memcache

package
v0.0.0-...-cacfd1a Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Jul 13, 2021 License: Apache-2.0 Imports: 13 Imported by: 0

Documentation

Overview

Package memcache provides a client for the memcached cache server.

Index

Constants

View Source
const (
	// DefaultTimeout is the default socket read/write timeout.
	DefaultTimeout = 100 * time.Millisecond

	// DefaultMaxIdleConns is the default maximum number of idle connections
	// kept for any single address.
	DefaultMaxIdleConns = 2
)
View Source
const (
	MIN_WORKER_COUNT = 3
	// At most 100 requests can be waiting inside of a chan to be unqueued
	MAX_BACKLOG_SIZE = 100

	// There can be at most 100 requests in flight per worker. Not tuned.
	MAX_BACKLOG_PER_WORKER = 100

	// Needed in case of spurious errors, has to be finite in case memcache cluster is down.
	MAX_RETRY_COUNT = 3
)

Variables

View Source
var (
	// ErrCacheMiss means that a Get failed because the item wasn't present.
	ErrCacheMiss = errors.New("memcache: cache miss")

	// ErrCASConflict means that a CompareAndSwap call failed due to the
	// cached value being modified between the Get and the CompareAndSwap.
	// If the cached value was simply evicted rather than replaced,
	// ErrNotStored will be returned instead.
	ErrCASConflict = errors.New("memcache: compare-and-swap conflict")

	// ErrNotStored means that a conditional write operation (i.e. Add or
	// CompareAndSwap) failed because the condition was not satisfied.
	ErrNotStored = errors.New("memcache: item not stored")

	// ErrServer means that a server error occurred.
	ErrServerError = errors.New("memcache: server error")

	// ErrNoStats means that no statistics were available.
	ErrNoStats = errors.New("memcache: no statistics available")

	// ErrMalformedKey is returned when an invalid key is used.
	// Keys must be at maximum 250 bytes long and not
	// contain whitespace or control characters.
	ErrMalformedKey = errors.New("malformed: key is too long or contains invalid characters")
)
View Source
var ErrPreviousRequestFailed = errors.New("A previous request failed")

Functions

func DebugLog

func DebugLog(message string)

func InitWorkerManager

func InitWorkerManager(manager *WorkerManager, maxWorkers int, connFactory ConnectionFactory)

func NewWorkerConnAndProcessor

func NewWorkerConnAndProcessor(cf ConnectionFactory) (workerConnAndProcessor, error)

func ResolveServerAddr

func ResolveServerAddr(server string) (net.Addr, error)

SetServers changes a ServerList's set of servers at runtime and is safe for concurrent use by multiple goroutines.

Each server is given equal weight. A server is given more weight if it's listed multiple times.

SetServers returns an error if any of the server names fail to resolve. No attempt is made to connect to the server. If any error is returned, no changes are made to the ServerList.

Types

type BufferedReader

type BufferedReader struct {
	// contains filtered or unexported fields
}

func (*BufferedReader) Read

func (reader *BufferedReader) Read(p []byte) (int, error)

func (*BufferedReader) ReadBytes

func (reader *BufferedReader) ReadBytes(delim byte) ([]byte, error)

ReadBytes returns a brand new byte slice that does not overlap with other byte slices

type ClientInterface

type ClientInterface interface {
	// TODO generalize
	SendProxiedMessageAsync(command *message.SingleMessage)

	Get(key string) (item *Item, err error)
	GetMulti(keys []string) (map[string]*Item, error)
	GetMultiArray(keys []string) ([]*Item, error)
	Set(item *Item) error
	Add(item *Item) error
	Replace(item *Item) error
	Increment(key string, delta uint64) (newValue uint64, err error)
	Decrement(key string, delta uint64) (newValue uint64, err error)
	Delete(key string) error
	DeleteAll() error
	Touch(key string, seconds int32) error
	Finalize()
}

type ConnectTimeoutError

type ConnectTimeoutError struct {
	Addr net.Addr
}

ConnectTimeoutError is the error type used when it takes too long to connect to the desired host. This level of detail can generally be ignored.

func (*ConnectTimeoutError) Error

func (cte *ConnectTimeoutError) Error() string

type ConnectionFactory

type ConnectionFactory interface {
	// contains filtered or unexported methods
}

interface for testability

type Item

type Item struct {
	// Key is the Item's key (250 bytes maximum).
	Key string

	// Value is the Item's value.
	Value []byte

	// Flags are server-opaque flags whose semantics are entirely
	// up to the app.
	Flags uint32

	// Expiration is the cache expiration time, in seconds: either a relative
	// time from now (up to 1 month), or an absolute Unix epoch time.
	// Zero means the Item has no expiration time.
	Expiration int32
	// contains filtered or unexported fields
}

Item is an item to be got or stored in a memcached server.

type PipeliningClient

type PipeliningClient struct {
	// Timeout specifies the socket read/write timeout.
	// If zero, DefaultTimeout is used.
	Timeout time.Duration

	// MaxIdleConns specifies the maximum number of idle connections that will
	// be maintained per address. If less than one, DefaultMaxIdleConns will be
	// used.
	//
	// Consider your expected traffic rates and latency carefully. This should
	// be set to a number higher than your peak parallel requests.
	MaxIdleConns int

	// For ketama
	Label  string
	Weight int
	// contains filtered or unexported fields
}

PipeliningClient is a memcache client with pipelining. It is safe for unlocked use by multiple concurrent goroutines.

func New

func New(server string, serverConnections int, timeout time.Duration) *PipeliningClient

New returns a memcache client using the provided server.

func (*PipeliningClient) Add

func (c *PipeliningClient) Add(item *Item) error

Add writes the given item, if no value already exists for its key. ErrNotStored is returned if that condition is not met.

func (*PipeliningClient) CompareAndSwap

func (c *PipeliningClient) CompareAndSwap(item *Item) error

CompareAndSwap writes the given item that was previously returned by Get, if the value was neither modified or evicted between the Get and the CompareAndSwap calls. The item's Key should not change between calls but all other item fields may differ. ErrCASConflict is returned if the value was modified in between the calls. ErrNotStored is returned if the value was evicted in between the calls.

func (*PipeliningClient) Decrement

func (c *PipeliningClient) Decrement(key string, delta uint64) (newValue uint64, err error)

Decrement atomically decrements key by delta. The return value is the new value after being decremented or an error. If the value didn't exist in memcached the error is ErrCacheMiss. The value in memcached must be an decimal number, or an error will be returned. On underflow, the new value is capped at zero and does not wrap around.

func (*PipeliningClient) Delete

func (c *PipeliningClient) Delete(key string) error

Delete deletes the item with the provided key. The error ErrCacheMiss is returned if the item didn't already exist in the cache.

func (*PipeliningClient) DeleteAll

func (c *PipeliningClient) DeleteAll() error

DeleteAll deletes all items in the cache.

func (*PipeliningClient) Finalize

func (c *PipeliningClient) Finalize()

Finalize is called in unit tests to free up open connections.

func (*PipeliningClient) FlushAll

func (c *PipeliningClient) FlushAll() error

func (*PipeliningClient) Get

func (c *PipeliningClient) Get(key string) (item *Item, err error)

Get gets the item for the given key. ErrCacheMiss is returned for a memcache cache miss. The key must be at most 250 bytes in length.

func (*PipeliningClient) GetMulti

func (c *PipeliningClient) GetMulti(keys []string) (map[string]*Item, error)

GetMulti is a batch version of Get. The returned map from keys to items may have fewer elements than the input slice, due to memcache cache misses. Each key must be at most 250 bytes in length. If no error is returned, the returned map will also be non-nil.

func (*PipeliningClient) GetMultiArray

func (c *PipeliningClient) GetMultiArray(keys []string) ([]*Item, error)

func (*PipeliningClient) GetServer

func (c *PipeliningClient) GetServer() string

GetServer returns the address of the server originally passed to memcache.New().

func (*PipeliningClient) Increment

func (c *PipeliningClient) Increment(key string, delta uint64) (newValue uint64, err error)

Increment atomically increments key by delta. The return value is the new value after being incremented or an error. If the value didn't exist in memcached the error is ErrCacheMiss. The value in memcached must be an decimal number, or an error will be returned. On 64-bit overflow, the new value wraps around.

func (*PipeliningClient) Replace

func (c *PipeliningClient) Replace(item *Item) error

Replace writes the given item, but only if the server *does* already hold data for this key

func (*PipeliningClient) SendProxiedMessageAsync

func (c *PipeliningClient) SendProxiedMessageAsync(command *message.SingleMessage)

func (*PipeliningClient) Set

func (c *PipeliningClient) Set(item *Item) error

Set writes the given item, unconditionally.

func (*PipeliningClient) Touch

func (c *PipeliningClient) Touch(key string, seconds int32) (err error)

Touch updates the expiry for the given key. The seconds parameter is either a Unix timestamp or, if seconds is less than 1 month, the number of seconds into the future at which time the item will expire. ErrCacheMiss is returned if the key is not in the cache. The key must be at most 250 bytes in length.

type WorkerManager

type WorkerManager struct {
	// contains filtered or unexported fields
}

func (*WorkerManager) Finalize

func (manager *WorkerManager) Finalize()

Directories

Path Synopsis
proxy listens on a socket and forwards data to one or more memcache servers (TODO: Actually shard requests)
proxy listens on a socket and forwards data to one or more memcache servers (TODO: Actually shard requests)
message
message contains utilities for proxying the sending out of a memcached request and receiving a memcached response
message contains utilities for proxying the sending out of a memcached request and receiving a memcached response
responsequeue
responsequeue ensures that responses to requests are sent in the order they are received
responsequeue ensures that responses to requests are sent in the order they are received

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL