github-starred/better-auth
2
0
Fork 0
mirror of https://github.com/better-auth/better-auth.git synced 2026-05-23 15:42:09 -05:00
Code Issues 2.5k Packages Projects Releases 112 Wiki Activity

[GH-ISSUE #8420] Session is set to null on network reconnect (mobile) #28408

New Issue
Closed
opened 2026-04-17 19:51:59 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-17 19:51:59 -05:00
Owner
Copy Link

Originally created by @florian-deroo on GitHub (Mar 5, 2026).
Original GitHub issue: https://github.com/better-auth/better-auth/issues/8420

Originally assigned to: @bytaesu on GitHub.

Is this suited for github?

  • Yes, this is suited for github

To Reproduce

  1. Authenticate successfully on a mobile device (React Native / Expo)
  2. Turn off internet (WiFi + cellular)
  3. Navigate around the app — session is intact, useSession still returns data ✓
  4. Turn internet back on
  5. → useSession immediately sets data: null → app redirects to login screen ✗

Bonus step to confirm the token is still valid:
6. Force-quit and reopen the app → session is automatically restored ✓
(proves the token in SecureStore was never invalidated server-side)

Current vs. Expected behavior

Current behavior:

When a mobile device loses internet and reconnects, useSession sets data: null and triggers a logout — even though the session token in SecureStore is still perfectly valid.

Expected behavior:

A transient network failure during session revalidation should preserve the previously cached session data. Only an explicit 401 Unauthorized response from the server should invalidate the session and trigger a logout.

What version of Better Auth are you using?

1.4.16

System info

{
  "system": {
    "platform": "darwin",
    "arch": "arm64",
    "version": "Darwin Kernel Version 25.3.0: Wed Jan 28 20:49:24 PST 2026; root:xnu-12377.81.4~5/RELEASE_ARM64_T8132",
    "release": "25.3.0",
    "cpuCount": 10,
    "cpuModel": "Apple M4",
    "totalMemory": "24.00 GB",
    "freeMemory": "0.42 GB"
  },
  "node": {
    "version": "v22.22.0",
    "env": "development"
  },
  "packageManager": {
    "name": "npm",
    "version": "10.9.4"
  },
  "frameworks": [
    {
      "name": "react",
      "version": "19.1.1"
    }
  ],
  "databases": null,
  "betterAuth": {
    "version": "catalog:",
    "config": null
  }
}

Which area(s) are affected? (Select all that apply)

Client

Auth config (if applicable)

import { createAuthClient } from "better-auth/react";
import { expoClient } from "@better-auth/expo/client";
import * as SecureStore from "expo-secure-store";
import {
	adminClient,
	inferAdditionalFields,
	oneTimeTokenClient,
	usernameClient,
	emailOTPClient,
	phoneNumberClient,
	twoFactorClient,
	anonymousClient,
} from "better-auth/client/plugins";

export const authClient = createAuthClient({
	baseURL: process.env.EXPO_PUBLIC_BASE_URL,
	disableDefaultFetchPlugins: true,
	plugins: [
		expoClient({
			scheme: "test",
			storagePrefix: "test",
			storage: SecureStore,
		}),
		anonymousClient(),
		usernameClient(),
		emailOTPClient(),
		phoneNumberClient(),
		oneTimeTokenClient(),
		twoFactorClient(),
	],
});

Additional context

Root cause

After digging into the source, I found the issue in node_modules/better-auth/dist/client/query.mjs:

async onError(context) {
  const retryAttempts = typeof request.retry === "number" ? request.retry : request.retry?.attempts;
  const retryAttempt = request.retryAttempt || 0;

  if (retryAttempts && retryAttempt < retryAttempts) return;

  // No network error check — data is unconditionally set to null on any failure
  value.set({
    error: context.error,
    data: null, // <-- session wiped on network error
    isPending: false,
    isRefetching: false,
    refetch: value.value.refetch
  });

  await opts?.onError?.(context);
},
Originally created by @florian-deroo on GitHub (Mar 5, 2026). Original GitHub issue: https://github.com/better-auth/better-auth/issues/8420 Originally assigned to: @bytaesu on GitHub. ### Is this suited for github? - [x] Yes, this is suited for github ### To Reproduce 1. Authenticate successfully on a mobile device (React Native / Expo) 2. Turn off internet (WiFi + cellular) 3. Navigate around the app — session is intact, useSession still returns data ✓ 4. Turn internet back on 5. → useSession immediately sets data: null → app redirects to login screen ✗ Bonus step to confirm the token is still valid: 6. Force-quit and reopen the app → session is automatically restored ✓ (proves the token in SecureStore was never invalidated server-side) ### Current vs. Expected behavior **Current behavior:** When a mobile device loses internet and reconnects, useSession sets data: null and triggers a logout — even though the session token in SecureStore is still perfectly valid. **Expected behavior:** A transient network failure during session revalidation should preserve the previously cached session data. Only an explicit 401 Unauthorized response from the server should invalidate the session and trigger a logout. ### What version of Better Auth are you using? 1.4.16 ### System info ```bash { "system": { "platform": "darwin", "arch": "arm64", "version": "Darwin Kernel Version 25.3.0: Wed Jan 28 20:49:24 PST 2026; root:xnu-12377.81.4~5/RELEASE_ARM64_T8132", "release": "25.3.0", "cpuCount": 10, "cpuModel": "Apple M4", "totalMemory": "24.00 GB", "freeMemory": "0.42 GB" }, "node": { "version": "v22.22.0", "env": "development" }, "packageManager": { "name": "npm", "version": "10.9.4" }, "frameworks": [ { "name": "react", "version": "19.1.1" } ], "databases": null, "betterAuth": { "version": "catalog:", "config": null } } ``` ### Which area(s) are affected? (Select all that apply) Client ### Auth config (if applicable) ```typescript import { createAuthClient } from "better-auth/react"; import { expoClient } from "@better-auth/expo/client"; import * as SecureStore from "expo-secure-store"; import { adminClient, inferAdditionalFields, oneTimeTokenClient, usernameClient, emailOTPClient, phoneNumberClient, twoFactorClient, anonymousClient, } from "better-auth/client/plugins"; export const authClient = createAuthClient({ baseURL: process.env.EXPO_PUBLIC_BASE_URL, disableDefaultFetchPlugins: true, plugins: [ expoClient({ scheme: "test", storagePrefix: "test", storage: SecureStore, }), anonymousClient(), usernameClient(), emailOTPClient(), phoneNumberClient(), oneTimeTokenClient(), twoFactorClient(), ], }); ``` ### Additional context ### Root cause After digging into the source, I found the issue in `node_modules/better-auth/dist/client/query.mjs`: ```js async onError(context) { const retryAttempts = typeof request.retry === "number" ? request.retry : request.retry?.attempts; const retryAttempt = request.retryAttempt || 0; if (retryAttempts && retryAttempt < retryAttempts) return; // No network error check — data is unconditionally set to null on any failure value.set({ error: context.error, data: null, // <-- session wiped on network error isPending: false, isRefetching: false, refetch: value.value.refetch }); await opts?.onError?.(context); }, ```
GiteaMirror added the lockedbug labels 2026-04-17 19:51:59 -05:00
GiteaMirror commented 2026-04-17 19:52:01 -05:00
Author
Owner
Copy Link

@dosubot[bot] commented on GitHub (Mar 5, 2026):

Hi @florian-deroo! I'm Dosu and I’m helping the better-auth team.

Your diagnosis is spot on. This is a known issue in Better Auth — the onError handler in client/query.ts unconditionally sets data: null after retries are exhausted, without distinguishing between network errors and actual auth failures (401s).

There's actually an existing issue (#987) documenting similar behavior — network failures like TypeError: Failed to fetch don't trigger retries at all (Better Fetch only retries HTTP response errors, not fetch exceptions), causing sessions to appear null.

Workaround: You could wrap your auth client calls with a custom fetch that has exponential backoff for network errors, as suggested in the issue comments. Another option is to handle the onError callback in your session usage to preserve the previous session data when the error is a network failure rather than a 401.

To reply, just mention @dosu.

How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other

Chat with Dosu Open in Cursor Join Discord Share on X

<!-- gh-comment-id:4006468983 --> @dosubot[bot] commented on GitHub (Mar 5, 2026): <!-- Greeting --> Hi @florian-deroo! I'm [Dosu](https://go.dosu.dev/dosubot) and I’m helping the better-auth team. <!-- Answer --> Your diagnosis is spot on. This is a known issue in Better Auth — the [`onError` handler in `client/query.ts`](https://github.com/better-auth/better-auth/blob/c15f6d9c0addd4218077f38c13be51cc17b69b42/packages/better-auth/src/client/query.ts#L74-L90) unconditionally sets `data: null` after retries are exhausted, without distinguishing between network errors and actual auth failures (401s). There's actually an [existing issue (#987)](https://github.com/better-auth/better-auth/issues/987) documenting similar behavior — network failures like `TypeError: Failed to fetch` don't trigger retries at all (Better Fetch only retries HTTP response errors, not fetch exceptions), causing sessions to appear null. **Workaround**: You could wrap your auth client calls with a custom fetch that has exponential backoff for network errors, as [suggested in the issue comments](https://github.com/better-auth/better-auth/issues/987#issuecomment-2558400053). Another option is to handle the `onError` callback in your session usage to preserve the previous session data when the error is a network failure rather than a 401. <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot).* --- <sup>How did I do? [Good](https://app.dosu.dev/response-feedback/26406ecd-1d24-467a-8350-13eb28e0b594?feedback_type=great_response) | [Irrelevant](https://app.dosu.dev/response-feedback/26406ecd-1d24-467a-8350-13eb28e0b594?feedback_type=irrelevant_answer) | [Incorrect](https://app.dosu.dev/response-feedback/26406ecd-1d24-467a-8350-13eb28e0b594?feedback_type=incorrect_sources) | [Verbose](https://app.dosu.dev/response-feedback/26406ecd-1d24-467a-8350-13eb28e0b594?feedback_type=too_verbose) | [Hallucination](https://app.dosu.dev/response-feedback/26406ecd-1d24-467a-8350-13eb28e0b594?feedback_type=hallucination) | [Report 🐛](https://app.dosu.dev/response-feedback/26406ecd-1d24-467a-8350-13eb28e0b594?feedback_type=bug_report) | [Other](https://app.dosu.dev/response-feedback/26406ecd-1d24-467a-8350-13eb28e0b594?feedback_type=other)</sup> [![Chat with Dosu](https://dosu.dev/dosu-chat-badge.svg)](https://app.dosu.dev/cdda13d9-dd27-4d31-b09a-5d8bec92de21/ask?utm_source=github)&nbsp;[![Open in Cursor](https://dosu.dev/dosu-in-cursor.svg)](https://cursor.com/link/prompt?text=Your%20diagnosis%20is%20spot%20on.%20This%20is%20a%20known%20issue%20in%20Better%20Auth%20%E2%80%94%20the%20%5B%60onError%60%20handler%20in%20%60client/query.ts%60%5D%28https%3A//github.com/better-auth/better-auth/blob/c15f6d9c0addd4218077f38c13be51cc17b69b42/packages/better-auth/src/client/query.ts%23L74-L90%29%20unconditionally%20sets%20%60data%3A%20null%60%20after%20retries%20are%20exhausted%2C%20without%20distinguishing%20between%20network%20errors%20and%20actual%20auth%20failures%20%28401s%29.%0A%0AThere%27s%20actually%20an%20%5Bexisting%20issue%20%28%23987%29%5D%28https%3A//github.com/better-auth/better-auth/issues/987%29%20documenting%20similar%20behavior%20%E2%80%94%20network%20failures%20like%20%60TypeError%3A%20Failed%20to%20fetch%60%20don%27t%20trigger%20retries%20at%20all%20%28Better%20Fetch%20only%20retries%20HTTP%20response%20errors%2C%20not%20fetch%20exceptions%29%2C%20causing%20sessions%20to%20appear%20null.%0A%0A%2A%2AWorkaround%2A%2A%3A%20You%20could%20wrap%20your%20auth%20client%20calls%20with%20a%20custom%20fetch%20that%20has%20exponential%20backoff%20for%20network%20errors%2C%20as%20%5Bsuggested%20in%20the%20issue%20comments%5D%28https%3A//github.com/better-auth/better-auth/issues/987%23issuecomment-2558400053%29.%20Another%20option%20is%20to%20handle%20the%20%60onError%60%20callback%20in%20your%20session%20usage%20to%20preserve%20the%20previous%20session%20data%20when%20the%20error%20is%20a%20network%20failure%20rather%20than%20a%20401.)&nbsp;[![Join Discord](https://img.shields.io/badge/join-5865F2?logo=discord&logoColor=white&label=)](https://go.dosu.dev/discord-bot)&nbsp;[![Share on X](https://img.shields.io/badge/X-share-black)](https://twitter.com/intent/tweet?text=%40dosu_ai%20helped%20me%20solve%20this%20issue!&url=https%3A//github.com/better-auth/better-auth/issues/8420)
GiteaMirror commented 2026-04-17 19:52:01 -05:00
Author
Owner
Copy Link

@bytaesu commented on GitHub (Mar 6, 2026):

Hi @florian-deroo,

I tested this with an actual Expo app and made a fix!

However, since the change exists outside the Expo plugin, I'll review it again with the team before applying it.

<!-- gh-comment-id:4009174428 --> @bytaesu commented on GitHub (Mar 6, 2026): Hi @florian-deroo, I tested this with an actual Expo app and made a fix! However, since the change exists outside the Expo plugin, I'll review it again with the team before applying it.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
adapter
astro
awaiting external contributor
blocked
breaking
breaking change
bug
c-devops
core
credentials
database
dependencies
devops
devtools
docs
documentation
duplicate
elysia
enhancement
enterprise
expo
express
fastify
good first issue
help wanted
hono
identity
infra
integration
invalid
javascript
locked
maintenance
need-more-information
needs: info
needs: repro
nextjs
nuxt
oauth
organization
P0
payments
perf
platform
plugin
pull-request
Mirrored from GitHub Pull Request
 question
ready
regression
remix
security
social-provider
solid
stale
svelte
tanstack-start
tracking
version-bump
vue
wontfix
No Label bug locked
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/better-auth#28408
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?