SKILL.md
Storage Sync
Synchronize memories between Turso (durable) and redb (cache) storage layers.
When to Sync
- On startup - After system initialization
- Periodic - Scheduled background sync
- Cache staleness - When redb appears outdated
- Recovery - After storage failures
- Manual trigger - When explicitly requested
Sync Process
- Check connection health:
``rust tursoclient.ping().await?; redbenv.check_integrity()?; ``
- Fetch latest from Turso:
``rust let episodes = tursoclient .query("SELECT * FROM episodes ORDER BY timestamp DESC LIMIT ?") .bind(maxepisodes_cache) .await?; ``
- Update redb cache:
``rust tokio::task::spawnblocking(move || { let writetxn = redb.beginwrite()?; let mut table = writetxn.opentable(EPISODESTABLE)?; for episode in episodes { table.insert(episode.id.asbytes(), episode.tobytes())?; } write_txn.commit()?; }); ``
- Sync patterns and embeddings if enabled
Configuration
pub struct SyncConfig {
pub max_episodes_cache: usize, // Default: 1000
pub batch_size: usize, // Default: 100
pub sync_patterns: bool, // Default: true
pub sync_embeddings: bool, // Default: true
}
Error Handling
| Error | Handling |
|---|---|
| Turso unavailable | Skip sync, log warning, retry later |
| redb corruption | Attempt repair, rebuild from Turso |
| Partial sync | Track progress, resume from last point |
Performance Tips
- Batch small operations
- Incremental sync (only changes)
- Parallel fetch with Tokio
- Write-ahead preparation
Validation
After sync, verify:
- Episode count matches
- Latest episodes present
- Pattern counts consistent
- No orphaned embeddings
Troubleshooting
| Issue | Solution |
|---|---|
| Slow syncs | Reduce cache size, increase batch |
| redb lock errors | Use dedicated write task |
| Memory pressure | Stream large results, smaller batches |