tsuru.conf reference

tsuru uses a configuration file in YAML format. This document describes what each option means, and how it should look.

Notation

tsuru uses a colon to represent nesting in YAML. So, whenever this document says something like key1:key2, it refers to the value of the key2 that is nested in the block that is the value of key1. For example, database:url means:

database:
  url: <value>

tsuru configuration

This section describes tsuru’s core configuration. Other sections will include configuration of optional components, and finally, a full sample file.

HTTP server

tsuru provides a REST API, that supports HTTP and HTTP/TLS (a.k.a. HTTPS). Here are the options that affect how tsuru’s API behaves:

listen

listen defines in which address tsuru webserver will listen. It has the form <host>:<port>. You may omit the host (example: :8080). This setting has no default value.

use-tls

use-tls indicates whether tsuru should use TLS or not. This setting is optional, and defaults to “false”.

tls:cert-file

tls:cert-file is the path to the X.509 certificate file configured to serve the domain. This setting is optional, unless use-tls is true.

tls:key-file

tls:key-file is the path to private key file configured to serve the domain. This setting is optional, unless use-tls is true.

server:read-timeout

server:read-timeout is the timeout of reading requests in the server. This is the maximum duration of any request to the tsuru server.

This is useful to avoid leaking connections, in case clients drop the connection before end sending the request. The default value is 0, meaning no timeout.

server:write-timeout

server:write-timeout is the timeout of writing responses in the server.

This is useful to avoid leaking connections, in case clients drop the connection before reading the response from tsuru. The default value is 0, meaning no timeout.

server:app-log-buffer-size

The maximum number of received log messages from applications to hold in memory waiting to be sent to the log database. The default value is 500000.

disable-index-page

tsuru API serves an index page with some basic instructions on how to use the current target. It’s possible to disable this page by setting the disable-index-page flag to true. It’s also possible to customize which template will be used in the index page, see the next configuration entry for more details.

This setting is optional, and defaults to false.

index-page-template

index-page-template is the template that will be used for the index page. It must use the Go template syntax, and tsuru will provide the following variables in the context of the template:

  • tsuruTarget: the target URL of the tsuru API serving the index page
  • userCreate: a boolean indicating whether user registration is enabled or disabled
  • nativeLogin: a boolean indicating whether the API is configured to use the native authentication scheme
  • keysEnabled: a boolean indicating whether the API is configured to manage SSH keys

It will also include a function used for querying configuration values, named getConfig. Here is an example of the function usage:

<body>
    {{if getConfig "use-tls"}}
    <p>we're safe</p>
    {{else}}
    <p>we're not safe</p>
    {{end}}
</body>

This setting is optional. When index-page-template is not defined, tsuru will use the default template.

Database access

tsuru uses MongoDB as a database manager to store information like users, machines, containers, etc. You need to describe how tsuru will connect to your database server. Therefore, it’s necessary to provide a MongoDB connection string. Database related options are listed below:

database:url

database:url is the database connection string. It is a mandatory setting and it has no default value. Examples of strings include basic 127.0.0.1 and more advanced mongodb://user:password@127.0.0.1:27017/database. Please refer to MongoDB documentation for more details and examples of connection strings.

database:name

database:name is the name of the database that tsuru uses. It is a mandatory setting and has no default value. An example of value is “tsuru”.

database:logdb-url

This setting is optional. If database:logdb-url is specified, tsuru will use it as the connection string to the MongoDB server responsible for storing application logs. If this value is not set, tsuru will use database:url instead.

This setting is useful because tsuru may have to process a very large number of log messages depending on the number of units deployed and applications behavior. Every log message will trigger a insertion in MongoDB and this may negatively impact the database performance. Other measures will be implemented in the future to improve this, but for now, having the ability to use an exclusive database server for logs will help mitigate the negative impact of log writing.

database:logdb-name

This setting is optional. If database:logdb-name is specified, tsuru will use it as the database name for storing application logs. If this value is not set, tsuru will use database:name instead.

Email configuration

tsuru sends email to users when they request password recovery. In order to send those emails, tsuru needs to be configured with some SMTP settings. Omitting these settings won’t break tsuru, but users will not be able to reset their password.

smtp:server

The SMTP server to connect to. It must be in the form <host>:<port>. Example: “smtp.gmail.com:587”.

smtp:user

The user to authenticate with the SMTP sever. Currently, tsuru requires authenticated sessions.

smtp:password

The password for authentication within the SMTP server.

Repository configuration

tsuru optionally uses Gandalf to manage git repositories. Gandalf exposes a REST API for repositories management and tsuru needs information about the Gandalf HTTP server endpoint.

repo-manager

repo-manager represents the repository manager that tsuru-server should use. For backward compatibility reasons, the default value is “gandalf”. Users can disable repository and SSH key management by setting “repo-manager” to “none”. For more details, please refer to the repository management page in the documentation.

git:api-server

git:api-server is the address of the Gandalf API. It should define the entire address, including protocol and port. Examples of value: http://localhost:9090 and https://gandalf.tsuru.io:9595.

Authentication configuration

tsuru has support for native, oauth and saml authentication schemes.

The default scheme is native and it supports the creation of users in tsuru’s internal database. It hashes passwords brcypt. Tokens are generated during authentication and are hashed using SHA512.

The auth section also controls whether user registration is on or off. When user registration is off, only admin users are able to create new users.

auth:scheme

The authentication scheme to be used. The default value is native, the other supported value is oauth.

auth:user-registration

This flag indicates whether user registration is enabled. This setting is optional, and defaults to false.

auth:hash-cost

Required only with native chosen as auth:scheme.

This number indicates how many CPU time you’re willing to give to hashing calculation. It is an absolute number, between 4 and 31, where 4 is faster and less secure, while 31 is very secure and very slow.

auth:token-expire-days

Required only with native chosen as auth:scheme.

Whenever a user logs in, tsuru generates a token for him/her, and the user may store the token. auth:token-expire-days setting defines the amount of days that the token will be valid. This setting is optional, and defaults to “7”.

auth:max-simultaneous-sessions

tsuru can limit the number of simultaneous sessions per user. This setting is optional, and defaults to “unlimited”.

auth:oauth

Every config entry inside auth:oauth are used when the auth:scheme is set to “oauth”. Please check rfc6749 for more details.

auth:oauth:client-id

The client id provided by your OAuth server.

auth:oauth:client-secret

The client secret provided by your OAuth server.

auth:oauth:scope

The scope for your authentication request.

auth:oauth:auth-url

The URL used in the authorization step of the OAuth flow. tsuru CLI will receive this URL and trigger the opening a browser on this URL with the necessary parameters.

During the authorization step, tsuru CLI will start a server locally and set the callback to http://localhost:<port>, if auth:oauth:callback-port is set tsuru CLI will use its value as <port>. If auth:oauth:callback-port isn’t present tsuru CLI will automatically choose an open port.

The callback URL should be registered on your OAuth server.

If the chosen server requires the callback URL to match the same host and port as the registered one you should register “http://localhost:<chosen port>” and set the auth:oauth:callback-port accordingly.

If the chosen server is more lenient and allows a different port to be used you should register simply “http://localhost” and leave auth:oauth:callback-port empty.

auth:oauth:token-url

The URL used in the exchange token step of the OAuth flow.

auth:oauth:info-url

The URL used to fetch information about the authenticated user. tsuru expects a json response containing a field called email.

tsuru will also make call this URL on every request to the API to make sure the token is still valid and hasn’t been revoked.

auth:oauth:collection

The database collection used to store valid access tokens. Defaults to “oauth_tokens”.

auth:oauth:callback-port

The port used in the callback URL during the authorization step. Check docs for auth:oauth:auth-url for more details.

auth:saml

Every config entry inside auth:saml are used when the auth:scheme is set to “saml”. Please check SAML V2.0 specification for more details.

auth:saml:sp-publiccert

Service provider public certificate path.

auth:saml:sp-privatekey

Service provider private key path.

auth:saml:idp-ssourl

Identity provider url.

auth:saml:sp-display-name

Service provider display name. The default value is Tsuru.

auth:saml:sp-description

Service provider description. The default values is Tsuru Platform as a Service software.

auth:saml:idp-publiccert

Identity provider public certificate.

auth:saml:sp-entityid

Service provider entitiy id.

auth:saml:sp-sign-request

Boolean value that indicates to service provider signs the request. The default value is false.

auth:saml:idp-sign-response

Boolean value that indicates to identity provider signs the response. The default value is false.

auth:saml:idp-deflate-encoding

Boolean value that indicates to identity provider to enable deflate encoding. The default value is false.

Queue configuration

tsuru uses a work queue for asynchronous tasks.

queue:* groups configuration settings for a MongoDB server that will be used as storage for delayed execution of queued jobs.

This queue is used to manage creation and destruction of IaaS machines, but tsuru may start using it in more places in the future.

It’s not mandatory to configure the queue, however creating and removing machines using a IaaS provider will not be possible.

queue:mongo-url

Connection url for MongoDB server used to store task information.

queue:mongo-database

Database name used in MongoDB. This value will take precedence over any database name already specified in the connection url.

pubsub

pubsub configuration is optional and depends on a redis server instance. It’s used only for following application logs (running tsuru app-log -f). If this is not configured tsuru will fail when running tsuru app-log -f.

Previously the configuration for this redis server was inside redis-queue:* keys shown below. Using these keys is deprecated and tsuru will start ignoring them before 1.0 release.

pubsub:redis-*

The Redis server to be used for pub/sub. For details on all available options for connecting to redis check common redis configuration

redis-queue:host

Deprecated. See pubsub:redis-*.

redis-queue:port

Deprecated. See pubsub:redis-*.

redis-queue:password

Deprecated. See pubsub:redis-*.

redis-queue:db

Deprecated. See pubsub:redis-*.

Quota management

tsuru can, optionally, manage quotas. Currently, there are two available quotas: apps per user and units per app.

tsuru administrators can control the default quota for new users and new apps in the configuration file, and use tsuru-admin command to change quotas for users or apps. Quota management is disabled by default, to enable it, just set the desired quota to a positive integer.

quota:units-per-app

quota:units-per-app is the default value for units per-app quota. All new apps will have at most the number of units specified by this setting. This setting is optional, and defaults to “unlimited”.

quota:apps-per-user

quota:apps-per-user is the default value for apps per-user quota. All new users will have at most the number of apps specified by this setting. This setting is optional, and defaults to “unlimited”.

Logging

Tsuru supports three logging flavors, that can be enabled or disabled altogether. The default behavior of tsuru is to send all logs to syslog, but it can also send logs to the standard error stream or a file. It’s is possible to use any combination of the three flavors at any time in tsuru configuration (e.g.: write logs both to stderr and syslog, or a file and stderr, or to all of the flavors simultaneously).

There’s also the possibility to enable or disable debugging log, via the debug flag.

debug

false is the default value, so you won’t see any noises on logs, to turn it on set it to true, e.g.: debug: true

log:file

Use this to specify a path to a log file. If no file is specified, tsuru-server won’t write logs to any file.

log:disable-syslog

log:disable-syslog indicates whether tsuru-server should disable the use of syslog. false is the default value. If it’s true, tsuru-server won’t send any logs to syslog.

log:syslog-tag

log:syslog-tag is the tag that will be attached to every log line. The default value is “tsr”.

log:use-stderr

log:use-stderr indicates whether tsuru-server should write logs to standard error stream. The default value is false.

Routers

As of 0.10.0, all your router configuration should live under entries with the format routers:<router name>.

routers:<router name>:type (type: hipache, galeb, vulcand)

Indicates the type of this router configuration. The standard router supported by tsuru is hipache. There is also experimental support for galeb and vulcand).

Depending on the type, there are some specific configuration options available.

routers:<router name>:domain (type: hipache, galeb, vulcand)

The domain of the server running your router. Applications created with tsuru will have a address of http://<app-name>.<domain>

routers:<router name>:redis-* (type: hipache)

Redis server used by Hipache router. This same server (or a redis slave of it), must be configured in your hipache.conf file. For details on all available options for connecting to redis check common redis configuration

routers:<router name>:api-url (type: galeb, vulcand)

The URL for the Galeb or vulcand manager API.

routers:<router name>:username (type: galeb)

Galeb manager username.

routers:<router name>:password (type: galeb)

Galeb manager password.

routers:<router name>:environment (type: galeb)

Galeb manager environment used to create virtual hosts and backend pools.

routers:<router name>:farm-type (type: galeb)

Galeb manager farm type used to create virtual hosts and backend pools.

routers:<router name>:plan (type: galeb)

Galeb manager plan used to create virtual hosts and backend pools.

routers:<router name>:project (type: galeb)

Galeb manager project used to create virtual hosts, backend pools and pools.

routers:<router name>:load-balance-policy (type: galeb)

Galeb manager load balancing policy used to create backend pools.

routers:<router name>:rule-type (type: galeb)

Galeb manager rule type used to create rules.

Hipache

hipache:redis-server

Redis server used by Hipache router. This same server (or a redis slave of it), must be configured in your hipache.conf file.

This setting is deprecated in favor of routers:<router name>:type = hipache and routers:<router name>:redis-server.

hipache:domain

The domain of the server running your hipache server. Applications created with tsuru will have a address of http://<app-name>.<hipache:domain>.

This setting is deprecated in favor of routers:<router name>:type = hipache and routers:<router name>:domain

Defining the provisioner

tsuru has extensible support for provisioners. A provisioner is a Go type that satisfies the provision.Provisioner interface. By default, tsuru will use DockerProvisioner (identified by the string “docker”), and now that’s the only supported provisioner (Ubuntu Juju was supported in the past but its support has been removed from tsuru).

provisioner

provisioner is the string the name of the provisioner that will be used by tsuru. This setting is optional and defaults to “docker”.

Docker provisioner configuration

docker:collection

Database collection name used to store containers information.

docker:port-allocator

The choice of port allocator. There are two possible values:

  • docker: trust Docker to allocate ports. Meaning that whenever a container restarts, the port might change (usually, it changes).
  • tsuru: leverage port allocation to tsuru, so ports mapped to containers never change.

The default value is “docker”.

docker:registry

For tsuru to work with multiple docker nodes, you will need a docker-registry. This should be in the form of hostname:port, the scheme cannot be present.

docker:registry-max-try

Number of times tsuru will try to send a image to registry.

docker:registry-auth:username

The username used for registry authentication. This setting is optional, for registries with authentication disabled, it can be omitted.

docker:registry-auth:password

The password used for registry authentication. This setting is optional, for registries with authentication disabled, it can be omitted.

docker:registry-auth:email

The email used for registry authentication. This setting is optional, for registries with authentication disabled, it can be omitted.

docker:repository-namespace

Docker repository namespace to be used for application and platform images. Images will be tagged in docker as <docker:repository-namespace>/<platform-name> and <docker:repository-namespace>/<app-name>

docker:max-layers

The maximum number of layers in Docker images. This number represents the number of times that Tsuru will reuse the previous image on application deployment. The default value is 10.

docker:bs:image

This setting is deprecated in favor of dynamically configuring with tsuru-admin docker-node-update big-sibling --image <image>.

docker:bs:socket

This setting is deprecated in favor of dynamically configuring with tsuru-admin docker-node-update big-sibling --volume <local>:<remote> --env DOCKER_ENDPOINT=<remote>.

docker:bs:syslog-port

docker:bs:syslog-port is the port in the Docker node that will be used by the bs container for collecting logs. The default value is 1514.

If this value is changed bs node containers must be update with tsuru-admin docker-node-update big-sibling --env SYSLOG_LISTEN_ADDRESS=udp://0.0.0.0:<port>.

docker:max-workers

Maximum amount of threads to be created when starting new containers, so tsuru doesn’t start too much threads in the process of starting 1000 units, for instance. Defaults to 0 which means unlimited.

docker:router

Default router to be used to distribute requests to units. This should be the name of a router configured under the routers:<name> key, see routers.

For backward compatibility reasons, the value hipache is also supported, and it will use either configuration available under router:hipache:* or hipache:*, in this order.

Note that as of 0.10.0, routers may be associated to plans, if when creating an application the chosen plan has a router value it will be used instead of the value set in docker:router.

The router defined in docker:router will only be used if the chosen plan doesn’t specify one.

docker:deploy-cmd

The command that will be called in your platform when a new deploy happens. The default value for platforms supported in tsuru’s basebuilder repository is /var/lib/tsuru/deploy.

docker:security-opts

This setting describes a list of security options that will be passed to containers. This setting must be a list, and has no default value. If one wants to specify just one value, it’s still needed to use the list notation:

docker:
  ...
  security-opts:
    - apparmor:PROFILE

For more details on the available options, please refer to the Docker documentation: <https://docs.docker.com/reference/run/#security-configuration>.

docker:segregate

Deprecated. As of tsuru 0.11.1, using segregate scheduler is the default setting. See Segregate Scheduler for details.

docker:scheduler:total-memory-metadata

This value describes which metadata key will describe the total amount of memory, in bytes, available to a docker node.

docker:scheduler:max-used-memory

This should be a value between 0.0 and 1.0 which describes which fraction of the total amount of memory available to a server should be reserved for app units.

The amount of memory available is found based on the node metadata described by docker:scheduler:total-memory-metadata config setting.

If this value is set, tsuru will try to find a node with enough unreserved memory to fit the creation of new units, based on how much memory is required by the plan used to create the application. If no node with enough unreserved memory is found, tsuru will ignore memory restrictions and let the scheduler choose any node.

This setting, along with docker:scheduler:total-memory-metadata, are also used by node auto scaling. See node auto scaling for more details.

docker:cluster:storage

This setting has been removed. You shouldn’t define it anymore, the only storage available for the docker cluster is now mongodb.

docker:cluster:mongo-url

Connection URL to the mongodb server used to store information about the docker cluster.

docker:cluster:mongo-database

Database name to be used to store information about the docker cluster.

docker:run-cmd:bin

The command that will be called on the application image to start the application. The default value for platforms supported in tsuru’s basebuilder repository is /var/lib/tsuru/start.

docker:run-cmd:port

The tcp port that will be exported by the container to the node network. The default value expected by platforms defined in tsuru’s basebuilder repository is 8888.

docker:user

The user tsuru will use to start the container. The value expected for basebuilder platforms is ubuntu.

docker:healing:heal-nodes

Boolean value that indicates whether tsuru should try to heal nodes that have failed a specified number of times. Healing nodes is only available if the node was created by tsuru itself using the IaaS configuration. Defaults to false.

docker:healing:active-monitoring-interval

Number of seconds between calls to <server>/_ping in each one of the docker nodes. If this value is 0 or unset tsuru will never call the ping URL. Defaults to 0.

docker:healing:disabled-time

Number of seconds tsuru disables a node after a failure. This setting is only valid if heal-nodes is set to true. Defaults to 30 seconds.

docker:healing:max-failures

Number of consecutive failures a node should have before triggering a healing operation. Only valid if heal-nodes is set to true. Defaults to 5.

docker:healing:wait-new-time

Number of seconds tsuru should wait for the creation of a new node during the healing process. Only valid if heal-nodes is set to true. Defaults to 300 seconds (5 minutes).

docker:healing:heal-containers-timeout

Number of seconds a container should be unresponsive before triggering the recreation of the container. A container is deemed unresponsive if it doesn’t call the set unit status URL (/apps/{app}/units/{unit}) with a started status. If this value is 0 or unset tsuru will never try to heal unresponsive containers. Defaults to 0.

docker:healing:events_collection

Collection name in mongodb used to store information about triggered healing events. Defaults to healing_events.

docker:healthcheck:max-time

Maximum time in seconds to wait for deployment time health check to be successful. Defaults to 120 seconds.

docker:image-history-size

Number of images available for rollback using tsuru app-deploy-rollback. tsuru will try to delete older images, but it may not be able to due to it being used as a layer to a newer image. tsuru will keep trying to remove these old images until they are not used as layers anymore. Defaults to 10 images.

docker:auto-scale:enabled

Enable node auto scaling. See node auto scaling for more details. Defaults to false.

docker:auto-scale:wait-new-time

Number of seconds tsuru should wait for the creation of a new node during the scaling up process. Defaults to 300 seconds (5 minutes).

docker:auto-scale:group-by-metadata

Deprecated. The pool is used to group nodes.

docker:auto-scale:metadata-filter

The name of a pool where auto scale will be enabled. Leave unset to allow dynamically configuring with tsuru-admin docker-autoscale-rule-set.

docker:auto-scale:max-container-count

Maximum number of containers per node, for count based scaling. See node auto scaling for more details. Leave unset to allow dynamically configuring with tsuru-admin docker-autoscale-rule-set.

docker:auto-scale:prevent-rebalance

Prevent rebalancing from happening when adding new nodes, or if a rebalance is needed. See node auto scaling for more details. Leave unset to allow dynamically configuring with tsuru-admin docker-autoscale-rule-set.

docker:auto-scale:run-interval

Number of seconds between two periodic runs of the auto scaling algorithm. Defaults to 3600 seconds (1 hour).

docker:auto-scale:scale-down-ratio

Ratio used when scaling down. Must be greater than 1.0. See node auto scaling for more details. Defaults to 1.33. Leave unset to allow dynamically configuring with tsuru-admin docker-autoscale-rule-set.

docker:limit:actions-per-host

The maximum number of simultaneous actions to run on a docker node. When the number of running actions is greater then the limit further actions will block until another action has finished. Setting this limit may help the stability of docker nodes with limited resources. If this value is set to 0 the limit is disabled. Default value is 0.

docker:limit:mode

The way tsuru will ensure docker:limit:actions-per-host limit is being respected. Possible values are local and global. Defaults to local. In local mode tsuru will only limit simultaneous actions from the current tsurud process. global mode uses MongoDB to ensure all tsurud servers using respects the same limit.

IaaS configuration

tsuru uses IaaS configuration to automatically create new docker nodes and adding them to your cluster when using docker-node-add command. See adding nodes for more details about how to use this command.

Attention

You should configure queue to be able to use IaaS.

General settings

iaas:default

The default IaaS tsuru will use when calling docker-node-add without specifying iaas=<iaas_name> as a metadata. Defaults to ec2.

iaas:node-protocol

Which protocol to use when accessing the docker api in the created node. Defaults to http.

iaas:node-port

In which port the docker API will be accessible in the created node. Defaults to 2375.

iaas:collection

Collection name on database containing information about created machines. Defaults to iaas_machines.

EC2 IaaS

iaas:ec2:key-id

Your AWS key id.

iaas:ec2:secret-key

Your AWS secret key.

iaas:ec2:user-data

A url for which the response body will be sent to ec2 as user-data. Defaults to a script which will run tsuru now installation.

iaas:ec2:wait-timeout

Number of seconds to wait for the machine to be created. Defaults to 300 (5 minutes).

CloudStack IaaS

iaas:cloudstack:api-key

Your api key.

iaas:cloudstack:secret-key

Your secret key.

iaas:cloudstack:url

The url for the cloudstack api.

iaas:cloudstack:user-data

A url for which the response body will be sent to cloudstack as user-data. Defaults to a script which will run tsuru now installation.

iaas:cloudstack:wait-timeout

Number of seconds to wait for the machine to be created. Defaults to 300 (5 minutes).

DigitalOcean IaaS

iaas:digitalocean:token

The access token used for communication with the DigitalOcean API.

iaas:digitalocean:url

The URL of the DigitalOcean API. This is optional, and defaults to “https://api.digitalocean.com/”.

iaas:digitalocean:user-data

A URL for which the response body will be sent to DigitalOcean as user-data. Defaults to a script which will run tsuru now installation.

Custom IaaS

You can define a custom IaaS based on an existing provider. Any configuration keys with the format iaas:custom:<name> will create a new IaaS with name.

iaas:custom:<name>:provider

The base provider name, it can be one of the supported providers: cloudstack or ec2.

iaas:custom:<name>:<any_other_option>

This will overwrite the value of iaas:<provider>:<any_other_option> for this IaaS. As an example, having the configuration below would allow you to call tsuru-admin docker-node-add iaas=region1_cloudstack ...:

iaas:
    custom:
        region1_cloudstack:
            provider: cloudstack
            url: http://region1.url/
            secret-key: mysecretkey
    cloudstack:
        api-key: myapikey

Common redis configuration options

<prefix>:redis-server

Connect to a single redis server. The redis server address should be in the format host:port. This parameter is mutually exclusive with <prefix>:redis-sentinel-addrs and <prefix>:redis-cluster-addrs.

<prefix>:redis-host

Alternative way to specify a single redis server to connect. Only the host name should be informed.

<prefix>:redis-port

The port used when <prefix>:redis-host is defined.

<prefix>:redis-sentinel-addrs

Connect to a farm of redis sentinel servers. It’s a comma separated list of host:port pairs. e.g.: 10.0.0.1:26379,10.0.0.2:26379. This parameter is mutually exclusive with <prefix>:redis-server and <prefix>:redis-cluster-addrs.

<prefix>:redis-sentinel-master

The master name for a sentinel farm. This parameter is mandatory when <prefix>:redis-sentinel-addrs is defined.

<prefix>:redis-cluster-addrs

Connect to a farm of redis cluster servers. It’s a comma separated list of host:port pairs. e.g.: 10.0.0.1:6379,10.0.0.2:6379. This parameter is mutually exclusive with <prefix>:redis-server and <prefix>:redis-sentinel-addrs.

<prefix>:redis-db

The db number selected when connecting to redis.

<prefix>:redis-password

The password used when connecting to redis.

<prefix>:redis-pool-size

The maximum number of simultaneously open connections to a redis server.

<prefix>:redis-max-retries

The number of times an unsuccessful command will be sent to redis again.

<prefix>:redis-pool-timeout

Duration in seconds to wait for a free redis connection once the maximum pool size defined in <prefix>:redis-pool-size is reached.

<prefix>:redis-pool-idle-timeout

Duration in seconds after which an idle connection will be discarded from the pool.

<prefix>:redis-dial-timeout

Duration in seconds after which an error will be returned if a connection to redis cannot be established.

<prefix>:redis-read-timeout

Duration in seconds after which an error will be returned if tsuru is still waiting for the response for an issued command.

<prefix>:redis-write-timeout

Duration in seconds after which an error will be returned if tsuru is still sending a command to redis.

Sample file

Here is a complete example:

listen: "0.0.0.0:8080"
debug: true
host: http://<machine-public-addr>:8080 # This port must be the same as in the "listen" conf
auth:
    user-registration: true
    scheme: native
database:
    url: <your-mongodb-server>:27017
    name: tsurudb
pubsub:
    redis-host: <your-redis-server>
    redis-port: 6379
queue:
    mongo-url: <your-mongodb-server>:27017
    mongo-database: queuedb
git:
    api-server: http://<your-gandalf-server>:8000
provisioner: docker
docker:
    router: hipache
    collection: docker_containers
    repository-namespace: tsuru
    deploy-cmd: /var/lib/tsuru/deploy
    cluster:
        storage: mongodb
        mongo-url: <your-mongodb-server>:27017
        mongo-database: cluster
    run-cmd:
        bin: /var/lib/tsuru/start
        port: "8888"
routers:
    hipache:
        type: hipache
        domain: <your-hipache-server-ip>.xip.io
        redis-server: <your-redis-server-with-port>