When a useQuery
query fails (the query function throws an error), TanStack Query will automatically retry the query if that query's request has not reached the max number of consecutive retries (defaults to 3
) or a function is provided to determine if a retry is allowed.
You can configure retries both on a global level and an individual query level.
- Setting
retry = false
will disable retries. - Setting
retry = 6
will retry failing requests 6 times before showing the final error thrown by the function. - Setting
retry = true
will infinitely retry failing requests. - Setting
retry = (failureCount, error) => ...
allows for custom logic based on why the request failed.
On the server, retries default to
0
to make server rendering as fast as possible.
1import { useQuery } from '@tanstack/react-query'
2
3// Make a specific query retry a certain number of times
4const result = useQuery({
5 queryKey: ['todos', 1],
6 queryFn: fetchTodoListPage,
7 retry: 10, // Will retry failed requests 10 times before displaying an error
8})
Info: Contents of the
error
property will be part offailureReason
response property ofuseQuery
until the last retry attempt. So in above example any error contents will be part offailureReason
property for first 9 retry attempts (Overall 10 attempts) and finally they will be part oferror
after last attempt if error persists after all retry attempts.
Retry Delay #
By default, retries in TanStack Query do not happen immediately after a request fails. As is standard, a back-off delay is gradually applied to each retry attempt.
The default retryDelay
is set to double (starting at 1000
ms) with each attempt, but not exceed 30 seconds:
1// Configure for all queries
2import {
3 QueryCache,
4 QueryClient,
5 QueryClientProvider,
6} from '@tanstack/react-query'
7
8const queryClient = new QueryClient({
9 defaultOptions: {
10 queries: {
11 retryDelay: (attemptIndex) => Math.min(1000 * 2 ** attemptIndex, 30000),
12 },
13 },
14})
15
16function App() {
17 return <QueryClientProvider client={queryClient}>...</QueryClientProvider>
18}
Though it is not recommended, you can obviously override the retryDelay
function/integer in both the Provider and individual query options. If set to an integer instead of a function the delay will always be the same amount of time:
1const result = useQuery({
2 queryKey: ['todos'],
3 queryFn: fetchTodoList,
4 retryDelay: 1000, // Will always wait 1000ms to retry, regardless of how many retries
5})