Struct LiteConfig 

pub struct LiteConfig {

    pub memory_budget: usize,
    pub hnsw_percent: usize,
    pub csr_percent: usize,
    pub loro_percent: usize,
    pub query_percent: usize,
    pub kv_percent: usize,
    pub document_percent: usize,
    pub strict_percent: usize,
    pub columnar_percent: usize,
    pub timeseries_percent: usize,
    pub spatial_percent: usize,
    pub fts_percent: usize,
    pub array_percent: usize,
    pub sparse_percent: usize,
    pub sync_enabled: bool,
    pub argon2_m_cost: u32,
    pub argon2_t_cost: u32,
    pub argon2_p_cost: u32,
    pub kv_cache_capacity: usize,
    pub outbound_queue_cap: usize,
    pub auto_flush_ms: u64,
    pub auto_compact_ms: u64,
    pub corruption_policy: CorruptionPolicy,
}

Runtime configuration for a NodeDB-Lite instance.

All percentage fields express a fraction of memory_budget allocated to the corresponding engine. The remaining percentage is headroom (untracked).

Example

use nodedb_lite::config::LiteConfig;

let cfg = LiteConfig {
    memory_budget: 256 * 1024 * 1024, // 256 MiB
    ..LiteConfig::default()
};

Fields

memory_budget: usize

Total memory budget in bytes. Default: 100 MiB.

hnsw_percent: usize

Percentage of memory_budget reserved for HNSW vector index. Default: 40.

csr_percent: usize

Percentage of memory_budget reserved for CSR graph index. Default: 15.

loro_percent: usize

Percentage of memory_budget reserved for Loro CRDT engine. Default: 15.

query_percent: usize

Percentage of memory_budget reserved for query scratch space. Default: 15.

kv_percent: usize

Percentage of memory_budget reserved for the key-value engine. Default: 2.

document_percent: usize

Percentage of memory_budget reserved for the schemaless document engine. Default: 2.

strict_percent: usize

Percentage of memory_budget reserved for the strict document engine. Default: 2.

columnar_percent: usize

Percentage of memory_budget reserved for the columnar engine. Default: 2.

timeseries_percent: usize

Percentage of memory_budget reserved for the timeseries engine. Default: 1.

spatial_percent: usize

Percentage of memory_budget reserved for the spatial engine. Default: 1.

fts_percent: usize

Percentage of memory_budget reserved for the full-text search engine. Default: 1.

array_percent: usize

Percentage of memory_budget reserved for the array engine. Default: 1.

sparse_percent: usize

Percentage of memory_budget reserved for the sparse-vector metadata engine. Default: 1.

sync_enabled: bool

Open the durable outbound sync queues. Default: true.

When true, KV, columnar, timeseries, vector, FTS and spatial writes record themselves in their outbound queue, and the push loop sends them to Origin. When false, those writes stay local. Documents sync through Loro either way.

argon2_m_cost: u32

Argon2id memory cost in KiB. Default: 19 MiB (19_456 KiB). Corresponds to the OWASP recommended minimum for interactive login.

argon2_t_cost: u32

Argon2id iteration count. Default: 2.

argon2_p_cost: u32

Argon2id parallelism lanes. Default: 1.

kv_cache_capacity: usize

Maximum entries in the in-memory KV read cache. Default: 10_000.

Each entry holds the raw encoded value (typically ~80 bytes for a 64-byte payload + 8-byte TTL framing), so 10 000 entries ≈ 800 KB.

A value of 0 is rejected at open time; use 1 as the effective minimum.

outbound_queue_cap: usize

Maximum number of pending entries in each durable outbound queue (columnar and timeseries). Default: 100_000.

When a queue reaches this cap, write operations return [LiteError::Backpressure] until the sync transport drains entries. This bounds RAM usage to the key/pointer overhead regardless of how long the device stays offline; the payloads themselves are on disk.

Can also be set via the NODEDB_LITE_OUTBOUND_QUEUE_CAP environment variable.

auto_flush_ms: u64

Scheduled interval between background flushes, in milliseconds. Default: 1000 (one second).

Opening a database starts a task that applies buffered document, CRDT, and KV state and persists derived checkpoints. Scheduling delays and storage errors prevent a hard wall-clock durability bound. The task stops when the last database handle disappears.

Strict row writes commit before success. Native filesystem pagedb commits survive reopening without an explicit flush. Generic storage backends provide their own commit durability guarantees. Unindexed KV puts and all KV deletes remain buffered. Indexed KV puts and SQL KV writes commit directly. Derived checkpoint maintenance remains separate from strict commits. Later maintenance errors do not roll back committed rows.

Set to 0 to disable background flushing. Call flush() before dropping buffered state.

auto_compact_ms: u64

Interval between automatic background compactions, in milliseconds. Default: 0 (disabled).

When non-zero, opening a database spawns a background task that calls the global compact() every auto_compact_ms milliseconds, reclaiming dead pages and truncating the backing file to bound on-disk growth. Unlike auto-flush this is opt-in: compaction is a heavier operation (it repacks B+ trees and truncates the file, and no-ops while a reader pins the reclaimable range), so it is not imposed on every embedder by default.

Enable it when writing one commit per entry (where the deferred-free list would otherwise grow unbounded). A much larger interval than auto_flush_ms is appropriate — e.g. minutes, not seconds. Set to 0 to leave compaction fully manual via compact().

corruption_policy: CorruptionPolicy

What to do when the store being opened cannot be read. Default: CorruptionPolicy::FailClosed.

The default refuses to open and leaves the damaged store untouched, so an embedder that has no Origin to re-sync from does not silently start against an empty database. Selecting CorruptionPolicy::DiscardStoreAndRecreate is a statement that the data can be rebuilt from somewhere else.

Implementations

impl LiteConfig

pub fn from_env() -> Self

Load configuration from environment variables, falling back to defaults for any variable that is absent or malformed.

Handled variables:

• NODEDB_LITE_MEMORY_MB — total memory budget in mebibytes (parsed as usize)
• NODEDB_LITE_AUTO_FLUSH_MS — auto-flush interval in milliseconds (parsed as u64; 0 = disabled)
• NODEDB_LITE_AUTO_COMPACT_MS — auto-compact interval in milliseconds (parsed as u64; 0 = disabled, the default)
• NODEDB_LITE_OUTBOUND_QUEUE_CAP — max pending entries per durable outbound queue (parsed as usize; must be > 0)

impl LiteConfig

pub fn to_governor_config(&self) -> GovernorConfig

Build a GovernorConfig from this config’s budget and percentages.

Every _percent field maps to one EngineId Lite runs. Wal and Bridge get a zero limit: Lite has no WAL crate and no cross-plane bridge, so neither engine ever allocates here.

impl LiteConfig

pub fn validate(&self) -> NodeDbResult<()>

Validate that percentage fields are coherent.

Returns an error if:

• Any individual percentage exceeds 100
• The sum of all engine percentages exceeds MAX_TOTAL_ENGINE_PERCENT

Trait Implementations

impl Clone for LiteConfig

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for LiteConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for LiteConfig

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for LiteConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Eq for LiteConfig

impl PartialEq for LiteConfig

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for LiteConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for LiteConfig