Skip to main content
Declare Redis in floo.app.toml. The next deploy provisions it and injects the connection URL. See Managed services for the shared lifecycle: declaring, attaching credentials, and removal. Managed Redis is cache-only: treat its contents as disposable. floo makes no backup, restore, or durability guarantee for it; keep authoritative data in Postgres.
floo.app.toml

​
What gets injected

A named Redis handle provides REDIS_URL_<NAME> in upper case. [managed.cache] injects REDIS_URL_CACHE.
Every Redis service gets isolated dev and prod databases. The dev database is created with the service; the prod database is created on the first release or promote, and dev credentials are never copied to prod.

​
One handle is one instance

A managed Redis handle maps to exactly one instance, so every service that declares managed = ["redis:cache"] receives the same REDIS_URL_CACHE. A web process enqueuing a job and a worker draining it share one Redis by construction. The entire queue can be lost, so queued work must be regenerable from authoritative records in Postgres.

​
Separate concerns with named instances

To keep cache eviction away from your job queue, declare separate instances:
That injects REDIS_URL_CACHE and REDIS_URL_QUEUE, two physically separate databases. floo’s managed Redis is one logical database per instance. There is no SELECT n index switching, so named instances are the way to separate concerns. A Rails app can point its cache store at REDIS_URL_CACHE and Action Cable at REDIS_URL_QUEUE. Use Sidekiq on REDIS_URL_QUEUE only for jobs your app can regenerate from Postgres; a separate instance adds no durability guarantee.

​
Health and throughput limits

Managed Redis is billed as provider usage passed through at cost, and floo can reject data-plane traffic when a database exceeds its request rate or its command allowance. A TCP connection or a PING does not prove application commands are getting through. floo runs a health check against every ready database once a minute. It writes a random token under a reserved key, reads it back, and expires it. It never reads, lists, overwrites, or deletes one of your keys, and its commands are billed to floo rather than to your app. Run the doctor when Redis work or cron coordination fails:
throttle_request_rate means requests are arriving faster than the database allows. Reduce request concurrency or batch commands, then contact team@getfloo.com to raise managed throughput. throttle_daily_limit means the database exhausted its daily or monthly command allowance. Reduce command volume, then contact team@getfloo.com for paid capacity. throttle_temporary is throttling with no more specific published reason. Retry later, and contact the same address if it persists. Authentication, permission, configuration, missing, or stale-canary failures need floo intervention. The doctor gives the matching action. Its output never contains credentials or raw upstream responses.

​
Debug a connection issue

Check that the service has env.managed including the Redis handle, and that REDIS_URL_<NAME> appears in floo env list.

Managed services

The shared lifecycle: declare, attach, inspect, remove.

Postgres

Connection values, pgvector, and preview database branches.