magic-link verify deadlocks: runWithTransaction nests adapter.transaction (1.6.11…
- User
magic-link verify deadlocks: runWithTransaction nests adapter.transaction (1.6.11 consumeVerificationValue) instead of reusing the active transaction
Is this suited for github?
- Yes, this is suited for github
Reproduction
- better-auth 1.6.11 with the
magicLinkplugin. - A database adapter whose
transactionis a real interactive DB transaction that holds a dedicated connection for its duration (e.g. the Kysely adapter, or@zenstackhq/better-auth, backed bypg). - Limit the client connection pool to one connection:
new Pool({ max: 1 })-- standard when using Cloudflare Hyperdrive, which pools server-side so the worker keeps a tiny client pool. await auth.api.signInMagicLink({ body: { email } })-> works (single insert).await auth.api.magicLinkVerify({ query: { token } })(orGET /api/auth/magic-link/verify?token=...).
Minimal trigger: any setup where
adapter.transactionacquires a connection and the pool cannot hand out two connections at once.Current vs. Expected behavior
Current: the verify call hangs forever and never returns. Postgres shows the backend stuck
idle in transaction(wait_event = ClientRead) right after the verificationSELECT; the connection is never released, so every attempt leaks one idle-in-transaction connection.Trace of the cause:
internalAdapter.consumeVerificationValue(added in 1.6.11) runs insiderunWithTransaction-> opens transaction tx1 (BEGIN, takes the only pooled connection).- Consuming the row (
consumeOne) triggers anotherrunWithTransaction-> callsadapter.transactionagain -> transaction tx2, which tries to acquire a second connection. runWithTransactionin@better-auth/core(context/transaction.mjs) always opens a freshadapter.transaction; it does NOT checkadapterAsyncStorage.getStore()to see whether a transaction is already active in the current async context. So tx2 is nested inside tx1 and blocks forever waiting for the connection tx1 holds -> deadlock.
Note the CRUD path already reuses the active transaction:
getCurrentAdapter(adapter)returns thetrxadapter from the ALS store.runWithTransactionis inconsistent with this -- it doesn't reuse.Expected:
runWithTransactionshould be reentrant -- when an adapter/transaction is already active in the async-context store, runfnwithin it instead of opening a nestedadapter.transaction(mirroringgetCurrentAdapter). Then verify needs only one connection and works on any pool size. (Alternatively/additionally, document that adapters MUST support reentrant nested transactions -- currently adapters like@zenstackhq/better-authopen a brand-new transaction on the base client for eachadapter.transactioncall, which is a reasonable contract but incompatible with the current nesting.)This is security-relevant: the 1.6.11 change that introduced this nesting was the magic-link "mint multiple sessions from one token" race fix, so the fix effectively breaks magic-link verification on connection-limited adapters/pools.
What version of Better Auth are you using?
1.6.11
System info
> The `auth info` CLI can't introspect this project headlessly because the auth instance is built dynamically via a factory (no static `auth.ts` export), so the values below are assembled manually { "better-auth": { "version": "1.6.11", "@better-auth/core": "1.6.11" }, "adapter": { "name": "@zenstackhq/better-auth", "version": "3.6.4", "orm": "@zenstackhq/orm 3.6.4 (Kysely 0.28.17)" }, "database": { "driver": "pg 8.20.0", "pool": { "max": 1 }, "fronted_by": "Cloudflare Hyperdrive" }, "runtime": { "platform": "Cloudflare Workers (workerd)", "wrangler": "4.95.0", "node_for_tooling": "v22.22.1" }, "plugins": ["magicLink"], "os": "macOS (Darwin 25.5.0)" }Which area(s) are affected? (Select all that apply)
Backend
Auth config (if applicable)
import { betterAuth } from "better-auth" export const auth = betterAuth({ emailAndPassword: { enabled: true }, });Additional context
- Bisected away unrelated suspects: the hang reproduces on older framework code and is independent of DI scope changes -- it's purely
1.6.10 -> 1.6.11. - Reproduced deterministically in a real
workerdintegration test (request magic link -> capture token -> verify) that times out before the change. - Relevant code:
@better-auth/corecontext/transaction.mjsrunWithTransaction(opensadapter.transactionunconditionally) vsgetCurrentAdapter(reuses the activetrx);internalAdapter.consumeVerificationValue->consumeOneWithHooks-> adapterconsumeOne. - Worked around downstream by making the adapter's transaction reentrant (reuse the in-flight transaction via
AsyncLocalStorage), which is also the correct semantics (nested transaction = one atomic unit) -- but it seems likerunWithTransactionitself should reuse the active transaction.
Interface
No new interfaces are introduced.
- Assistant
I’ll trace the transaction context and existing tests, add a regression that exposes the nested call, then patch the smallest core implementation and run focused tests plus typecheck.
- Assistant
rgisn’t installed in this workspace, so I’m using targetedfind/grepsearches. The repository guide matches the supplied instructions; there are no extra local constraints. - Assistant
The key distinction is that every auth request already has an adapter context via
runWithAdapter; treating any store as a transaction would silently disable transactions. I’m modeling transaction state explicitly so only nested transactions reuse the active adapter, while the first transaction still opens normally. - Assistant
The regression file is in place. The first direct root-level Vitest invocation hit the package’s self-reference export resolution before collecting tests, so I’m switching to the core package’s configured test command to get a meaningful red/green result.
- Assistant
The focused regression now passes: nested
runWithTransactioncalls invoke the adapter transaction exactly once and both levels observe the same transaction adapter. I’m checking the broader transaction semantics and repository typecheck next. - Assistant
Typecheck is clean and the regression passes. I’m doing one final focused pass over the neighboring context tests and the magic-link verification suite to catch integration-level behavior changes.
- Assistant
The core checks are green. The magic-link suite itself couldn’t collect because its local Kysely workspace dependency has no built entry yet; I’m building that package locally and retrying rather than treating the resolver failure as a product failure.
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.
Identical first prompt
- composer-2.5Compare side by side →
- composer-2.5Compare side by side →
- composer-2.5Compare side by side →
- composer-2.5Compare side by side →
- composer-2.5Compare side by side →
- gpt-5.6-solCompare side by side →
- gpt-5.6-solCompare side by side →
- gpt-5.6-solCompare side by side →
- gpt-5.6-solCompare side by side →
- fable-5Compare side by side →
- fable-5Compare side by side →