Caching

Two read endpoints are Redis-backed; everything else hits Postgres directly. Cache invalidation is structurally enforced via shared.TxRunner.Run — every write declares a CacheScope at the call site, and the post-commit Invalidator.Apply busts exactly the right keys.

Cached endpoints

Endpoint	Key pattern	TTL	Invalidator scope
`GET /portfolios/{id}`	`portfolio:detail:{id}`	30s	Detail
`GET /portfolios/{id}/summary`	(same — `FindSummary` delegates to `FindDetail`)	30s	Detail
`GET /portfolios/{id}/valuation?range=R`	`portfolio:valuation:{id}:{R}`	60s for `1D`, 5min for daily ranges	Valuation (busts all 8 ranges)

R ∈ {1D, 1W, 1M, 3M, 6M, 1Y, YTD, ALL}. There are 8 separate cache keys per portfolio for valuation — one per range — and they are always invalidated as a group.

Cache scopes

Defined in internal/modules/portfolio/shared/cache.go:

type CacheScope int

const (
    CacheScopeNone CacheScope = iota          // no bust — for reads
    CacheScopeDetail                          // bust portfolio:detail:{id}
    CacheScopeDetailAndValuation              // bust detail + all valuation ranges
)

Which scope each write uses

Path	Scope	Reason
Create	None	New portfolio has no cache yet
Update (name, description, isDefault)	Detail	Header changes affect bundled detail
Delete	DetailAndValuation	All derived numbers go to zero
Transaction Add / Update / Delete	DetailAndValuation	Source data changed; all derivations stale

Pagination snapshot

FindAll (portfolios) and Find (transactions) wrap their rows-query and count-query in one read-only REPEATABLE READ transaction via TxRunner.RunReadSnapshot. A concurrent insert/delete between the two queries would otherwise yield an inconsistent (rows, count) pair. No Redis caching here — Postgres consistency is the guarantee.

Cache miss path

The FindDetail flow short-circuits on cache hit:

if found := cache.GetJSON(detailKey, &snapshot); found {
    return snapshotToDetail(snapshot)
}
// otherwise: read transactions + prices, BuildState, cache.SetJSON, return

Cache misses never block writes — slog.Warn on cache read/write failure and fall through to Postgres.