Skip to main content
Valkey is an open source datastore that can be used high-performance data retrieval and key-value storage. Upsun supports two different Valkey configurations:
  • Persistent: to set up fast persistent storage for your application
  • Ephemeral: to set up a non-persistent cache for your application

​
Supported versions

You can select the major and minor version. Patch versions are applied periodically for bug fixes and the like. When you deploy your app, you always get the latest available patches.

​
Service types

Depending on your needs, you can set up Valkey as persistent or ephemeral.

​
Relationship reference

For each service defined via a relationship to your application, Upsun automatically generates corresponding environment variables within your application container, in the $<RELATIONSHIP-NAME>_<SERVICE-PROPERTY> format. Here is example information available through the service environment variables themselves, or through the PLATFORM_RELATIONSHIPS environment variable.
You can obtain the complete list of available service environment variables in your app container by running upsun ssh env.Note that the information about the relationship can change when an app is redeployed or restarted or the relationship is changed. So your apps should only rely on the service environment variables directly rather than hard coding any values.
The format of the relationship is identical whether your Valkey service is ephemeral or persistent.

Database access

It should be noted that when you set up a relationship connection, access to all of the databases is automatically granted.

​
Persistent Valkey

By default, Valkey is an ephemeral service that stores data in memory. This allows for fast data retrieval,but also means data can be lost when a container is moved or shut down. To solve this issue, configure your Valkey service as persistent. Persistent Valkey stores data on a disk, restoring it if the container restarts. To switch from persistent to ephemeral Valkey, set up a new service with a different name.

Warning

Upsun sets the maximum amount of memory (maxmemory) Valkey can use for the data set,and it cannot be amended. It is defined by comparing the following values and keeping the lower of the two:
  • Disk size
  • The amount of memory allocated to the service container
For instance, if your Valkey container has 3072 MB of disk space and 1024 MB of memory, only 512 MB of RAM are actually available to the service (3072/6 = 512).But if your Valkey container has 3072 MB of disk space and 256 MB of memory, only 256 MB of Valkey are actually available to the service (as per the container limit).

​
Usage example

​
1. Configure the service

To define the service, use the valkey-persistent endpoint: Note that changing the name of the service replaces it with a brand new service and all existing data is lost. Back up your data before changing the service.

​
2. Define the relationship

To define the relationship, use the valkey endpoint :
You can define SERVICE_NAME as you like, so long as it’s unique between all defined services and matches in both the application and services configuration.The example above leverages default endpoint configuration for relationships. That is, it uses default endpoints behind the scenes, providing a relationship (the network address a service is accessible from) that is identical to the name of that service.Depending on your needs, instead of default endpoint configuration, you can use explicit endpoint configuration.With the above definition, the application container now has access to the service via the relationship SERVICE_NAME and its corresponding service environment variables.
For PHP, enable the extension for the service:

​
Configuration example

​
Ephemeral Valkey

By default, Valkey is an ephemeral service that serves as a non-persistent cache. Ephemeral Valkey stores data only in memory and requires no disk space. When the service reaches its memory limit, it triggers a cache cleanup. To customize those cache cleanups, set up an eviction policy. Make sure your app doesn’t rely on ephemeral Vedis for persistent storage as it can cause issues. For example, if a container is moved during region maintenance,the deploy and post_deploy hooks don’t run and an app that treats the cache as permanent shows errors. To prevent data from getting lost when a container is moved or shut down, you can use the persistent Valkey configuration. Persistent Valkey provides a cache with persistent storage.

​
Usage example

​
1. Configure the service

To define the service, use the valkey endpoint: Note that changing the name of the service replaces it with a brand new service and all existing data is lost. Back up your data before changing the service.

​
2. Define the relationship

To define the relationship, use the valkey endpoint :
You can define SERVICE_NAME as you like, so long as it’s unique between all defined services and matches in both the application and services configuration.The example above leverages default endpoint configuration for relationships. That is, it uses default endpoints behind the scenes, providing a relationship (the network address a service is accessible from) that is identical to the name of that service.Depending on your needs, instead of default endpoint configuration, you can use explicit endpoint configuration.With the above definition, the application container now has access to the service via the relationship SERVICE_NAME and its corresponding service environment variables.
For PHP, enable the extension for the service:

​
Configuration example

​
Use in app

To use the configured service in your app, add a configuration file similar to the following to your project. This configuration defines a single application (myapp), whose source code exists in the <PROJECT_ROOT>/myapp directory.
myapp has access to the valkey service, via a relationship whose name is identical to the service name (as per default endpoint configuration for relationships). From this, myapp can retrieve access credentials to the service through the relationship environment variables.
myapp/.environment
# Set environment variables for individual credentials.
# For more information, please visit https://docs.upsun.com/development/variables.html#service-environment-variables.
export CACHE_HOST="${VALKEY_HOST}"
export CACHE_PORT="${VALKEY_PORT}"
export CACHE_PASSWORD="${VALKEY_PASSWORD}"
export CACHE_SCHEME="${VALKEY_SCHEME}"

# Surface a Valkey connection string for use in app.
export CACHE_URL="${CACHE_SCHEME}://${CACHE_PASSWORD}@${CACHE_HOST}:${CACHE_PORT}"
The above file — .environment in the myapp directory — is automatically sourced by Upsun into the runtime environment, so that the variable CACHE_URL can be used within the application to connect to the service. Note that CACHE_URL, and all Upsun service environment variables like VALKEY_HOST, are environment-dependent. Unlike the build produced for a given commit, they can’t be reused across environments and only allow your app to connect to a single service instance on a single environment. A file very similar to this is generated automatically for your when using the upsun ify command to migrate a codebase to Upsun.

​
Eviction policy

When Valkey reaches its memory limit, it triggers a cache cleanup. To customize those cache cleanups, set up an eviction policy such as the following: The following table presents the possible values:
ValuePolicy description
allkeys-lruRemoves the oldest cache items first. This is the default policy when maxmemory_policy isn’t set.
noevictionNew items aren’t saved when the memory limit is reached.
allkeys-lfuRemoves least frequently used cache items first.
volatile-lruRemoves least recently used cache items with the expire field set to true.
volatile-lfuRemoves least frequently used cache items with the expire field set to true.
allkeys-randomRandomly removes cache items to make room for new data.
volatile-randomRandomly removes cache items with the expire field set to true.
volatile-ttlRemoves cache items with the expire field set to true and the shortest remaining time-to -live value.
For more information on the different policies, see the official Valkey documentation.

​
Access your Valkey service

After you’ve configured your Valkey service, you can access it using either the Upsun CLI or through the Valkey CLI.

​
Upsun CLI

Unlike the Valkey CLI, connecting via the Upsun CLI does not require additional authentication steps if you are already authenticated in your terminal. Access your Valkey service by running the command: 
upsun valkey

​
Valkey CLI

Retrieve the hostname and port you can connect to through the PLATFORM_RELATIONSHIPS environment variable. To do so, run the upsun relationships command. After you’ve retrieved the hostname and port, open an SSH session. To access your Valkey service, run the following command: To get the current configuration, run the following command:

​
Use Valkey as a handler for PHP sessions

A PHP session allows you to store different data for each user through a unique session ID. By default, PHP handles sessions using files. But you can use Valkey as a session handler, which means Valkey stores and retrieves the data saved into sessions. To set up Valkey as your session handler, add a configuration similar to the following:

​
Migrate from Redis to Valkey

It is possible for a user to switch from redis-persistent to valkey-persistent without losing data. To make this switch, change the type of the service from redis-persistent to valkey-persistent (also note the version change), while keeping the same service name. For example, replace this: with the following:

​
Exporting Data Persistent

This section applies only to persistent Valkey services configured with type: valkey-persistent. Ephemeral Valkey (type: valkey) stores data only in memory and does not need to be exported. The procedure is identical to Redis (persistent), using the same BGSAVE / RDB approach:
  1. Open an SSH tunnel:
Terminal
upsun tunnel:single --relationship <RELATIONSHIP_NAME>
  1. Trigger a background save:
Terminal
redis-cli -h 127.0.0.1 -p 30000 BGSAVE
  1. Download the dump.rdb file from the container as described in the Redis section.

​
Further resources

​
Documentation

Last modified on March 11, 2026