Stop waiting on the network eleven times per page

The app got slower as pages grew, and the reason was not the queries. It
was their number.

A transaction is pinned to one connection, and a connection runs queries
one after another. Every Promise.all in a withUser block looked like
concurrency and was a queue. Measured against the real database: the
round trip is ~36 ms, ten trivial `select 1` over one connection take
343 ms, over ten connections 39 ms. Nothing here is slow — the whole
dashboard payload is under 200 kB, and every table is around a thousand
rows.

More connections is the wrong answer: the RLS session context is per
transaction, so parallel reads mean parallel transactions, and those
multiply the connections the database will grant. Fewer round trips
instead. Postgres will return each sub-select as its own JSON column of
one result.

Per page view, counting the transaction frame:

  shell (paid by every page)  10 → 4
  overview                    14 → 5
  employee file               14 → 7
  employee list                8 → 6

The overview plus its shell went from 24 round trips to 9 — about 860 ms
of pure waiting down to about 320 ms.

The one trap is documented where it bites: inside json_agg, Postgres
formats values itself and the driver's parsers (lib/db/pool.ts) never
see them. Dates, numerics and uuids come out identical; timestamptz does
not — "+00:00" where the driver gives "…Z". Timestamps are compared as
strings in lib/history.ts to decide what happened later, and those two
forms sort against each other wrongly. Every timestamptz in a bundled
query therefore goes through zeitstempel(), which was checked
character-for-character against the driver.

Four loaders moved out of their pages into lib/ so the number of round
trips can be measured without building a React tree, and so the new path
could be held against the old one field by field: same rows, same order,
same strings, for the overview and for four employee files chosen to
differ (with history, a chief, a planned entry, one with dependents).

withUser now counts the queries in each transaction and says so in
development past a threshold. Without that, this grows back: each new
tile brings its own query, and nobody notices until everybody does.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
2026-08-16 19:31:36 +02:00
parent 6957b95a97
commit 91b2b3406b
13 changed files with 753 additions and 336 deletions

View File

@@ -6,14 +6,12 @@ import { DraftsCard } from "@/components/dashboard/DraftsCard";
import { Card, CARD_CLASS, CardTitle } from "@/components/ui/Card";
import { actionBadgeStyle } from "@/lib/colors";
import { istEingeschraenkt, parseArten, parseZeitraum } from "@/lib/dashboard-filter";
import { loadDashboardData } from "@/lib/dashboard-data";
import { addDaysIso, fmtDate, fmtName, todayIso } from "@/lib/format";
import { divisionOf, loadOrgMaps } from "@/lib/org";
import { loadPlacements } from "@/lib/placement";
import { loadOpenPositions } from "@/lib/positions";
import { divisionOf } from "@/lib/org";
import { deriveStatusAsOf } from "@/lib/reports";
import { currentUserId } from "@/lib/auth/session";
import { withUser } from "@/lib/db";
import type { HistoryEventType } from "@/lib/supabase/types";
// Each KPI carries a colour already; the accent bar repeats it in a second
// channel so the tiles are scannable as a row rather than six identical
@@ -63,7 +61,6 @@ export default async function DashboardPage({
const zeitraum = parseZeitraum(params.tage);
const arten = parseArten(params.arten);
const bisIso = addDaysIso(today, zeitraum);
const zeigt = (art: "hire" | "exit" | "return") => arten.includes(art);
const userId = await currentUserId();
@@ -87,114 +84,7 @@ export default async function DashboardPage({
upcomingExits,
upcomingReturns,
history,
} = await withUser(userId, async (tx) => {
const countIn = (types: readonly HistoryEventType[]) =>
tx
.selectFrom("employee_history")
.select(({ fn }) => fn.countAll<string>().as("anzahl"))
.where("event_type", "in", [...types])
.where("event_date", ">=", yearStart)
.where("event_date", "<=", yearEnd)
.executeTakeFirst();
const [
drafts,
staffRows,
hiresYtd,
exitsYtd,
openPositions,
orgMaps,
placements,
upcomingHires,
upcomingExits,
upcomingReturns,
history,
] = await Promise.all([
userId
? tx
.selectFrom("hire_drafts")
.select(["id", "step", "payload", "updated_at"])
.where("created_by", "=", userId)
.orderBy("updated_at", "desc")
.execute()
: Promise.resolve([]),
tx
.selectFrom("employees")
.select(["id", "weekly_hours", "entry_date", "exit_date", "karenz_start_date", "karenz_return_date"])
.orderBy("id")
.execute(),
// Entries/exits count history events, which is what the linked report
// counts too. `entry_date` would also sweep up rehires, whose event is
// logged as 'Wiedereintritt' — the tile and its destination then showed
// different numbers for the same year.
countIn(["Eintritt", "Wiedereintritt"]),
countIn(["Austritt"]),
loadOpenPositions(tx),
loadOrgMaps(tx),
loadPlacements(tx, { asOf: today }),
// Abgewählte Arten werden gar nicht erst gelesen — die Karte zeigt sie
// ohnehin nicht, und eine Abfrage, deren Ergebnis niemand ansieht, ist
// eine Abfrage zu viel.
zeigt("hire")
? tx
.selectFrom("employees")
.select(["id", "first_name", "last_name", "entry_date"])
.where("status", "=", "Geplant")
.where("entry_date", ">=", today)
.where("entry_date", "<=", bisIso)
.execute()
: Promise.resolve([]),
zeigt("exit")
? tx
.selectFrom("employees")
.select(["id", "first_name", "last_name", "exit_date"])
.where("exit_date", "is not", null)
.where("exit_date", ">=", today)
.where("exit_date", "<=", bisIso)
.execute()
: Promise.resolve([]),
zeigt("return")
? tx
.selectFrom("employees")
.select(["id", "first_name", "last_name", "karenz_return_date"])
.where("status", "=", "Karenz")
.where("karenz_return_date", "is not", null)
.where("karenz_return_date", ">=", today)
.where("karenz_return_date", "<=", bisIso)
.execute()
: Promise.resolve([]),
tx
.selectFrom("employee_history as h")
.leftJoin("employees as e", "e.id", "h.employee_id")
.select(["h.id", "h.employee_id", "h.event_date", "h.event_type", "h.description", "e.first_name", "e.last_name"])
.orderBy("h.event_date", "desc")
.orderBy("h.created_at", "desc")
.limit(10)
.execute(),
]);
return {
drafts,
staffRows,
hiresYtd: Number(hiresYtd?.anzahl ?? 0),
exitsYtd: Number(exitsYtd?.anzahl ?? 0),
openPositions,
orgMaps,
placements,
upcomingHires,
upcomingExits,
upcomingReturns,
history,
};
});
} = await withUser(userId, (tx) => loadDashboardData(tx, { userId, today, yearStart, yearEnd, bisIso, arten }));
// "Aktiv" means status Aktiv — somebody on Karenz is employed but not
// active, and is counted by its own tile instead. FTE follows the same
// set: Karenz contributes no capacity, so including it would overstate