I am on a Windows platform. I am testing Redis Sentinels. I have a master,
slave, and 3 sentinels all monitoring the master. All of these Redis
servers are running as services.
I connection to one of the sentinels and issue the command "SENTINEL
masters". This command lists the masters that the sentinel is monitoring.
For my simple case it lists the master, the master address and the master
Hi!
I have tried for a while now to get my head around redis-sentinel. In my current configuration I have 1 master and 2 slaves. I have got it to work so that when
the master dies one of the slaves will take over. The problem is that if the newly promoted master would also die then the left alone slave won't get promoted to slave, shouldn't it?
I have one sentinel instance running on a separate
Hi
I am using Redis redis-3.0.0 in master-slave (1 master 2 slaves) mode. Also
followed the exact steps mentioned in http://redis.io/topics/sentinel with
different IPs.
sentinel.conf
========port 26379
daemonize yes
pidfile "/var/run/redis/sentinel.pid"
logfile "/var/log/redis/sentinel.log"
bind 10.1.161.246
dir "/data/redis"
sentinel monitor mymaster 10.1.161.244 6379 2
sentinel
Hi,
I am somewhat new to Redis and am trying to test failover. I have a 3 node
cluster. Each node has a sentinel and a redis-server. Sometimes (not often)
the failover takes minutes even though I have tuned the timers way down.
What is happening here?
My logs and configs are below from redis-server and sentinel.
Interesting timers from the redis configs:
timeout 0
tcp-keepalive
Hello redis-gurus,
I'm naive to redis and am working on setting up redis cluster with
automatic failover/failback via sentinel
server1 - redis & sentinel
server2 - redis & sentinel
appservers - atleast 10
At present I m running sentinel on redis server only, but would also move
it to appserver to comply quorum logic.
1. redis-sentinel 2.8.8
2. redis-server 2.8.21
____ server1
Hi,
wondering how every one is monitoring sentinels. In particular I'm
interested in alerts when we loose enough quorum to failover and other
situations that can put failover at risk…
Any suggestions?
Thanks,
Pedro Melo
@pedromelo
http://www.simplicidade.org/
xmpp:
[email protected]
mailto:
[email protected]
You received this message because you are subscribed to the Google
Hi,
I am experimenting Redis Sentinel.
Everything seems to be good at first, but after second restart of my
master, nothing seems to be working.
First time, I shut down my master and the slave became master. After few
seconds, I started my previous master and it became master. Second time,
when I shutdown, slave didn't became master and afterwards, it's not
getting elected as master
Hi all,
I have a Redis HA configuration: 2 Redis instances, 3 Sentinels.
NODE A
Redis Master + Sentinel
NODE B
Redis Slave + Sentinel
NODE C
Sentinel
Everything is ok.
Then I stop Master and Slave (at the same time).
After few seconds I restart only Slave (in its redis.conf file there is
"slaveof " setting).
Redis Slave continuosly logs:
*Error reply to PING from master
Hello:
We have Sentinel and Redis in production for more than 3 months. Today
we got max coonection error from Sentinel, even can't telnet to it.
After some analysis, looks like the Redis Sentinel didn't detect the dead
tcp
connection and stay there for ever.
Is there a configuration for Sentinel to detect the dead tcp
connection?
Best Regards,
Michael
You received this message
I'm analyzing actual output of Sentinel events during failover, and I think
I've found some things which could be improved a bit. Specifically the
+elected-leader, +vote-for-leader, and +try-failover events. I think they
should be instance-type sentinel, indeed possibly all of the events
published during a failover should have the left-side instance details be
that of the sentinel, with the master
Hi,
I have a 3 server redis sentinel setup:
Server 1 : 10 master redis / redis sentinel
Server 2: 10 slave redis / redis sentinel
Server 3: Application / redis sentinel
When I kill redis servers the sentinels are doing failover so no problem
here. However, when I intentionally pull the network cable from lets say
Server 1, there are times that all redis slaves becomes the
Hi,
I'm trying to learn about redis sentinel - plan to use it in my application.
I have setup a 2-vm setup locally to try out sentinel configuration and
corresponding changes in my app.
o. Both VMs are running redis-server - one working as master, the other as
slave.
o. Both VMs have sentinel running.
o. My desktop has the 3rd sentinel running but no redis-server - to test
out the
Good day!
I'm trying to use client list command to list sentinel connections and it's
not working in version 2.8.8 and 2.8.24 (the current versions that I have
set up).
I've tested in the most recent version and client list from the redis-cli
while connecting to the sentinel port works, but it doesn't in the version
I have setup.
Is CLIENT LIST supported for sentinel in redis 2.8.24?
I'm having trouble attempting to manually fail over Redis Sentinel using the�SENTINEL failover �command. The failover hangs on the�+failover-state-wait-promotionstep and ultimately ends up aborting with a�-failover-abort-slave-timeout�message.
This is running Redis 2.8.12 on the 64-bit release of Red Hat Enterprise Linux 6.2. I've included the relevant commands, log excerpts, configurations, and
I've been playing around with Redis/Sentinel instances, and I've come across some interesting situations. �Here's the configuration:
A - master
B - slave of A
C - Sentinel 1 connecting to A
D - Sentinel 2 connecting to A
(Quorum is 2)
Pre-steps:
Start ABCD
A dies
B is promoted to master
A is restarts and gets demoted to a slave of B
We now have a successful master resurrection:
A - slave of B
B -
Hi,
How long does Sentinel take to complete�the master failover?
In my test, it took about 20 secondes for Sentinel to complete master failover. Details are shown as follow:
3s +sdown - master mymaster 127.0.0.1 6379
15s +failover-state-wait-start - master mymaster 127.0.0.1 6379 #starting in 14753 milliseconds
2s +switch-master - mymaster 127.0.0.1 6379 127.0.0.1 6381
I am wondering is there any
Hi all!
I'm now working on redis-py sentinel support. And I have a question.
In guidelines for sentinel clients it is recommended to use�get-master-addr-by-name:
http://redis.io/topics/sentinel-clients
I'm not sure it's a good idea. I think that first of all we must check that sentinel is ok itself.
Example:
We have 3 nodes, each runs redis-server and sentinel instance. One of them is a master
I think I am almost to the point that I understand the Redis server with
sentinels failover process but I am a little shaky on what is available the
the StackExchange.Redis library to handle such a case. My initial attempt
would be to have a ConnectionMultiplexer for the sentinels and for the
master/save servers. I would start out by getting the master server from
the master/slave ConnectionMultiplexer
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"
This may sound strange, but the set up fits into our situation. We only
need to use Redis as a cache system, where replication is not needed. Since
every restart on Redis we like it to be empty to begin with. Also we don't
do replication due to memory concern (we are setting aside 128 GB of RAM
for the master due to our object size). However, I like to ensure the high
system (not data) availability
