877"

   17) "last-ping-reply"

   18) "877"

   19) "down-after-milliseconds"

   20) "30000"

   21) "info-refresh"

   22) "8632"

   23) "role-reported"

   24) "master"

   25) "role-reported-time"

   26) "1123888"

   27) "config-epoch"

   28) "35"

   29) "num-slaves"

   30) "1"

   31) "num-other-sentinels"

   32) "1"

   33) "quorum"

   34) "2"

   35) "failover-timeout"

   36) "180000"

   37) "parallel-syncs"

   38) "1"

   39) "notification-script"

   40) "/usr/local/bin/redis-sentinel-notify"

So, when a failover takes place, I'd actually expect the defined notification-script to be triggered, but nothing like this happens according to the logs:

[3425] 08 Apr 10:19:46.843 # +sdown master mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:46.916 # +odown master mymaster 172.16.12.102 6379 #quorum 2/2

[3425] 08 Apr 10:19:46.916 # +new-epoch 35

[3425] 08 Apr 10:19:46.916 # +try-failover master mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:47.155 # +vote-for-leader 63f239c769985e600aed93b0e63e0f6d071cf43c 35

[3425] 08 Apr 10:19:47.209 # 172.16.12.96:26379 voted for 63f239c769985e600aed93b0e63e0f6d071cf43c 35

[3425] 08 Apr 10:19:47.279 # +elected-leader master mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:47.280 # +failover-state-select-slave master mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:47.346 # +selected-slave slave 172.16.12.96:6379 172.16.12.96 6379 @ mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:47.347 * +failover-state-send-slaveof-noone slave 172.16.12.96:6379 172.16.12.96 6379 @ mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:47.405 * +failover-state-wait-promotion slave 172.16.12.96:6379 172.16.12.96 6379 @ mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:48.269 # +promoted-slave slave 172.16.12.96:6379 172.16.12.96 6379 @ mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:48.269 # +failover-state-reconf-slaves master mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:48.291 # +failover-end master mymaster 172.16.12.102 6379

[3425] 08 Apr 10:19:48.291 # +switch-master mymaster 172.16.12.102 6379 172.16.12.96 6379

[3425] 08 Apr 10:19:48.291 * +slave slave 172.16.12.102:6379 172.16.12.102 6379 @ mymaster 172.16.12.96 6379

Are there any hints why this is not happening?

Thanks very much in advance & Best Regards,

Dominic

[Anton Iutin]

You need sentinel client-reconfig-script