* async create conn
* update default values and testcase
* fix comments
* fix data race
* remove context.WithoutCancel, which is a function introduced in Go 1.21
* fix TestDialerRetryConfiguration/DefaultDialerRetries, because tryDial are likely done in async flow
* change to share failed to delivery connection to other waiting
* remove chinese comment
* fix: optimize WantConnQueue benchmarks to prevent memory exhaustion
- Fix BenchmarkWantConnQueue_Dequeue timeout issue by limiting pre-population
- Use object pooling in BenchmarkWantConnQueue_Enqueue to reduce allocations
- Optimize BenchmarkWantConnQueue_EnqueueDequeue with reusable wantConn pool
- Prevent GitHub Actions benchmark failures due to excessive memory usage
Before: BenchmarkWantConnQueue_Dequeue ran for 11+ minutes and was killed
After: All benchmarks complete in ~8 seconds with consistent performance
* format
* fix turn leaks
---------
Co-authored-by: Nedyalko Dyakov <1547186+ndyakov@users.noreply.github.com>
Co-authored-by: Hristo Temelski <hristo.temelski@redis.com>
* internal/proto/peek_push_notification_test : Refactor test helpers to use fmt.Fprintf for buffers
Replaced buf.WriteString(fmt.Sprintf(...)) with fmt.Fprintf or fmt.Fprint in test helper functions for improved clarity and efficiency. This change affects push notification and RESP3 test utilities.
* peek_push_notification_test: revert prev formatting
* all: replace buf.WriteString with fmt.FprintF for consistency
---------
Co-authored-by: Nedyalko Dyakov <1547186+ndyakov@users.noreply.github.com>
* perf(pool): replace hookManager RWMutex with atomic.Pointer and add predefined state slices
- Replace hookManager RWMutex with atomic.Pointer for lock-free reads in hot paths
- Add predefined state slices to avoid allocations (validFromInUse, validFromCreatedOrIdle, etc.)
- Add Clone() method to PoolHookManager for atomic updates
- Update AddPoolHook/RemovePoolHook to use copy-on-write pattern
- Update all hookManager access points to use atomic Load()
Performance improvements:
- Eliminates RWMutex contention in Get/Put/Remove hot paths
- Reduces allocations by reusing predefined state slices
- Lock-free reads allow better CPU cache utilization
* perf(pool): eliminate mutex overhead in state machine hot path
The state machine was calling notifyWaiters() on EVERY Get/Put operation,
which acquired a mutex even when no waiters were present (the common case).
Fix: Use atomic waiterCount to check for waiters BEFORE acquiring mutex.
This eliminates mutex contention in the hot path (Get/Put operations).
Implementation:
- Added atomic.Int32 waiterCount field to ConnStateMachine
- Increment when adding waiter, decrement when removing
- Check waiterCount atomically before acquiring mutex in notifyWaiters()
Performance impact:
- Before: mutex lock/unlock on every Get/Put (even with no waiters)
- After: lock-free atomic check, only acquire mutex if waiters exist
- Expected improvement: ~30-50% for Get/Put operations
* perf(pool): use predefined state slices to eliminate allocations in hot path
The pool was creating new slice literals on EVERY Get/Put operation:
- popIdle(): []ConnState{StateCreated, StateIdle}
- putConn(): []ConnState{StateInUse}
- CompareAndSwapUsed(): []ConnState{StateIdle} and []ConnState{StateInUse}
- MarkUnusableForHandoff(): []ConnState{StateInUse, StateIdle, StateCreated}
These allocations were happening millions of times per second in the hot path.
Fix: Use predefined global slices defined in conn_state.go:
- validFromInUse
- validFromCreatedOrIdle
- validFromCreatedInUseOrIdle
Performance impact:
- Before: 4 slice allocations per Get/Put cycle
- After: 0 allocations (use predefined slices)
- Expected improvement: ~30-40% reduction in allocations and GC pressure
* perf(pool): optimize TryTransition to reduce atomic operations
Further optimize the hot path by:
1. Remove redundant GetState() call in the loop
2. Only check waiterCount after successful CAS (not before loop)
3. Inline the waiterCount check to avoid notifyWaiters() call overhead
This reduces atomic operations from 4-5 per Get/Put to 2-3:
- Before: GetState() + CAS + waiterCount.Load() + notifyWaiters mutex check
- After: CAS + waiterCount.Load() (only if CAS succeeds)
Performance impact:
- Eliminates 1-2 atomic operations per Get/Put
- Expected improvement: ~10-15% for Get/Put operations
* perf(pool): add fast path for Get/Put to match master performance
Introduced TryTransitionFast() for the hot path (Get/Put operations):
- Single CAS operation (same as master's atomic bool)
- No waiter notification overhead
- No loop through valid states
- No error allocation
Hot path flow:
1. popIdle(): Try IDLE → IN_USE (fast), fallback to CREATED → IN_USE
2. putConn(): Try IN_USE → IDLE (fast)
This matches master's performance while preserving state machine for:
- Background operations (handoff/reauth use UNUSABLE state)
- State validation (TryTransition still available)
- Waiter notification (AwaitAndTransition for blocking)
Performance comparison per Get/Put cycle:
- Master: 2 atomic CAS operations
- State machine (before): 5 atomic operations (2.5x slower)
- State machine (after): 2 atomic CAS operations (same as master!)
Expected improvement: Restore to baseline ~11,373 ops/sec
* combine cas
* fix linter
* try faster approach
* fast semaphore
* better inlining for hot path
* fix linter issues
* use new semaphore in auth as well
* linter should be happy now
* add comments
* Update internal/pool/conn_state.go
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* address comment
* slight reordering
* try to cache time if for non-critical calculation
* fix wrong benchmark
* add concurrent test
* fix benchmark report
* add additional expect to check output
* comment and variable rename
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* fix(pool): wip, pool reauth should not interfere with handoff
* fix credListeners map
* fix race in tests
* better conn usable timeout
* add design decision comment
* few small improvements
* update marked as queued
* add Used to clarify the state of the conn
* rename test
* fix(test): fix flaky test
* lock inside the listeners collection
* address pr comments
* Update internal/auth/cred_listeners.go
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Update internal/pool/buffer_size_test.go
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* wip refactor entraid
* fix maintnotif pool hook
* fix mocks
* fix nil listener
* sync and async reauth based on conn lifecycle
* be able to reject connection OnGet
* pass hooks so the tests can observe reauth
* give some time for the background to execute commands
* fix tests
* only async reauth
* Update internal/pool/pool.go
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Update internal/auth/streaming/pool_hook.go
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Update internal/pool/conn.go
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* chore(redisotel): use metric.WithAttributeSet to avoid copy (#3552)
In order to improve performance replace `WithAttributes` with `WithAttributeSet`.
This avoids the slice allocation and copy that is done in `WithAttributes`.
For more information see https://github.com/open-telemetry/opentelemetry-go/blob/v1.38.0/metric/instrument.go#L357-L376
* chore(docs): explain why MaxRetries is disabled for ClusterClient (#3551)
Co-authored-by: Nedyalko Dyakov <1547186+ndyakov@users.noreply.github.com>
* exponential backoff
* address pr comments
* address pr comments
* remove rlock
* add some comments
* add comments
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Warnar Boekkooi <wboekkooi@impossiblecloud.com>
Co-authored-by: Justin <justindsouza80@gmail.com>
- At present, the `scan` command is dispatched to a random slot.
- As far as I can tell, the scanX family of commands are not cluster aware (e.g. don't redirect the client to the correct slot).
- You can see [here](869dc0bb66/src/main/java/redis/clients/jedis/ShardedCommandObjects.java (L101)), the Jedis client calling `processKey` on the match argument, and this is what this PR also does.
We've had this patch running in production, and it seems to work well for us.
For further thought:
- Continuing looking at other Redis clients (e.g. Jedis), they outright [reject as invalid](869dc0bb66/src/main/java/redis/clients/jedis/ShardedCommandObjects.java (L98)) any scan command that does not include a hash-tag. Presumably this has the advantage of users not being surprised when their scan produces no results when a random server is picked.
- Perhaps it would be sensible for go-redis to do the same also?
Co-authored-by: Nedyalko Dyakov <1547186+ndyakov@users.noreply.github.com>
* wip
* update documentation
* add streamingcredentialsprovider in options
* fix: put back option in pool creation
* add package level comment
* Initial re authentication implementation
Introduces the StreamingCredentialsProvider as the CredentialsProvider
with the highest priority.
TODO: needs to be tested
* Change function type name
Change CancelProviderFunc to UnsubscribeFunc
* add tests
* fix race in tests
* fix example tests
* wip, hooks refactor
* fix build
* update README.md
* update wordlist
* update README.md
* refactor(auth): early returns in cred listener
* fix(doctest): simulate some delay
* feat(conn): add close hook on conn
* fix(tests): simulate start/stop in mock credentials provider
* fix(auth): don't double close the conn
* docs(README): mark streaming credentials provider as experimental
* fix(auth): streamline auth err proccess
* fix(auth): check err on close conn
* chore(entraid): use the repo under redis org
