SQLite

This component allows using SQLite 3 as state store for Dapr.

The component is currently compiled with SQLite version 3.41.2.

Create a Dapr component

Create a file called sqlite.yaml, paste the following, and replace the <CONNECTION STRING> value with your connection string, which is the path to a file on disk.

If you want to also configure SQLite to store actors, add the actorStateStore option as in the example below.

apiVersion: dapr.io/v1alpha1
kind: Component
metadata:
  name: <NAME>
spec:
  type: state.sqlite
  version: v1
  metadata:
  # Connection string
  - name: connectionString
    value: "data.db"
  # Timeout for database operations, in seconds (optional)
  #- name: timeoutInSeconds
  #  value: 20
  # Name of the table where to store the state (optional)
  #- name: tableName
  #  value: "state"
  # Cleanup interval in seconds, to remove expired rows (optional)
  #- name: cleanupInterval
  #  value: "1h"
  # Set busy timeout for database operations
  #- name: busyTimeout
  #  value: "2s"
  # Uncomment this if you wish to use SQLite as a state store for actors (optional)
  #- name: actorStateStore
  #  value: "true"

Spec metadata fields

The connectionString parameter configures how to open the SQLite database.

• Normally, this is the path to a file on disk, relative to the current working directory, or absolute. For example: "data.db" (relative to the working directory) or "/mnt/data/mydata.db".
• The path is interpreted by the SQLite library, so it’s possible to pass additional options to the SQLite driver using “URI options” if the path begins with file:. For example: "file:path/to/data.db?mode=ro" opens the database at path path/to/data.db in read-only mode. Refer to the SQLite documentation for all supported URI options.
• The special case ":memory:" launches the component backed by an in-memory SQLite database. This database is not persisted on disk, not shared across multiple Dapr instances, and all data is lost when the Dapr sidecar is stopped. When using an in-memory database, Dapr automatically sets the cache=shared URI option.

Advanced

TTLs and cleanups

This state store supports Time-To-Live (TTL) for records stored with Dapr. When storing data using Dapr, you can set the ttlInSeconds metadata property to indicate when the data should be considered “expired”.

Because SQLite doesn’t have built-in support for TTLs, this is implemented in Dapr by adding a column in the state table indicating when the data is to be considered “expired”. Records that are “expired” are not returned to the caller, even if they’re still physically stored in the database. A background “garbage collector” periodically scans the state table for expired rows and deletes them.

The cleanupInterval metadata property sets the expired records deletion interval, which is disabled by default.

• Longer intervals require less frequent scans for expired rows, but can cause the database to store expired records for longer, potentially requiring more storage space. If you plan to store many records in your state table, with short TTLs, consider setting cleanupInterval to a smaller value, for example 5m.
• If you do not plan to use TTLs with Dapr and the SQLite state store, you should consider setting cleanupInterval to a value <= 0 (e.g. 0 or -1) to disable the periodic cleanup and reduce the load on the database. This is the default behavior.

The expiration_time column in the state table, where the expiration date for records is stored, does not have an index by default, so each periodic cleanup must perform a full-table scan. If you have a table with a very large number of records, and only some of them use a TTL, you may find it useful to create an index on that column. Assuming that your state table name is state (the default), you can use this query:

CREATE INDEX idx_expiration_time
  ON state (expiration_time);

Dapr does not automatically vacuum SQLite databases.

Sharing a SQLite database and using networked filesystems

Although you can have multiple Dapr instances accessing the same SQLite database (for example, because your application is scaled horizontally or because you have multiple apps accessing the same state store), there are some caveats you should keep in mind.

SQLite works best when all clients access a database file on the same, locally-mounted disk. Using virtual disks that are mounted from a SAN (Storage Area Network), as is common practice in virtualized or cloud environments, is fine.

However, storing your SQLite database in a networked filesystem (for example via NFS or SMB, but these examples are not an exhaustive list) should be done with care. The official SQLite documentation has a page dedicated to recommendations and caveats for running SQLite over a network.

Given the risk of data corruption that running SQLite over a networked filesystem (such as via NFS or SMB) comes with, we do not recommend doing that with Dapr in production environment. However, if you do want to do that, you should configure your SQLite Dapr component with disableWAL set to true.

Related links

• Basic schema for a Dapr component
• Read this guide for instructions on configuring state store components
• State management building block