Manage Server Dataadopt Rtk Query

Core Patterns

Keep one API slice per base URL and extend it

import { createApi, fetchBaseQuery } from '@reduxjs/toolkit/query/react'

export const api = createApi({
  reducerPath: 'api',
  baseQuery: fetchBaseQuery({ baseUrl: '/api/' }),
  endpoints: () => ({}),
})

export const postsApi = api.injectEndpoints({
  endpoints: (build) => ({
    getPosts: build.query<{ id: string; title: string }[], void>({
      query: () => 'posts',
    }),
  }),
})

Split files with injectEndpoints, not by making multiple createApi roots for the same backend.

Use tags for cache invalidation

type Post = { id: string; title: string }

export const api = createApi({
  reducerPath: 'api',
  baseQuery: fetchBaseQuery({ baseUrl: '/api/' }),
  tagTypes: ['Post'],
  endpoints: (build) => ({
    getPosts: build.query<Post[], void>({
      query: () => 'posts',
      providesTags: (result) =>
        result
          ? [...result.map(({ id }) => ({ type: 'Post' as const, id })), 'Post']
          : ['Post'],
    }),
    updatePost: build.mutation<Post, Pick<Post, 'id' | 'title'>>({
      query: ({ id, title }) => ({
        url: `posts/${id}`,
        method: 'PATCH',
        body: { title },
      }),
      invalidatesTags: (_result, _error, { id }) => [{ type: 'Post', id }],
    }),
  }),
})

Treat tags as the normal invalidation path before reaching for manual cache patching.

Do optimistic updates in endpoint lifecycles

type Post = { id: string; title: string }

export const api = createApi({
  reducerPath: 'api',
  baseQuery: fetchBaseQuery({ baseUrl: '/api/' }),
  tagTypes: ['Post'],
  endpoints: (build) => ({
    getPosts: build.query<Post[], void>({
      query: () => 'posts',
      providesTags: ['Post'],
    }),
    updatePostTitle: build.mutation<Post, Pick<Post, 'id' | 'title'>>({
      query: ({ id, title }) => ({
        url: `posts/${id}`,
        method: 'PATCH',
        body: { title },
      }),
      async onQueryStarted({ id, title }, { dispatch, queryFulfilled }) {
        const patch = dispatch(
          api.util.updateQueryData('getPosts', undefined, (draft) => {
            const post = draft.find((item) => item.id === id)
            if (post) {
              post.title = title
            }
          }),
        )

        try {
          await queryFulfilled
        } catch {
          patch.undo()
        }
      },
    }),
  }),
})

Keep optimistic and pessimistic cache updates inside endpoint lifecycle handlers so they stay coupled to the request.

Common Mistakes

CRITICAL Creating multiple API slices for one backend

Wrong:

import { createApi, fetchBaseQuery } from '@reduxjs/toolkit/query/react'

type User = { id: string; name: string }

const baseQuery = fetchBaseQuery({ baseUrl: '/api/' })

const postsApi = createApi({
  reducerPath: 'api',
  baseQuery,
  endpoints: () => ({}),
})

const usersApi = createApi({
  reducerPath: 'api',
  baseQuery,
  endpoints: () => ({}),
})

Correct:

import { createApi, fetchBaseQuery } from '@reduxjs/toolkit/query/react'

type User = { id: string; name: string }

const baseQuery = fetchBaseQuery({ baseUrl: '/api/' })

const api = createApi({
  reducerPath: 'api',
  baseQuery,
  endpoints: () => ({}),
})

const usersApi = api.injectEndpoints({
  endpoints: (build) => ({
    getUsers: build.query<User[], void>({ query: () => 'users' }),
  }),
})

One API slice per base URL preserves invalidation behavior and avoids duplicated middleware work.

Source: reduxjs/redux-toolkit:docs/rtk-query/api/createApi.mdx

HIGH Forgetting api.reducer or api.middleware

Wrong:

import { configureStore } from '@reduxjs/toolkit'

const store = configureStore({
  reducer: {},
})

Correct:

import { configureStore } from '@reduxjs/toolkit'

const store = configureStore({
  reducer: {
    [api.reducerPath]: api.reducer,
  },
  middleware: (getDefaultMiddleware) =>
    getDefaultMiddleware().concat(api.middleware),
})

RTK Query hooks need both the reducer and middleware to manage cache state and request lifecycles.

Source: reduxjs/redux-toolkit:docs/tutorials/rtk-query.mdx

MEDIUM Persisting browser API cache by default

Wrong:

const storage = window.localStorage

const persistConfig = {
  key: 'root',
  storage,
}

Correct:

import { createApi, fetchBaseQuery } from '@reduxjs/toolkit/query/react'

const api = createApi({
  reducerPath: 'api',
  baseQuery: fetchBaseQuery({ baseUrl: '/api/' }),
  endpoints: () => ({}),
})

Persisting RTK Query cache in browsers often keeps stale data around longer than users expect; treat persistence as a special case, not the default.

Source: reduxjs/redux-toolkit:docs/rtk-query/usage/persistence-and-rehydration.mdx

HIGH Patching cache from components

Wrong:

import { useEffect } from 'react'
import { useAppDispatch } from '../../app/hooks'

const dispatch = useAppDispatch()

useEffect(() => {
  dispatch(api.util.updateQueryData('getPosts', undefined, (draft) => {
    draft.push({ id: 'p3', title: 'Patched from component' })
  }))
}, [dispatch])

Correct:

updatePostTitle: build.mutation<Post, Pick<Post, 'id' | 'title'>>({
  query: ({ id, title }) => ({
    url: `posts/${id}`,
    method: 'PATCH',
    body: { title },
  }),
  async onQueryStarted({ id, title }, { dispatch, queryFulfilled }) {
    const patch = dispatch(
      api.util.updateQueryData('getPosts', undefined, (draft) => {
        const post = draft.find((item) => item.id === id)
        if (post) {
          post.title = title
        }
      }),
    )

    try {
      await queryFulfilled
    } catch {
      patch.undo()
    }
  },
})

Component-level cache patches drift away from the mutation lifecycle that should own them.

Source: reduxjs/redux-toolkit:docs/rtk-query/usage/manual-cache-updates.mdx

HIGH Expecting invalidation to refetch unsubscribed queries

Wrong:

import { api } from './api'
import { store } from './store'

const subscription = store.dispatch(api.endpoints.getPosts.initiate())
subscription.unsubscribe()
store.dispatch(api.util.invalidateTags(['Post']))

Correct:

import { api } from './api'
import { store } from './store'

store.dispatch(api.endpoints.getPosts.initiate())
store.dispatch(api.util.invalidateTags(['Post']))

Invalidation only refetches actively subscribed queries; if no component is using that cache entry, RTK Query drops it and fetches again next time it is needed.

Source: reduxjs/redux-toolkit:docs/rtk-query/usage/automated-refetching.mdx

References

• Endpoint lifecycle details and cache tradeoffs