Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Tree: 8487142eac
Fetching contributors…

Cannot retrieve contributors at this time

447 lines (335 sloc) 23.528 kB

Redis Sentinel Documentation

Redis Sentinel is a system designed to help managing Redis instances. It performs the following three tasks:

  • Monitoring. Sentinel constantly check if your master and slave instances are working as expected.
  • Notification. Sentinel can notify the system administrator, or another computer program, via an API, that something is wrong with one of the monitored Redis instances.
  • Automatic failover. If a master is not working as expected, Sentinel can start a failover process where a slave is promoted to master, the other additional slaves are reconfigured to use the new master, and the applications using the Redis server informed about the new address to use when connecting.

Redis Sentinel is a distributed system, this means that usually you want to run multiple Sentinel processes across your infrastructure, and this processes will use agreement protocols in order to understand if a master is down and to perform the failover.

Redis Sentinel is shipped as a stand-alone executable called redis-sentinel but actually it is a special execution mode of the Redis server itself, and can be also invoked using the --sentinel option of the normal redis-sever executable.

WARNING: Redis Sentinel is currently a work in progress. This document describes how to use what we already have and may change as the Sentinel implementation changes.

Redis Sentinel is compatible with Redis 2.4.16 or greater, and redis 2.6.0-rc6 or greater.

Obtaining Sentinel

Currently Sentinel is part of the Redis unstable branch at github. To compile it you need to clone the unstable branch and compile Redis. You'll see a redis-sentinel executable in your src directory.

Alternatively you can use directly the redis-server executable itself, starting it in Sentinel mode as specified in the next paragraph.

Running Sentinel

If you are using the redis-sentinel executable (or if you have a symbolic link with that name to the redis-server executable) you can run Sentinel with the following command line:

redis-sentinel /path/to/sentinel.conf

Otherwise you can use directly the redis-server executable starting it in Sentinel mode:

redis-server /path/to/sentine.conf --sentinel

Both ways work the same.

Configuring Sentinel

In the root of the Redis source distribution you will find a sentinel.conf file that is a self-documented example configuration file you can use to configure Sentinel, however a typical minimal configuration file looks like the following:

sentinel monitor mymaster 6379 2
sentinel down-after-milliseconds mymaster 60000
sentinel failover-timeout mymaster 900000
sentinel can-failover mymaster yes
sentinel parallel-syncs mymaster 1

sentinel monitor resque 6380 4
sentinel down-after-milliseconds resque 10000
sentinel failover-timeout resque 900000
sentinel can-failover resque yes
sentinel parallel-syncs resque 5

The first line is used to tell Redis to monitor a master called mymaster, that is at address and port 6379, with a level of agreement needed to detect this master as failing of 2 sentinels (if the agreement is not reached the automatic failover does not start).

The other options are almost always in the form:

sentinel <option_name> <master_name> <option_value>

And are used for the following purposes:

  • down-after-milliseconds is the time in milliseconds an instance should not be reachable (either does not reply to our PINGs or it is replying with an error) for a Sentinel starting to think it is down. After this time has elapsed the Sentinel will mark an instance as subjectively down (also known as SDOWN), that is not enough to start the automatic failover. However if enough instances will think that there is a subjectively down condition, then the instance is marked as objectively down. The number of sentinels that needs to agree depends on the configured agreement for this master.
  • can-failover tells this Sentinel if it should start a failover when an instance is detected as objectively down (also called ODOWN for simplicity). You may configure all the Sentinels to perform the failover if needed, or you may have a few Sentinels used only to reach the agreement, and a few more that are actually in charge to perform the failover.
  • parallel-syncs sets the number of slaves that can be reconfigured to use the new master after a failover at the same time. The lower the number, the more time it will take for the failover process to complete, however if the slaves are configured to serve old data, you may not want all the slaves to resync at the same time with the new master, as while the replication process is mostly non blocking for a slave, there is a moment when it stops to load the bulk data from the master during a resync. You may make sure only one slave at a time is not reachable by setting this option to the value of 1.

The other options are described in the rest of this document and documented in the example sentinel.conf file shipped with the Redis distribution.


As already briefly mentioned in this document Redis Sentinel has two different concepts of being down, one is called a Subjectively Down condition (SDOWN) and is a down condition that is local to a given Sentinel instance. Another is called Objectively Down condition (ODOWN) and is reached when enough Sentinels (at least the number configured as the quorum parameter of the monitored master) have an SDOWN condition, and get feedbacks from other Sentinels using the SENTINEL is-master-down-by-addr command.

From the point of view of a Sentienl an SDOWN condition is reached if we don't receive a valid reply to PING requests for the number of seconds specified in the configuration as is-master-down-after-milliseconds parameter.

An acceptable reply to PING is one of the following:

  • PING replied with +PONG.
  • PING replied with -LOADING error.
  • PING replied with -MASTERDOWN error.

Any other reply (or no reply) is considered non valid.

Note that SDOWN requires that no acceptable reply is received for the whole interval configured, so for instance if the interval is 30000 milliseconds (30 seconds) and we receive an acceptable ping reply every 29 seconds, the instance is considered to be working.

The ODOWN condition only applies to masters. For other kind of instances Sentinel don't require any agreement, so the ODOWN state is never reached for slaves and other sentinels.

Sentinels and Slaves auto discovery

While Sentinels stay connected with other Sentinels in order to reciprocally check the availability of each other, and to exchange messages, you don't need to configure the other Sentinel addresses in every Sentinel instance you run, as Sentinel uses the Redis master Pub/Sub capabilities in order to discover the other Sentinels that are monitoring the same master.

This is obtained by sending Hello Messages into the channel named __sentinel__:hello.

Similarly you don't need to configure what is the list of the slaves attached to a master, as Sentinel will auto discover this list querying Redis.

Sentinel API

By default Sentinel runs using TCP port 26379 (note that 6379 is the normal Redis port). Sentinels accept commands using the Redis protocol, so you can use redis-cli or any other unmodified Redis client in order to talk with Sentinel.

There are two ways to talk with Sentinel: it is possible to directly query it to check what is the state of the monitored Redis instances from its point of view, to see what other Sentinels it knows, and so forth.

An alternative is to use Pub/Sub to receive push style notifications from Sentinels, every time some event happens, like a failover, or an instance entering an error condition, and so forth.

Sentinel commands

The following is a list of accepted commands:

  • PING this command simply returns PONG.
  • SENTINEL masters show a list of monitored masters and their state.
  • SENTIENL slaves <master name> show a list of slaves for this master, and their state.
  • SENTINEL is-master-down-by-addr <ip> <port> return a two elements multi bulk reply where the first is 0 or 1 (0 if the master with that address is known and is in SDOWN state, 1 otherwise). The second element of the reply is the subjective leader for this master, that is, the runid of the Redis Sentinel instance that should perform the failover accordingly to the queried instance.
  • SENTINEL get-master-addr-by-name <master name> return the ip and port number of the master with that name. If a failover is in progress or terminated successfully for this master it returns the address and port of the promoted slave.
  • SENTINEL reset <pattern> this command will reset all the masters with matching name. The pattern argument is a glob-style pattern. The reset process clears any previous state in a master (including a failover in progress), and removes every slave and sentinel already discovered and associated with the master.

Pub/Sub Messages

A client can use a Sentinel as it was a Redis compatible Pub/Sub server (but you can't use PUBLISH) in order to SUBSCRIBE or PSUBSCRIBE to channels and get notified about specific events.

The channel name is the same as the name of the event. For instance the channel named +sdown will receive all the notifications related to instances entering an SDOWN condition.

To get all the messages simply subscribe using PSUBSCRIBE *.

The following is a list of channels and message formats you can receive using this API. The first word is the channel / event name, the rest is the format of the data.

Note: where instance details is specified it means that the following arguments are provided to identify the target instance:

<instance-type> <name> <ip> <port> @ <master-name> <master-ip> <master-port>

The part identifying the master (from the @ argument to the end) is optional and is only specified if the instance is not a master itself.

  • +reset-master <instance details> -- The master was reset.
  • +slave <instance details> -- A new slave was detected and attached.
  • +failover-state-reconf-slaves <instance details> -- Failover state changed to reconf-slaves state.
  • +failover-detected <instance details> -- A failover started by another Sentinel or any other external entity was detected (An attached slave turned into a master).
  • +salve-reconf-sent <instance details> -- The leader sentinel sent the SLAVEOF command to this instance in order to reconfigure it for the new slave.
  • +salve-reconf-inprog <instance details> -- The slave being reconfigured showed to be a slave of the new master ip:port pair, but the synchronization process is not yet complete.
  • +salve-reconf-done <instance details> -- The slave is now synchronized with the new master.
  • -dup-sentinel <instance details> -- One or more sentinels for the specified master were removed as duplicated (this happens for instance when a Sentinel instance is restarted).
  • +sentinel <instance details> -- A new sentinel for this master was detected and attached.
  • +sdown <instance details> -- The specified instance is now in Subjectively Down state.
  • -sdown <instance details> -- The specified instance is no longer in Subjectively Down state.
  • +odown <instance details> -- The specified instance is now in Objectively Down state.
  • -odown <instance details> -- The specified instance is no longer in Objectively Down state.
  • +failover-takedown <instance details> -- 25% of the configured failover timeout has elapsed, but this sentinel can't see any progress, and is the new leader. It starts to act as the new leader reconfiguring the remaining slaves to replicate with the new master.
  • +failover-triggered <instance details> -- We are starting a new failover as a the leader sentinel.
  • +failover-state-wait-start <instance details> -- New failover state is wait-start: we are waiting a fixed number of seconds, plus a random number of seconds before starting the failover.
  • +failover-state-select-slave <instance details> -- New failover state is select-slave: we are trying to find a suitable slave for promotion.
  • no-good-slave <instance details> -- There is no good slave to promote. Currently we'll try after some time, but probably this will change and the state machine will abort the failover at all in this case.
  • selected-slave <instance details> -- We found the specified good slave to promote.
  • failover-state-send-slaveof-noone <instance details> -- We are trynig to reconfigure the promoted slave as master, waiting for it to switch.
  • failover-end-for-timeout <instance details> -- The failover terminated for timeout. If we are the failover leader, we sent a best effort SLAVEOF command to all the slaves yet to reconfigure.
  • failover-end <instance details> -- The failover terminated with success. All the slaves appears to be reconfigured to replicate with the new master.
  • switch-master <master name> <oldip> <oldport> <newip> <newport> -- We are starting to monitor the new master, using the same name of the old one. The old master will be completely removed from our tables.
  • failover-abort-x-sdown <instance details> -- The failover was undoed (aborted) because the promoted slave appears to be in extended SDOWN state.
  • -slave-reconf-undo <instance details> -- The failover aborted so we sent a SLAVEOF command to the specified instance to reconfigure it back to the original master instance.
  • +tilt -- Tilt mode entered.
  • -tilt -- Tilt mode exited.

The Redis CLIENT SENTINELS command

  • Work in progress, not yet implemented in Redis instances.

Sentinel failover

The failover process consists on the following steps:

  • Recognize that the master is in ODOWN state.
  • Understand who is the Sentinel that should start the failover, called The Leader. All the other Sentinels will be The Observers.
  • The leader selects a slave to promote to master.
  • The promoted slave is turned into a master with the command SLAVEOF NO ONE.
  • The observers see that a slave was turned into a master, so they know the failover started. Note: this means that any event that turns one of the slaves of a monitored master into a master (SLAVEOF NO ONE command) will be sensed as the start of a failover process.
  • All the other slaves attached to the original master are configured with the SLAVEOF command in order to start the replication process with the new master.
  • The leader terminates the failover process when all the slaves are reconfigured. It removes the old master from the table of monitored masters and adds the new master, under the same name of the original master.
  • The observers detect the end of the failover process when all the slaves are reconfigured. They remove the old master from the table and start monitoring the new master, exactly as the leader does.

The election of the Leader is performed using the same mechanism used to reach the ODOWN state, that is, the SENTINEL is-master-down-by-addr command. It returns the leader from the point of view of the queried Sentinel, we call it the Subjective Leader, and is selected using the following rule:

  • We remove all the Sentinels that can't failover for configuration (this information is propagated using the Hello Channel to all the Sentinels).
  • We remove all the Sentinels in SDOWN, disconnected, or with the last ping reply received more than SENTINEL_INFO_VALIDITY_TIME milliseconds ago (currently defined as 5 seconds).
  • Of all the remaining instances, we get the one with the lowest runid, lexicographically (every Redis instance has a Run ID, that is an identifier of every single execution).

For a Sentinel to sense to be the Objective Leader, that is, the Sentinel that should start the failove process, the following conditions are needed.

  • It thinks it is the subjective leader itself.
  • It receives acknowledges from other Sentinels about the fact it is the leader: at least 50% plus one of all the Sentinels that were able to reply to the SENTINEL is-master-down-by-addr request shoudl agree it is the leader, and additionally we need a total level of agreement at least equal to the configured quorum of the master instance that we are going to failover.

Once a Sentinel things it is the Leader, the failover starts, but there is always a delay of five seconds plus an additional random delay. This is an additional layer of protection because if during this period we see another instance turning a slave into a master, we detect it as another instance staring the failover and turn ourselves into an observer instead.

End of failover

The failover process is considered terminated from the point of view of a single Sentinel if:

  • The promoted slave is not in SDOWN condition.
  • A slave was promoted as new master.
  • All the other slaves are configured to use the new master.

Note: Slaves that are in SDOWN state are ignored.

Also the failover state is considered terminate if:

  • The promoted slave is not in SDOWN condition.
  • A slave was promoted as new master.
  • At least failover-timeout milliseconds elapsed since the last progress.

The failover-timeout value can be configured in sentinel.conf for every different slave.

Note that when a leader terminates a failover for timeout, it sends a SLAVEOF command in a best-effort way to all the slaves yet to be configured, in the hope that they'll receive the command and replicate with the new master eventually.

Leader failing during failover

If the leader fails when it has yet to promote the slave into a master, and it fails in a way that makes it in SDOWN state from the point of view of the other Sentinels, if enough Sentinels remained to reach the quorum the failover will automatically continue using a new leader (the subjective leader of all the remaining Sentinels will change because of the SDOWN state of the previous leader).

If the failover was already in progress and the slave was already promoted, and possibly a few other slaves were already reconfigured, an observer that is the new objective leader will continue the failover in case no progresses are made for more than 25% of the time specified by the failover-timeout configuration option.

Note that this is safe as multiple Sentinels trying to reconfigure slaves with duplicated SLAVEOF commands do not create any race condition, but at the same time we want to be sure that all the slaves are reconfigured in the case the original leader is no longer working.

Promoted slave failing during failover

If the promoted slave has an active SDOWN condition, a Sentinel will never sense the failover as terminated.

Additionally if there is an extended SDOWN condition (that is an SDOWN that lasts for more than ten times down-after-milliseconds milliseconds) the failover is aborted (this happens for leaders and observers), and the master starts to be monitored again as usually, so that a new failover can start with a different slave in case the master is still failing.

Note that when this happens it is possible that there are a few slaves already configured to replicate from the (now failing) promoted slave, so when the leader sentinel aborts a failover it sends a SLAVEOF command to all the slaves already reconfigured or in the process of being reconfigured to switch the configuration back to the original master.

Manual interactions

  • TODO: Manually triggering a failover with SENTINEL FAILOVER.
  • TODO: Pausing Sentinels with SENTINEL PAUSE, RESUME.

The failback process

  • TODO: Sentinel does not perform automatic Failback.
  • TODO: Document correct steps for the failback.

Clients configuration update

Work in progress.

TILT mode

Redis Sentinel is heavily dependent on the computer time: for instance in order to understand if an instance is available it remembers the time of the latest successful reply to the PING command, and compares it with the current time to understand how old it is.

However if the computer time changes in an unexpected way, or if the computer is very busy, or the process blocked for some reason, Sentinel may start to behave in an unexpected way.

The TILT mode is a special "protection" mode that a Sentinel can enter when something odd is detected that can lower the reliability of the system. The Sentinel timer interrupt is normally called 10 times per second, so we expect that more or less 100 milliseconds will elapse between two calls to the timer interrupt.

What a Sentinel does is to register the previous time the timer interrupt was called, and compare it with the current call: if the time difference is negative or unexpectedly big (2 seconds or more) the TILT mode is entered (or if it was already entered the exit from the TILT mode postponed).

When in TILT mode the Sentinel will continue to monitor everything, but:

  • It stops acting at all.
  • It starts to reply negatively to SENTINEL is-master-down-by-addr requests as the ability to detect a failure is no longer trusted.

If everything appears to be normal for 30 second, the TILT mode is exited.

Handling of -BUSY state

(Warning: Yet not implemented)

The -BUSY error is returned when a script is running for more time than the configured script time limit. When this happens before triggering a fail over Redis Sentinel will try to send a "SCRIPT KILL" command, that will only succeed if the script was read-only.

Notifications via user script

Work in progress.

Suggested setup

Work in progress.

APPENDIX A - Get started with Sentinel in five minutes

Work in progress.

APPENDIX B - Implementation and algorithms

Duplicate Sentinels removal

In order to reach the configured quorum we absolutely want to make sure that the quorum is reached by different physical Sentinel instances. Under no circumstance we should get agreement from the same instance that for some reason appears to be two or multiple distinct Sentinel instances.

This is enforced by an aggressive removal of duplicated Sentinels: every time a Sentinel sends a message in the Hello Pub/Sub channel with its address and runid, if we can't find a perfect match (same runid and address) inside the Sentinels table for that master, we remove any other Sentinel with the same runid OR the same address. And later add the new Sentinel.

For instance if a Sentinel instance is restarted, the Run ID will be different, and the old Sentinel with the same IP address and port pair will be removed.

Selection of the Slave to promote

If a master has multiple slaves, the slave to promote to master is selected checking the slave priority (a new configuration option of Redis instances that is propagated via INFO output, still not implemented), and picking the one with lower priority value (it is an integer similar to the one of the MX field of the DNS system).

All the slaves that appears to be disconnected from the master for a long time are discarded.

If slaves with the same priority exist, the one with the lexicographically smaller Run ID is selected.

Note: because currently slave priority is not implemented, the selection is performed only discarding unreachable slaves and picking the one with the lower Run ID.

Jump to Line
Something went wrong with that request. Please try again.