MediaWiki  master
WANObjectCache Class Reference

Multi-datacenter aware caching interface. More...

Inheritance diagram for WANObjectCache:
Collaboration diagram for WANObjectCache:

Public Member Functions

 __construct (array $params)
 
 adaptiveTTL ( $mtime, $maxTTL, $minTTL=30, $factor=0.2)
 Get a TTL that is higher for objects that have not changed recently. More...
 
 clearLastError ()
 Clear the "last error" registry. More...
 
 clearProcessCache ()
 Clear the in-process caches; useful for testing. More...
 
 delete ( $key, $ttl=self::HOLDOFF_TTL)
 Purge a key from all datacenters. More...
 
 get ( $key, &$curTTL=null, array $checkKeys=[], &$info=[])
 Fetch the value of a key from cache. More...
 
 getCheckKeyTime ( $key)
 Fetch the value of a timestamp "check" key. More...
 
 getLastError ( $watchPoint=0)
 Get the "last error" registry. More...
 
 getMulti (array $keys, &$curTTLs=[], array $checkKeys=[], &$info=[])
 Fetch the value of several keys from cache. More...
 
 getMultiCheckKeyTime (array $keys)
 Fetch the values of each timestamp "check" key. More...
 
 getMultiWithSetCallback (ArrayIterator $keyedIds, $ttl, callable $callback, array $opts=[])
 Method to fetch multiple cache keys at once with regeneration. More...
 
 getMultiWithUnionSetCallback (ArrayIterator $keyedIds, $ttl, callable $callback, array $opts=[])
 Method to fetch/regenerate multiple cache keys at once. More...
 
 getQoS ( $flag)
 
 getWarmupKeyMisses ()
 
 getWithSetCallback ( $key, $ttl, $callback, array $opts=[], array $cbParams=[])
 Method to fetch/regenerate a cache key. More...
 
 hash256 ( $component)
 Hash a possibly long string into a suitable component for makeKey()/makeGlobalKey() More...
 
 makeGlobalKey ( $collection,... $components)
 Make a cache key for the global keyspace and given components. More...
 
 makeKey ( $collection,... $components)
 Make a cache key using the "global" keyspace for the given components. More...
 
 makeMultiKeys (array $ids, $keyCallback)
 Get an iterator of (cache key => entity ID) for a list of entity IDs. More...
 
 multiRemap (array $ids, array $res)
 Get an (ID => value) map from (i) a non-unique list of entity IDs, and (ii) the list of corresponding entity values by first appearance of each ID in the entity ID list. More...
 
 reap ( $key, $purgeTimestamp, &$isStale=false)
 Set a key to soon expire in the local cluster if it pre-dates $purgeTimestamp. More...
 
 reapCheckKey ( $key, $purgeTimestamp, &$isStale=false)
 Set a "check" key to soon expire in the local cluster if it pre-dates $purgeTimestamp. More...
 
 resetCheckKey ( $key)
 Clear the last-purge timestamp of a "check" key in all datacenters. More...
 
 set ( $key, $value, $ttl=self::TTL_INDEFINITE, array $opts=[])
 Set the value of a key in cache. More...
 
 setLogger (LoggerInterface $logger)
 
 setMockTime (&$time)
 
 touchCheckKey ( $key, $holdoff=self::HOLDOFF_TTL)
 Increase the last-purge timestamp of a "check" key in all datacenters. More...
 
 useInterimHoldOffCaching ( $enabled)
 Enable or disable the use of brief caching for tombstoned keys. More...
 
 watchErrors ()
 Get a "watch point" token that can be used to get the "last error" to occur after now. More...
 

Static Public Member Functions

static getCollectionFromSisterKey (string $sisterKey)
 
static newEmpty ()
 Get an instance that wraps EmptyBagOStuff. More...
 

Public Attributes

const GRACE_TTL_NONE = 0
 Idiom for set()/getWithSetCallback() meaning "no post-expiration grace period". More...
 
const HOLDOFF_TTL = self::MAX_COMMIT_DELAY + self::MAX_READ_LAG + 1
 Seconds to tombstone keys on delete() and to treat keys as volatile after purges. More...
 
const HOLDOFF_TTL_NONE = 0
 Idiom for delete()/touchCheckKey() meaning "no hold-off period". More...
 
const KEY_AS_OF = 'asOf'
 Generation completion timestamp attribute for a key; keep value for b/c (< 1.36) More...
 
const KEY_CHECK_AS_OF = 'lastCKPurge'
 Highest "check" key timestamp for a key; keep value for b/c (< 1.36) More...
 
const KEY_CUR_TTL = 'curTTL'
 Remaining TTL attribute for a key; keep value for b/c (< 1.36) More...
 
const KEY_TOMB_AS_OF = 'tombAsOf'
 Tomstone timestamp attribute for a key; keep value for b/c (< 1.36) More...
 
const KEY_TTL = 'ttl'
 Logical TTL attribute for a key. More...
 
const KEY_VERSION = 'version'
 Version number attribute for a key; keep value for b/c (< 1.36) More...
 
const PASS_BY_REF = []
 Idiom for get()/getMulti() to return extra information by reference. More...
 
const STALE_TTL_NONE = 0
 Idiom for set()/getWithSetCallback() meaning "no post-expiration persistence". More...
 
const TTL_LAGGED = 30
 Max TTL, in seconds, to store keys when a data source has high replication lag. More...
 
- Public Attributes inherited from Wikimedia\LightweightObjectStore\StorageAwareness
const ATTR_DURABILITY = 2
 Durability of writes; see QOS_DURABILITY_* (higher means stronger) More...
 
const ATTR_EMULATION = 1
 Emulation/fallback mode; see QOS_EMULATION_*; higher is better. More...
 
const ERR_NO_RESPONSE = 1
 Storage medium failed to yield a complete response to an operation. More...
 
const ERR_NONE = 0
 No storage medium error. More...
 
const ERR_UNEXPECTED = 3
 Storage medium operation failed due to usage limitations or an I/O error. More...
 
const ERR_UNREACHABLE = 2
 Storage medium could not be reached to establish a connection. More...
 
const QOS_DURABILITY_DISK = 4
 Data is saved to disk and writes do not usually block on fsync() More...
 
const QOS_DURABILITY_NONE = 1
 Data is never saved to begin with (blackhole store) More...
 
const QOS_DURABILITY_RDBMS = 5
 Data is saved to disk and writes usually block on fsync(), like a standard RDBMS. More...
 
const QOS_DURABILITY_SCRIPT = 2
 Data is lost at the end of the current web request or CLI script. More...
 
const QOS_DURABILITY_SERVICE = 3
 Data is lost once the service storing the data restarts. More...
 
const QOS_EMULATION_SQL = 1
 Fallback disk-based SQL store. More...
 
const QOS_UNKNOWN = INF
 Generic "unknown" value; useful for comparisons (always "good enough") More...
 

Protected Member Functions

 fetchKeys (array $keys, array $checkKeys, $touchedCb=null)
 Fetch the value and key metadata of several keys from cache. More...
 
 getCurrentTime ()
 
 isLotteryRefreshDue ( $res, $lowTTL, $ageNew, $hotTTR, $now)
 Check if a key is due for randomized regeneration due to near-expiration/popularity. More...
 
 isValid ( $value, $asOf, $minAsOf)
 Check that a wrapper value exists and has an acceptable age. More...
 
 prependRoute (string $sisterKey, string $route)
 
 relayNonVolatilePurge (string $sisterKey)
 Remove a sister key from all datacenters. More...
 
 relayVolatilePurges (array $purgeBySisterKey, int $ttl)
 Set a sister key to a purge value in all datacenters. More...
 
 worthRefreshExpiring ( $curTTL, $logicalTTL, $lowTTL)
 Check if a key is nearing expiration and thus due for randomized regeneration. More...
 
 worthRefreshPopular ( $asOf, $ageNew, $timeTillRefresh, $now)
 Check if a key is due for randomized regeneration due to its popularity. More...
 

Protected Attributes

callable null $asyncHandler
 Function that takes a WAN cache callback and runs it later. More...
 
string null $broadcastRoute
 Routing prefix for operations that should be broadcasted to all data centers. More...
 
BagOStuff $cache
 The local datacenter cache. More...
 
int $coalesceScheme
 Scheme to use for key coalescing (Hash Tags or Hash Stops) More...
 
float $epoch
 Unix timestamp of the oldest possible valid values. More...
 
LoggerInterface $logger
 
MapCacheLRU[] $processCaches = []
 Map of group PHP instance caches. More...
 
string $secret
 Stable secret used for hashing long strings into key components. More...
 
StatsdDataFactoryInterface $stats
 
bool $useInterimHoldOffCaching = true
 Whether to use "interim" caching while keys are tombstoned. More...
 

Detailed Description

Multi-datacenter aware caching interface.

Using WANObjectCache

All operations go to the local datacenter cache, except for delete(), touchCheckKey(), and resetCheckKey(), which broadcast to all datacenters.

This class is intended for caching data from primary stores. If the get() method does not return a value, then the caller should query the new value and backfill the cache using set(). The preferred way to do this logic is through getWithSetCallback(). When querying the store on cache miss, the closest DB replica should be used. Try to avoid heavyweight DB primary or quorum reads.

To ensure consumers of the cache see new values in a timely manner, you either need to follow either the validation strategy, or the purge strategy.

The validation strategy refers to the natural avoidance of stale data by one of the following means:

  • A) The cached value is immutable. If the consumer has access to an identifier that uniquely describes a value, cached value need not change. Instead, the key can change. This also allows all servers to access their perceived current version. This is important in context of multiple deployed versions of your application and/or cross-dc database replication, to ensure deterministic values without oscillation.
  • B) Validity is checked against the source after get(). This is the inverse of A. The unique identifier is embedded inside the value and validated after on retrieval. If outdated, the value is recomputed.
  • C) The value is cached with a modest TTL (without validation). If value recomputation is reasonably performant, and the value is allowed to be stale, one should consider using TTL only – using the value's age as method of validation.

The purge strategy refers to the approach whereby your application knows that source data has changed and can react by purging the relevant cache keys. As purges are expensive, this strategy should be avoided if possible. The simplest purge method is delete().

No matter which strategy you choose, callers must not rely on updates or purges being immediately visible to other servers. It should be treated similarly as one would a database replica.

The need for immediate updates should be avoided. If needed, solutions must be sought outside WANObjectCache.

Deploying WANObjectCache

There are two supported ways to set up broadcasted operations:

  • A) Set up mcrouter as the cache backend, with a memcached BagOStuff class for the 'cache' parameter, and a wildcard routing prefix for the 'broadcastRoutingPrefix' parameter. Configure mcrouter as follows:
  • B) Set up dynomite as the cache backend, using a memcached BagOStuff class for the 'cache' parameter. Note that with this setup, all key setting operations will be broadcasted, rather than just purges. Writes will be eventually consistent via the Dynamo replication model. See https://github.com/Netflix/dynomite.

Broadcasted operations like delete() and touchCheckKey() are intended to run immediately in the local datacenter and asynchronously in remote datacenters.

This means that callers in all datacenters may see older values for however many milliseconds that the purge took to reach that datacenter. As with any cache, this should not be relied on for cases where reads are used to determine writes to source (e.g. non-cache) data stores, except when reading immutable data.

Internally, access to a given key actually involves the use of one or more "sister" keys. A sister key is constructed by prefixing the base key with "WANCache:" (used to distinguish WANObjectCache formatted keys) and suffixing a colon followed by a single-character sister key type. The sister key types include the following:

  • v: used to store "regular" values (metadata-wrapped) and temporary purge "tombstones".
  • t: used to store "last purge" timestamps for "check" keys.
  • m: used to store temporary mutex locks to avoid cache stampedes.
  • i: used to store temporary interim values (metadata-wrapped) for tombstoned keys.
  • c: used to store temporary "cool-off" indicators, which specify a period during which values cannot be stored, neither regularly nor using interim keys.
Stability: newable
Since
1.26

Definition at line 127 of file WANObjectCache.php.

Constructor & Destructor Documentation

◆ __construct()

WANObjectCache::__construct ( array  $params)
Stability: stable
to call
Parameters
array$params
  • cache : BagOStuff object for a persistent cache
  • logger : LoggerInterface object
  • stats : StatsdDataFactoryInterface object
  • asyncHandler : A function that takes a callback and runs it later. If supplied, whenever a preemptive refresh would be triggered in getWithSetCallback(), the current cache value is still used instead. However, the async-handler function receives a WAN cache callback that, when run, will execute the value generation callback supplied by the getWithSetCallback() caller. The result will be saved as normal. The handler is expected to call the WAN cache callback at an opportune time (e.g. HTTP post-send), though generally within a few 100ms. [optional]
  • broadcastRoutingPrefix: a routing prefix used to broadcast certain operations to all datacenters; See also https://github.com/facebook/mcrouter/wiki/Config-Files. This prefix takes the form /<datacenter>/<name of wan route>/, where datacenter is usually a wildcard to select all matching routes (e.g. the WAN cluster in all DCs). See also https://github.com/facebook/mcrouter/wiki/Multi-cluster-broadcast-setup. This is required when using mcrouter as a multi-region backing store proxy. [optional]
  • epoch: lowest UNIX timestamp a value/tombstone must have to be valid. [optional]
  • secret: stable secret used for hashing long strings into key components. [optional]
  • coalesceScheme: which key scheme to use in order to encourage the backend to place any "helper" keys for a "value" key within the same cache server. This reduces network overhead and reduces the chance the a single downed cache server causes disruption. Use "hash_stop" with mcrouter and "hash_tag" with dynomite. [default: "hash_stop"]
  • keyHighQps: reads/second assumed during a hypothetical cache write stampede for a single key. This is used to decide when the overhead of checking short-lived write throttling keys is worth it. [default: 100]
  • keyHighUplinkBps: maximum tolerable bytes/second to spend on a cache write stampede for a single key. This is used to decide when the overhead of checking short-lived write throttling keys is worth it. [default: (1/100 of a 1Gbps link)]

Definition at line 347 of file WANObjectCache.php.

References setLogger().

Member Function Documentation

◆ adaptiveTTL()

WANObjectCache::adaptiveTTL (   $mtime,
  $maxTTL,
  $minTTL = 30,
  $factor = 0.2 
)

Get a TTL that is higher for objects that have not changed recently.

This is useful for keys that get explicit purges and DB or purge relay lag is a potential concern (especially how it interacts with CDN cache)

Example usage:

// Last-modified time of page
$mtime = wfTimestamp( TS_UNIX, $page->getTimestamp() );
// Get adjusted TTL. If $mtime is 3600 seconds ago and $minTTL/$factor left at
// defaults, then $ttl is 3600 * .2 = 720. If $minTTL was greater than 720, then
// $ttl would be $minTTL. If $maxTTL was smaller than 720, $ttl would be $maxTTL.
$ttl = $cache->adaptiveTTL( $mtime, $cache::TTL_DAY );
wfTimestamp( $outputtype=TS_UNIX, $ts=0)
Get a timestamp string in one of various formats.
BagOStuff $cache
The local datacenter cache.

Another use case is when there are no applicable "last modified" fields in the DB, and there are too many dependencies for explicit purges to be viable, and the rate of change to relevant content is unstable, and it is highly valued to have the cached value be as up-to-date as possible.

Example usage:

$query = "<some complex query>";
$idListFromComplexQuery = $cache->getWithSetCallback(
$cache->makeKey( 'complex-graph-query', $hashOfQuery ),
GraphQueryClass::STARTING_TTL,
function ( $oldValue, &$ttl, array &$setOpts, $oldAsOf ) use ( $query, $cache ) {
$gdb = $this->getReplicaGraphDbConnection();
// Account for any snapshot/replica DB lag
$setOpts += GraphDatabase::getCacheSetOptions( $gdb );
$newList = iterator_to_array( $gdb->query( $query ) );
sort( $newList, SORT_NUMERIC ); // normalize
$minTTL = GraphQueryClass::MIN_TTL;
$maxTTL = GraphQueryClass::MAX_TTL;
if ( $oldValue !== false ) {
// Note that $oldAsOf is the last time this callback ran
$ttl = ( $newList === $oldValue )
// No change: cache for 150% of the age of $oldValue
? $cache->adaptiveTTL( $oldAsOf, $maxTTL, $minTTL, 1.5 )
// Changed: cache for 50% of the age of $oldValue
: $cache->adaptiveTTL( $oldAsOf, $maxTTL, $minTTL, .5 );
}
return $newList;
},
[
// Keep stale values around for doing comparisons for TTL calculations.
// High values improve long-tail keys hit-rates, though might waste space.
'staleTTL' => GraphQueryClass::GRACE_TTL
]
);
getWithSetCallback( $key, $exptime, $callback, $flags=0)
Get an item, regenerating and setting it if not found.
Definition: BagOStuff.php:212
makeKey( $collection,... $components)
Make a cache key for the global keyspace and given components.
Parameters
int | float | string | null$mtimeUNIX timestamp; null if none
int$maxTTLMaximum TTL (seconds)
int$minTTLMinimum TTL (seconds); Default: 30
float$factorValue in the range (0,1); Default: .2
Returns
int Adaptive TTL
Since
1.28

Definition at line 2626 of file WANObjectCache.php.

References getCurrentTime().

◆ clearLastError()

WANObjectCache::clearLastError ( )
final

Clear the "last error" registry.

Deprecated:
Since 1.38

Definition at line 2517 of file WANObjectCache.php.

◆ clearProcessCache()

WANObjectCache::clearProcessCache ( )

Clear the in-process caches; useful for testing.

Since
1.27

Definition at line 2526 of file WANObjectCache.php.

◆ delete()

WANObjectCache::delete (   $key,
  $ttl = self::HOLDOFF_TTL 
)
final

Purge a key from all datacenters.

This should only be called when the underlying data (being cached) changes in a significant way. This deletes the key and starts a hold-off period where the key cannot be written to for a few seconds (HOLDOFF_TTL). This is done to avoid the following race condition:

  • a) Some DB data changes and delete() is called on a corresponding key
  • b) A request refills the key with a stale value from a lagged DB
  • c) The stale value is stuck there until the key is expired/evicted

This is implemented by storing a special "tombstone" value at the cache key that this class recognizes; get() calls will return false for the key and any set() calls will refuse to replace tombstone values at the key. For this to always avoid stale value writes, the following must hold:

  • a) Replication lag is bounded to being less than HOLDOFF_TTL; or
  • b) If lag is higher, the DB will have gone into read-only mode already

Note that set() can also be lag-aware and lower the TTL if it's high.

Be aware that this does not clear the process cache. Even if it did, callbacks used by getWithSetCallback() might still return stale data in the case of either uncommitted or not-yet-replicated changes (callback generally use replica DBs).

When using potentially long-running ACID transactions, a good pattern is to use a pre-commit hook to issue the delete. This means that immediately after commit, callers will see the tombstone in cache upon purge relay. It also avoids the following race condition:

  • a) T1 begins, changes a row, and calls delete()
  • b) The HOLDOFF_TTL passes, expiring the delete() tombstone
  • c) T2 starts, reads the row and calls set() due to a cache miss
  • d) T1 finally commits
  • e) Stale value is stuck in cache

Example usage:

$dbw->startAtomic( __METHOD__ ); // start of request
... <execute some stuff> ...
// Update the row in the DB
$dbw->update( ... );
$key = $cache->makeKey( 'homes', $homeId );
// Purge the corresponding cache entry just before committing
$dbw->onTransactionPreCommitOrIdle( function() use ( $cache, $key ) {
$cache->delete( $key );
} );
... <execute some stuff> ...
$dbw->endAtomic( __METHOD__ ); // end of request
delete( $key, $flags=0)
Delete an item if it exists.

The $ttl parameter can be used when purging values that have not actually changed recently. For example, user-requested purges or cache cleanup scripts might not need to invoke a hold-off period on cache backfills, so they can use HOLDOFF_TTL_NONE.

Note that $ttl limits the effective range of 'lockTSE' for getWithSetCallback().

If called twice on the same key, then the last hold-off TTL takes precedence. For idempotence, the $ttl should not vary for different delete() calls on the same key.

Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
int$ttlTombstone TTL; Default: WANObjectCache::HOLDOFF_TTL
Returns
bool True if the item was purged or not found, false on failure

Definition at line 978 of file WANObjectCache.php.

References getCurrentTime(), relayNonVolatilePurge(), and relayVolatilePurges().

◆ fetchKeys()

WANObjectCache::fetchKeys ( array  $keys,
array  $checkKeys,
  $touchedCb = null 
)
protected

Fetch the value and key metadata of several keys from cache.

$checkKeys holds the "check" keys used to validate values of applicable keys. The integer indexes hold "check" keys that apply to all of $keys while the string indexes hold "check" keys that only apply to the cache key with that name.

Parameters
string[]$keysList/map with makeKey()/makeGlobalKey() cache keys as values
string[] | string[][]$checkKeysMap of (integer or cache key => "check" key(s)); "check" keys must also be made with makeKey()/makeGlobalKey()
callable | null$touchedCbCallback yielding a UNIX timestamp from a value, or null
Returns
array<string,array> Map of (key => WANObjectCache::RESULT_* map) in order of $keys
Note
Callable type hints are not used to avoid class-autoloading

Definition at line 554 of file WANObjectCache.php.

References $keys, $res, and getCurrentTime().

Referenced by get(), getMulti(), and getMultiWithUnionSetCallback().

◆ get()

WANObjectCache::get (   $key,
$curTTL = null,
array  $checkKeys = [],
$info = [] 
)
final

Fetch the value of a key from cache.

If supplied, $curTTL is set to the remaining TTL (current time left):

  • a) INF; if $key exists, has no TTL, and is not purged by $checkKeys
  • b) float (>=0); if $key exists, has a TTL, and is not purged by $checkKeys
  • c) float (<0); if $key is tombstoned, stale, or existing but purged by $checkKeys
  • d) null; if $key does not exist and is not tombstoned

If a key is tombstoned, $curTTL will reflect the time since delete().

The timestamp of $key will be checked against the last-purge timestamp of each of $checkKeys. Those $checkKeys not in cache will have the last-purge initialized to the current timestamp. If any of $checkKeys have a timestamp greater than that of $key, then $curTTL will reflect how long ago $key became invalid. Callers can use $curTTL to know when the value is stale. The $checkKeys parameter allow mass key purges by updating a single key:

  • a) Each "check" key represents "last purged" of some source data
  • b) Callers pass in relevant "check" keys as $checkKeys in get()
  • c) When the source data that "check" keys represent changes, the touchCheckKey() method is called on them

Source data entities might exists in a DB that uses snapshot isolation (e.g. the default REPEATABLE-READ in innoDB). Even for mutable data, that isolation can largely be maintained by doing the following:

However, pre-snapshot values might still be seen if an update was made in a remote datacenter but the purge from delete() didn't relay yet.

Consider using getWithSetCallback(), which has cache slam avoidance and key versioning features, instead of bare get()/set() calls.

Do not use this method on versioned keys accessed via getWithSetCallback().

When using the $info parameter, it should be passed in as WANObjectCache::PASS_BY_REF. In that case, it becomes a key metadata map. Otherwise, for backwards compatibility, $info becomes the value generation timestamp (null if the key is nonexistant/tombstoned). Key metadata map fields include:

Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
float | null&$curTTLSeconds of TTL left [returned]
string[]$checkKeysMap of (integer or cache key => "check" key(s)); "check" keys must also be made with makeKey()/makeGlobalKey()
array&$infoMetadata map [returned]
Returns
mixed Value of cache key; false on failure

Definition at line 449 of file WANObjectCache.php.

References $res, fetchKeys(), getCurrentTime(), and PASS_BY_REF.

◆ getCheckKeyTime()

WANObjectCache::getCheckKeyTime (   $key)
final

Fetch the value of a timestamp "check" key.

The key will be initialized to the current time if not set, so only call this method if this behavior is actually desired

The timestamp can be used to check whether a cached value is valid. Callers should not assume that this returns the same timestamp in all datacenters due to relay delays.

The level of staleness can roughly be estimated from this key, but if the key was evicted from cache, such calculations may show the time since expiry as ~0 seconds.

Note that "check" keys won't collide with other regular keys.

Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
Returns
float UNIX timestamp

Definition at line 1027 of file WANObjectCache.php.

References getMultiCheckKeyTime().

◆ getCollectionFromSisterKey()

static WANObjectCache::getCollectionFromSisterKey ( string  $sisterKey)
static
Parameters
string$sisterKeySister key from makeSisterKey()
Returns
string Key collection name
Access: internal
For use by WANObjectCache/BagOStuff only
Since
1.36

Definition at line 1824 of file WANObjectCache.php.

◆ getCurrentTime()

WANObjectCache::getCurrentTime ( )
protected
Returns
float UNIX timestamp

Definition at line 3137 of file WANObjectCache.php.

Referenced by adaptiveTTL(), delete(), fetchKeys(), get(), getMultiCheckKeyTime(), set(), and touchCheckKey().

◆ getLastError()

WANObjectCache::getLastError (   $watchPoint = 0)
final

Get the "last error" registry.

The method should be invoked by a caller as part of the following pattern:

  • The caller invokes watchErrors() to get a "since token"
  • The caller invokes a sequence of cache operation methods
  • The caller invokes getLastError() with the "since token"

External callers can also invoke this method as part of the following pattern:

Parameters
int$watchPointOnly consider errors from after this "watch point" [optional]
Returns
int BagOStuff:ERR_* constant for the "last error" registry
Note
Parameters added in 1.38: $watchPoint

Definition at line 2499 of file WANObjectCache.php.

References Wikimedia\LightweightObjectStore\StorageAwareness\ERR_NO_RESPONSE, Wikimedia\LightweightObjectStore\StorageAwareness\ERR_NONE, Wikimedia\LightweightObjectStore\StorageAwareness\ERR_UNEXPECTED, and Wikimedia\LightweightObjectStore\StorageAwareness\ERR_UNREACHABLE.

◆ getMulti()

WANObjectCache::getMulti ( array  $keys,
$curTTLs = [],
array  $checkKeys = [],
$info = [] 
)
final

Fetch the value of several keys from cache.

$curTTLs becomes a map of only present/tombstoned $keys to their current time-to-live.

$checkKeys holds the "check" keys used to validate values of applicable keys. The integer indexes hold "check" keys that apply to all of $keys while the string indexes hold "check" keys that only apply to the cache key with that name. The logic of "check" keys otherwise works the same as in WANObjectCache::get().

When using the $info parameter, it should be passed in as WANObjectCache::PASS_BY_REF. In that case, it becomes a mapping of all the $keys to their metadata maps, each in the style of WANObjectCache::get(). Otherwise, for backwards compatibility, $info becomes a map of only present/tombstoned $keys to their value generation timestamps.

See also
WANObjectCache::get()
Parameters
string[]$keysList/map with makeKey()/makeGlobalKey() cache keys as values
array<string,float>&$curTTLs Map of (key => seconds of TTL left) [returned]
string[] | string[][]$checkKeysMap of (integer or cache key => "check" key(s)); "check" keys must also be made with makeKey()/makeGlobalKey()
array<string,array>&$info Map of (key => metadata map) [returned]
Returns
array<string,mixed> Map of (key => value) for existing values in order of $keys

Definition at line 502 of file WANObjectCache.php.

References $res, fetchKeys(), and PASS_BY_REF.

◆ getMultiCheckKeyTime()

WANObjectCache::getMultiCheckKeyTime ( array  $keys)
final

Fetch the values of each timestamp "check" key.

This works like getCheckKeyTime() except it takes a list of keys and returns a map of timestamps instead of just that of one key

This might be useful if both:

  • a) a class of entities each depend on hundreds of other entities
  • b) these other entities are depended upon by millions of entities

The later entities can each use a "check" key to purge their dependee entities. However, it is expensive for the former entities to verify against all of the relevant "check" keys during each getWithSetCallback() call. A less expensive approach is to do these verifications only after a "time-till-verify" (TTV) has passed. This is a middle ground between using blind TTLs and using constant verification. The adaptiveTTL() method can be used to dynamically adjust the TTV. Also, the initial TTV can make use of the last-modified times of the dependent entities (either from the DB or the "check" keys).

Example usage:

$cache->makeGlobalKey( 'wikibase-item', $id ),
self::INITIAL_TTV, // initial time-till-verify
function ( $oldValue, &$ttv, &$setOpts, $oldAsOf ) use ( $checkKeys, $cache ) {
$now = microtime( true );
// Use $oldValue if it passes max ultimate age and "check" key comparisons
if ( $oldValue &&
$oldAsOf > max( $cache->getMultiCheckKeyTime( $checkKeys ) ) &&
( $now - $oldValue['ctime'] ) <= self::MAX_CACHE_AGE
) {
// Increase time-till-verify by 50% of last time to reduce overhead
$ttv = $cache->adaptiveTTL( $oldAsOf, self::MAX_TTV, self::MIN_TTV, 1.5 );
// Unlike $oldAsOf, "ctime" is the ultimate age of the cached data
return $oldValue;
}
$mtimes = []; // dependency last-modified times; passed by reference
$value = [ 'data' => $this->fetchEntityData( $mtimes ), 'ctime' => $now ];
// Guess time-till-change among the dependencies, e.g. 1/(total change rate)
$ttc = 1 / array_sum( array_map(
function ( $mtime ) use ( $now ) {
return 1 / ( $mtime ? ( $now - $mtime ) : 900 );
},
$mtimes
) );
// The time-to-verify should not be overly pessimistic nor optimistic
$ttv = min( max( $ttc, self::MIN_TTV ), self::MAX_TTV );
return $value;
},
[ 'staleTTL' => $cache::TTL_DAY ] // keep around to verify and re-save
);
makeGlobalKey( $collection,... $components)
Make a cache key for the default keyspace and given components.
See also
WANObjectCache::getCheckKeyTime()
WANObjectCache::getWithSetCallback()
Parameters
string[]$keysCache keys made with makeKey()/makeGlobalKey()
Returns
float[] Map of (key => UNIX timestamp)
Since
1.31

Definition at line 1092 of file WANObjectCache.php.

References $keys, and getCurrentTime().

Referenced by getCheckKeyTime().

◆ getMultiWithSetCallback()

WANObjectCache::getMultiWithSetCallback ( ArrayIterator  $keyedIds,
  $ttl,
callable  $callback,
array  $opts = [] 
)
final

Method to fetch multiple cache keys at once with regeneration.

This works the same as getWithSetCallback() except:

  • a) The $keys argument must be the result of WANObjectCache::makeMultiKeys()
  • b) The $callback argument expects a function that returns an entity value, using boolean "false" if it does not exist. The callback takes the following arguments:
    • $id: ID of the entity to query
    • $oldValue: prior cache value or false if none was present
    • &$ttl: reference to the TTL to be assigned to the new value (alterable)
    • &$setOpts: reference to the new value set() options (alterable)
    • $oldAsOf: generation UNIX timestamp of $oldValue or null if not present
  • c) The return value is a map of (cache key => value) in the order of $keyedIds
See also
WANObjectCache::getWithSetCallback()
WANObjectCache::getMultiWithUnionSetCallback()

Example usage:

$rows = $cache->getMultiWithSetCallback(
// Map of cache keys to entity IDs
$cache->makeMultiKeys(
$this->fileVersionIds(),
function ( $id ) use ( $cache ) {
return $cache->makeKey( 'file-version', $id );
}
),
// Time-to-live (in seconds)
$cache::TTL_DAY,
// Function that derives the new key value
function ( $id, $oldValue, &$ttl, array &$setOpts ) {
// Account for any snapshot/replica DB lag
$setOpts += Database::getCacheSetOptions( $dbr );
// Load the row for this file
$queryInfo = File::getQueryInfo();
$row = $dbr->selectRow(
$queryInfo['tables'],
$queryInfo['fields'],
[ 'id' => $id ],
__METHOD__,
[],
$queryInfo['joins']
);
return $row ? (array)$row : false;
},
[
// Process cache for 30 seconds
'pcTTL' => 30,
// Use a dedicated 500 item cache (initialized on-the-fly)
'pcGroup' => 'file-versions:500'
]
);
$files = array_map( [ __CLASS__, 'newFromRow' ], $rows );
wfGetDB( $db, $groups=[], $wiki=false)
Get a Database object.
const DB_REPLICA
Definition: defines.php:26
Parameters
ArrayIterator$keyedIdsResult of WANObjectCache::makeMultiKeys()
int$ttlSeconds to live for key updates
callable$callbackCallback that yields entity generation callbacks
array$optsOptions map similar to that of getWithSetCallback()
Returns
mixed[] Map of (cache key => value) in the same order as $keyedIds
Since
1.28

Definition at line 2057 of file WANObjectCache.php.

References getWithSetCallback().

◆ getMultiWithUnionSetCallback()

WANObjectCache::getMultiWithUnionSetCallback ( ArrayIterator  $keyedIds,
  $ttl,
callable  $callback,
array  $opts = [] 
)
final

Method to fetch/regenerate multiple cache keys at once.

This works the same as getWithSetCallback() except:

  • a) The $keys argument expects the result of WANObjectCache::makeMultiKeys()
  • b) The $callback argument expects a function that returns a map of (ID => new value), using boolean "false" for entities that could not be found, for all entity IDs in $ids. The callback takes the following arguments:
    • $ids: list of entity IDs that require value generation
    • &$ttls: reference to the (entity ID => new TTL) map (alterable)
    • &$setOpts: reference to the new value set() options (alterable)
  • c) The return value is a map of (cache key => value) in the order of $keyedIds
  • d) The "lockTSE" and "busyValue" options are ignored
See also
WANObjectCache::getWithSetCallback()
WANObjectCache::getMultiWithSetCallback()

Example usage:

$rows = $cache->getMultiWithUnionSetCallback(
// Map of cache keys to entity IDs
$cache->makeMultiKeys(
$this->fileVersionIds(),
function ( $id ) use ( $cache ) {
return $cache->makeKey( 'file-version', $id );
}
),
// Time-to-live (in seconds)
$cache::TTL_DAY,
// Function that derives the new key value
function ( array $ids, array &$ttls, array &$setOpts ) {
// Account for any snapshot/replica DB lag
$setOpts += Database::getCacheSetOptions( $dbr );
// Load the rows for these files
$rows = array_fill_keys( $ids, false );
$queryInfo = File::getQueryInfo();
$res = $dbr->select(
$queryInfo['tables'],
$queryInfo['fields'],
[ 'id' => $ids ],
__METHOD__,
[],
$queryInfo['joins']
);
foreach ( $res as $row ) {
$rows[$row->id] = $row;
$mtime = wfTimestamp( TS_UNIX, $row->timestamp );
$ttls[$row->id] = $this->adaptiveTTL( $mtime, $ttls[$row->id] );
}
return $rows;
},
]
);
$files = array_map( [ __CLASS__, 'newFromRow' ], $rows );
adaptiveTTL( $mtime, $maxTTL, $minTTL=30, $factor=0.2)
Get a TTL that is higher for objects that have not changed recently.
Parameters
ArrayIterator$keyedIdsResult of WANObjectCache::makeMultiKeys()
int$ttlSeconds to live for key updates
callable$callbackCallback that yields entity generation callbacks
array$optsOptions map similar to that of getWithSetCallback()
Returns
mixed[] Map of (cache key => value) in the same order as $keyedIds
Since
1.30

Definition at line 2160 of file WANObjectCache.php.

References $res, fetchKeys(), and getWithSetCallback().

◆ getQoS()

WANObjectCache::getQoS (   $flag)
Parameters
int$flagATTR_* class constant
Returns
int QOS_* class constant
Since
1.28

Definition at line 2559 of file WANObjectCache.php.

◆ getWarmupKeyMisses()

WANObjectCache::getWarmupKeyMisses ( )
final
Access: internal
For use by unit tests only
Returns
int
Since
1.30

Definition at line 2644 of file WANObjectCache.php.

◆ getWithSetCallback()

WANObjectCache::getWithSetCallback (   $key,
  $ttl,
  $callback,
array  $opts = [],
array  $cbParams = [] 
)
final

Method to fetch/regenerate a cache key.

On cache miss, the key will be set to the callback result via set() (unless the callback returns false) and that result will be returned. The arguments supplied to the callback are:

  • $oldValue: prior cache value or false if none was present
  • &$ttl: alterable reference to the TTL to be assigned to the new value
  • &$setOpts: alterable reference to the set() options to be used with the new value
  • $oldAsOf: generation UNIX timestamp of $oldValue or null if not present (since 1.28)
  • $params: custom field/value map as defined by $cbParams (since 1.35)

It is strongly recommended to set the 'lag' and 'since' fields to avoid race conditions that can cause stale values to get stuck at keys. Usually, callbacks ignore the current value, but it can be used to maintain "most recent X" values that come from time or sequence based source data, provided that the "as of" id/time is tracked. Note that preemptive regeneration and $checkKeys can result in a non-false current value.

Usage of $checkKeys is similar to get() and getMulti(). However, rather than the caller having to inspect a "current time left" variable (e.g. $curTTL, $curTTLs), a cache regeneration will automatically be triggered using the callback.

The $ttl argument and "hotTTR" option (in $opts) use time-dependent randomization to avoid stampedes. Keys that are slow to regenerate and either heavily used or subject to explicit (unpredictable) purges, may need additional mechanisms. The simplest way to avoid stampedes for such keys is to use 'lockTSE' (in $opts). If explicit purges are needed, also:

  • a) Pass $key into $checkKeys
  • b) Use touchCheckKey( $key ) instead of delete( $key )

This applies cache server I/O stampede protection against duplicate cache sets. This is important when the callback is slow and/or yields large values for a key.

Example usage (typical key):

// Key to store the cached value under
$cache->makeKey( 'cat-attributes', $catId ),
// Time-to-live (in seconds)
$cache::TTL_MINUTE,
// Function that derives the new key value
function ( $oldValue, &$ttl, array &$setOpts ) {
// Account for any snapshot/replica DB lag
$setOpts += Database::getCacheSetOptions( $dbr );
return $dbr->selectRow( ... );
}
);

Example usage (key that is expensive and hot):

$catConfig = $cache->getWithSetCallback(
// Key to store the cached value under
$cache->makeKey( 'site-cat-config' ),
// Time-to-live (in seconds)
$cache::TTL_DAY,
// Function that derives the new key value
function ( $oldValue, &$ttl, array &$setOpts ) {
// Account for any snapshot/replica DB lag
$setOpts += Database::getCacheSetOptions( $dbr );
return CatConfig::newFromRow( $dbr->selectRow( ... ) );
},
[
// Calling touchCheckKey() on this key purges the cache
'checkKeys' => [ $cache->makeKey( 'site-cat-config' ) ],
// Try to only let one datacenter thread manage cache updates at a time
'lockTSE' => 30,
// Avoid querying cache servers multiple times in a web request
'pcTTL' => $cache::TTL_PROC_LONG
]
);

Example usage (key with dynamic dependencies):

// Key to store the cached value under
$cache->makeKey( 'cat-state', $cat->getId() ),
// Time-to-live (seconds)
$cache::TTL_HOUR,
// Function that derives the new key value
function ( $oldValue, &$ttl, array &$setOpts ) {
// Determine new value from the DB
// Account for any snapshot/replica DB lag
$setOpts += Database::getCacheSetOptions( $dbr );
return CatState::newFromResults( $dbr->select( ... ) );
},
[
// The "check" keys that represent things the value depends on;
// Calling touchCheckKey() on any of them purges the cache
'checkKeys' => [
$cache->makeKey( 'sustenance-bowls', $cat->getRoomId() ),
$cache->makeKey( 'people-present', $cat->getHouseId() ),
$cache->makeKey( 'cat-laws', $cat->getCityId() ),
]
]
);

Example usage (key that is expensive with too many DB dependencies for "check" keys):

// Key to store the cached value under
$cache->makeKey( 'cat-toys', $catId ),
// Time-to-live (seconds)
$cache::TTL_HOUR,
// Function that derives the new key value
function ( $oldValue, &$ttl, array &$setOpts ) {
// Determine new value from the DB
// Account for any snapshot/replica DB lag
$setOpts += Database::getCacheSetOptions( $dbr );
return CatToys::newFromResults( $dbr->select( ... ) );
},
[
// Get the highest timestamp of any of the cat's toys
'touchedCallback' => function ( $value ) use ( $catId ) {
$ts = $dbr->selectField( 'cat_toys', 'MAX(ct_touched)', ... );
return wfTimestampOrNull( TS_UNIX, $ts );
},
// Avoid DB queries for repeated access
'pcTTL' => $cache::TTL_PROC_SHORT
]
);
wfTimestampOrNull( $outputtype=TS_UNIX, $ts=null)
Return a formatted timestamp, or null if input is null.

Example usage (hot key holding most recent 100 events):

$lastCatActions = $cache->getWithSetCallback(
// Key to store the cached value under
$cache->makeKey( 'cat-last-actions', 100 ),
// Time-to-live (in seconds)
10,
// Function that derives the new key value
function ( $oldValue, &$ttl, array &$setOpts ) {
// Account for any snapshot/replica DB lag
$setOpts += Database::getCacheSetOptions( $dbr );
// Start off with the last cached list
$list = $oldValue ?: [];
// Fetch the last 100 relevant rows in descending order;
// only fetch rows newer than $list[0] to reduce scanning
$rows = iterator_to_array( $dbr->select( ... ) );
// Merge them and get the new "last 100" rows
return array_slice( array_merge( $new, $list ), 0, 100 );
},
[
// Try to only let one datacenter thread manage cache updates at a time
'lockTSE' => 30,
// Use a magic value when no cache value is ready rather than stampeding
'busyValue' => 'computing'
]
);

Example usage (key holding an LRU subkey:value map; this can avoid flooding cache with keys for an unlimited set of (constraint,situation) pairs, thereby avoiding elevated cache evictions and wasted memory):

$catSituationTolerabilityCache = $this->cache->getWithSetCallback(
// Group by constraint ID/hash, cat family ID/hash, or something else useful
$this->cache->makeKey( 'cat-situation-tolerability-checks', $groupKey ),
WANObjectCache::TTL_DAY, // rarely used groups should fade away
// The $scenarioKey format is $constraintId:<ID/hash of $situation>
function ( $cacheMap ) use ( $scenarioKey, $constraintId, $situation ) {
$lruCache = MapCacheLRU::newFromArray( $cacheMap ?: [], self::CACHE_SIZE );
$result = $lruCache->get( $scenarioKey ); // triggers LRU bump if present
if ( $result === null || $this->isScenarioResultExpired( $result ) ) {
$result = $this->checkScenarioTolerability( $constraintId, $situation );
$lruCache->set( $scenarioKey, $result, 3 / 8 );
}
// Save the new LRU cache map and reset the map's TTL
return $lruCache->toArray();
},
[
// Once map is > 1 sec old, consider refreshing
'ageNew' => 1,
// Update within 5 seconds after "ageNew" given a 1hz cache check rate
'hotTTR' => 5,
// Avoid querying cache servers multiple times in a request; this also means
// that a request can only alter the value of any given constraint key once
'pcTTL' => WANObjectCache::TTL_PROC_LONG
]
);
$tolerability = isset( $catSituationTolerabilityCache[$scenarioKey] )
? $catSituationTolerabilityCache[$scenarioKey]
: $this->checkScenarioTolerability( $constraintId, $situation );
static newFromArray(array $values, $maxKeys)
Definition: MapCacheLRU.php:77
See also
WANObjectCache::get()
WANObjectCache::set()
Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
int$ttlNominal seconds-to-live for newly computed values. Special values are:
  • WANObjectCache::TTL_INDEFINITE: Cache forever (subject to LRU-style evictions)
  • WANObjectCache::TTL_UNCACHEABLE: Do not cache (if the key exists, it is not deleted)
callable$callbackValue generation function
array$optsOptions map:
  • checkKeys: List of "check" keys. The key at $key will be seen as stale when either touchCheckKey() or resetCheckKey() is called on any of the keys in this list. This is useful if thousands or millions of keys depend on the same entity. The entity can simply have its "check" key updated whenever the entity is modified. Default: [].
  • graceTTL: If the key is purged (by "checkKeys" or "touchedCallback") less than this many seconds ago, consider reusing the stale value. The odds of a refresh become more likely over time, becoming certain once the grace period is reached. This can reduce traffic spikes when millions of keys are compared to the same "check" key and touchCheckKey() or resetCheckKey() is called on that "check" key. This option is not useful for avoiding traffic spikes in the case of the key simply expiring on account of its TTL (use "lowTTL" instead). Default: WANObjectCache::GRACE_TTL_NONE.
  • lockTSE: Prefer the use of a mutex during value regeneration of the key if its TSE ("time since expiry") is less than the given number of seconds ago. The TSE of the key is influenced by purges (e.g. via delete(), "checkKeys", and "touchedCallback"), and various other options (e.g. "staleTTL"). A low enough TSE is assumed to indicate a high enough key access rate to justify stampede avoidance. A thread that tries and fails to acquire the mutex will use a stale value for the key, if there is one, and, if not, it will execute the callback. Note that no cache value exists after deletion or storage-layer expiration/eviction; to prevent stampedes during these cases, avoid using delete(), keep "lowTTL" enabled, and consider using "busyValue". Default: WANObjectCache::TSE_NONE.
  • busyValue: Specify a placeholder value to use when no value exists and another thread is currently regenerating it. This assures that cache stampedes cannot happen if the value falls out of cache. This also mitigates stampedes when value regeneration becomes very slow (greater than $ttl/"lowTTL"). If this is a closure, then it will be invoked to get the placeholder when needed. Default: null.
  • pcTTL: Process cache the value in this PHP instance for this many seconds. This avoids network I/O when a key is read several times. This will not cache when the callback returns false, however. Note that any purges will not be seen while process cached; since the callback should use replica DBs and they may be lagged or have snapshot isolation anyway, this should not typically matter. Default: WANObjectCache::TTL_UNCACHEABLE.
  • pcGroup: Process cache group to use instead of the primary one. If set, this must be of the format ALPHANUMERIC_NAME:MAX_KEY_SIZE, e.g. "mydata:10". Use this for storing large values, small yet numerous values, or some values with a high cost of eviction. It is generally preferable to use a class constant when setting this value. This has no effect unless pcTTL is used. Default: WANObjectCache::PC_PRIMARY.
  • version: Integer version number. This lets callers make breaking changes to the format of cached values without causing problems for sites that use non-instantaneous code deployments. Old and new code will recognize incompatible versions and purges from both old and new code will been seen by each other. When this method encounters an incompatibly versioned value at the provided key, a "variant key" will be used for reading from and saving to cache. The variant key is specific to the key and version number provided to this method. If the variant key value is older than that of the provided key, or the provided key is non-existant, then the variant key will be seen as non-existant. Therefore, delete() calls purge the provided key's variant keys. The "checkKeys" and "touchedCallback" options still apply to variant keys as usual. Avoid storing class objects, as this reduces compatibility (due to serialization). Default: null.
  • minAsOf: Reject values if they were generated before this UNIX timestamp. This is useful if the source of a key is suspected of having possibly changed recently, and the caller wants any such changes to be reflected. Default: WANObjectCache::MIN_TIMESTAMP_NONE.
  • hotTTR: Expected time-till-refresh (TTR) in seconds for keys that average ~1 hit per second (e.g. 1Hz). Keys with a hit rate higher than 1Hz will refresh sooner than this TTR and vise versa. Such refreshes won't happen until keys are "ageNew" seconds old. This uses randomization to avoid triggering cache stampedes. The TTR is useful at reducing the impact of missed cache purges, since the effect of a heavily referenced key being stale is worse than that of a rarely referenced key. Unlike simply lowering $ttl, seldomly used keys are largely unaffected by this option, which makes it possible to have a high hit rate for the "long-tail" of less-used keys. Default: WANObjectCache::HOT_TTR.
  • lowTTL: Consider pre-emptive updates when the current TTL (seconds) of the key is less than this. It becomes more likely over time, becoming certain once the key is expired. This helps avoid cache stampedes that might be triggered due to the key expiring. Default: WANObjectCache::LOW_TTL.
  • ageNew: Consider popularity refreshes only once a key reaches this age in seconds. Default: WANObjectCache::AGE_NEW.
  • staleTTL: Seconds to keep the key around if it is stale. This means that on cache miss the callback may get $oldValue/$oldAsOf values for keys that have already been expired for this specified time. This is useful if adaptiveTTL() is used on the old value's as-of time when it is verified as still being correct. Default: WANObjectCache::STALE_TTL_NONE
  • touchedCallback: A callback that takes the current value and returns a UNIX timestamp indicating the last time a dynamic dependency changed. Null can be returned if there are no relevant dependency changes to check. This can be used to check against things like last-modified times of files or DB timestamp fields. This should generally not be used for small and easily queried values in a DB if the callback itself ends up doing a similarly expensive DB query to check a timestamp. Usages of this option makes the most sense for values that are moderately to highly expensive to regenerate and easy to query for dependency timestamps. The use of "pcTTL" reduces timestamp queries. Default: null.
array$cbParamsCustom field/value map to pass to the callback (since 1.35) @phpcs:ignore Generic.Files.LineLength
Returns
mixed Value found or written to the key
Note
Options added in 1.28: version, busyValue, hotTTR, ageNew, pcGroup, minAsOf
Options added in 1.31: staleTTL, graceTTL
Options added in 1.33: touchedCallback
Callable type hints are not used to avoid class-autoloading

Definition at line 1502 of file WANObjectCache.php.

References makeGlobalKey().

Referenced by getMultiWithSetCallback(), and getMultiWithUnionSetCallback().

◆ hash256()

WANObjectCache::hash256 (   $component)

Hash a possibly long string into a suitable component for makeKey()/makeGlobalKey()

Parameters
string$componentA raw component used in building a cache key
Returns
string 64 character HMAC using a stable secret for public collision resistance
Since
1.34

Definition at line 2348 of file WANObjectCache.php.

◆ isLotteryRefreshDue()

WANObjectCache::isLotteryRefreshDue (   $res,
  $lowTTL,
  $ageNew,
  $hotTTR,
  $now 
)
protected

Check if a key is due for randomized regeneration due to near-expiration/popularity.

Parameters
array$resCurrent value WANObjectCache::RES_* data map
float$lowTTLConsider a refresh when $curTTL is less than this; the "low" threshold
int$ageNewAge of key when this might recommend refreshing (seconds)
int$hotTTRAge of key when it should be refreshed if popular (seconds)
float$nowThe current UNIX timestamp
Returns
bool

Definition at line 2793 of file WANObjectCache.php.

References $res, worthRefreshExpiring(), and worthRefreshPopular().

◆ isValid()

WANObjectCache::isValid (   $value,
  $asOf,
  $minAsOf 
)
protected

Check that a wrapper value exists and has an acceptable age.

Parameters
array | false$valueValue wrapper or false
float$asOfValue generation "as of" timestamp
float$minAsOfMinimum acceptable value "as of" UNIX timestamp
Returns
bool

Definition at line 2889 of file WANObjectCache.php.

◆ makeGlobalKey()

WANObjectCache::makeGlobalKey (   $collection,
  $components 
)

Make a cache key for the global keyspace and given components.

See also
IStoreKeyEncoder::makeGlobalKey()
Parameters
string$collectionKey collection name component
string|int...$components Additional, ordered, key components for entity IDs
Returns
string Colon-separated, keyspace-prepended, ordered list of encoded components
Since
1.27

Implements IStoreKeyEncoder.

Definition at line 2323 of file WANObjectCache.php.

Referenced by getWithSetCallback().

◆ makeKey()

WANObjectCache::makeKey (   $collection,
  $components 
)

Make a cache key using the "global" keyspace for the given components.

See also
IStoreKeyEncoder::makeKey()
Parameters
string$collectionKey collection name component
string|int...$components Additional, ordered, key components for entity IDs
Returns
string Colon-separated, keyspace-prepended, ordered list of encoded components
Since
1.27

Implements IStoreKeyEncoder.

Definition at line 2337 of file WANObjectCache.php.

◆ makeMultiKeys()

WANObjectCache::makeMultiKeys ( array  $ids,
  $keyCallback 
)
final

Get an iterator of (cache key => entity ID) for a list of entity IDs.

The $callback argument expects a function that returns the key for an entity ID via makeKey()/makeGlobalKey(). There should be no network nor filesystem I/O used in the callback. The entity ID/key mapping must be 1:1 or an exception will be thrown. Use the hash256() method for any hashing. The callback takes the following arguments:

Example usage for the default keyspace:

$keyedIds = $cache->makeMultiKeys(
function ( $module, $cache ) {
return $cache->makeKey( 'example-module', $module );
}
);
$modules

Example usage for mixed default and global keyspace:

$keyedIds = $cache->makeMultiKeys(
$filters,
function ( $filter, $cache ) {
return self::isCentral( $filter )
? $cache->makeGlobalKey( 'example-filter', $filter )
: $cache->makeKey( 'example-filter', $filter )
}
);

Example usage with hashing:

$keyedIds = $cache->makeMultiKeys(
$urls,
function ( $url, $cache ) {
return $cache->makeKey( 'example-url', $cache->hash256( $url ) );
}
);
See also
WANObjectCache::makeKey()
WANObjectCache::makeGlobalKey()
WANObjectCache::hash256()
Parameters
string[] | int[]$idsList of entity IDs
callable$keyCallbackFunction returning makeKey()/makeGlobalKey() on the input ID
Returns
ArrayIterator Iterator of (cache key => ID); order of $ids is preserved
Since
1.28

Definition at line 2403 of file WANObjectCache.php.

◆ multiRemap()

WANObjectCache::multiRemap ( array  $ids,
array  $res 
)
final

Get an (ID => value) map from (i) a non-unique list of entity IDs, and (ii) the list of corresponding entity values by first appearance of each ID in the entity ID list.

For use with getMultiWithSetCallback() and getMultiWithUnionSetCallback().

Only use this method if the entity ID/key mapping is trivially 1:1 without exception. Key generation method must utilize the full entity ID in the key (not a hash of it).

Example usage:

$poems = $cache->getMultiWithSetCallback(
$cache->makeMultiKeys(
$uuids,
function ( $uuid ) use ( $cache ) {
return $cache->makeKey( 'poem', $uuid );
}
),
$cache::TTL_DAY,
function ( $uuid ) use ( $url ) {
return $this->http->run( [ 'method' => 'GET', 'url' => "$url/$uuid" ] );
}
);
$poemsByUUID = $cache->multiRemap( $uuids, $poems );
See also
WANObjectCache::makeMultiKeys()
WANObjectCache::getMultiWithSetCallback()
WANObjectCache::getMultiWithUnionSetCallback()
Parameters
string[] | int[]$idsEntity ID list makeMultiKeys()
mixed[]$resResult of getMultiWithSetCallback()/getMultiWithUnionSetCallback()
Returns
mixed[] Map of (ID => value); order of $ids is preserved
Since
1.34

Definition at line 2459 of file WANObjectCache.php.

References $res.

◆ newEmpty()

static WANObjectCache::newEmpty ( )
static

◆ prependRoute()

WANObjectCache::prependRoute ( string  $sisterKey,
string  $route 
)
protected
Parameters
string$sisterKey
string$routeKey routing prefix
Returns
string

Definition at line 2707 of file WANObjectCache.php.

Referenced by relayNonVolatilePurge(), and relayVolatilePurges().

◆ reap()

WANObjectCache::reap (   $key,
  $purgeTimestamp,
$isStale = false 
)
final

Set a key to soon expire in the local cluster if it pre-dates $purgeTimestamp.

This sets stale keys' time-to-live at HOLDOFF_TTL seconds, which both avoids broadcasting in mcrouter setups and also avoids races with new tombstones.

Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
int | float$purgeTimestampUNIX timestamp of purge
bool&$isStaleWhether the key is stale
Returns
bool Success
Since
1.28

Definition at line 2260 of file WANObjectCache.php.

References HOLDOFF_TTL.

◆ reapCheckKey()

WANObjectCache::reapCheckKey (   $key,
  $purgeTimestamp,
$isStale = false 
)
final

Set a "check" key to soon expire in the local cluster if it pre-dates $purgeTimestamp.

Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
int$purgeTimestampUNIX timestamp of purge
bool&$isStaleWhether the key is stale
Returns
bool Success
Since
1.28

Definition at line 2292 of file WANObjectCache.php.

◆ relayNonVolatilePurge()

WANObjectCache::relayNonVolatilePurge ( string  $sisterKey)
protected

Remove a sister key from all datacenters.

This method should not wait for the operation to complete on remote datacenters

Parameters
string$sisterKeyA value key or "check" key
Returns
bool Success

Definition at line 2692 of file WANObjectCache.php.

References prependRoute().

Referenced by delete(), and resetCheckKey().

◆ relayVolatilePurges()

WANObjectCache::relayVolatilePurges ( array  $purgeBySisterKey,
int  $ttl 
)
protected

Set a sister key to a purge value in all datacenters.

This method should not wait for the operation to complete on remote datacenters

Since older purge values can sometimes arrive after newer ones, use a relative expiry so that even if the older value replaces the newer value, the TTL will greater than the remaining TTL on the older value (assuming that all purges for a key use the same TTL).

Parameters
array<string,string>$purgeBySisterKey Map of (sister key => result of makeTombstonePurgeValue()/makeCheckKeyPurgeValue())
int$ttlSeconds to keep the purge value around
Returns
bool Success

Definition at line 2663 of file WANObjectCache.php.

References prependRoute().

Referenced by delete(), and touchCheckKey().

◆ resetCheckKey()

WANObjectCache::resetCheckKey (   $key)
final

Clear the last-purge timestamp of a "check" key in all datacenters.

Similar to touchCheckKey(), in that keys fetched using get*() calls, that include the given "check" key, will be seen as purged. However, there are some differences:

  • a) The "check" key will be deleted from all caches and lazily re-initialized when accessed (rather than set everywhere)
  • b) Thus, dependent keys will be known to be stale, but not for how long (they are treated as "just" purged), which effects any lockTSE logic in getWithSetCallback()
  • c) Since "check" keys are initialized only on the server the key hashes to, any temporary ejection of that server will cause the value to be seen as purged as a new server will initialize the "check" key.

The advantage over touchCheckKey() is that the "check" keys, which have high TTLs, will only be created when a get*() method actually uses those keys. This is better when a large number of "check" keys must be changed in a short period of time.

Note that "check" keys won't collide with other regular keys.

See also
WANObjectCache::get()
WANObjectCache::getWithSetCallback()
WANObjectCache::touchCheckKey()
Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
Returns
bool True if the item was purged or not found, false on failure

Definition at line 1189 of file WANObjectCache.php.

References relayNonVolatilePurge().

◆ set()

WANObjectCache::set (   $key,
  $value,
  $ttl = self::TTL_INDEFINITE,
array  $opts = [] 
)
final

Set the value of a key in cache.

Simply calling this method when source data changes is not valid because the changes do not replicate to the other WAN sites. In that case, delete() should be used instead. This method is intended for use on cache misses.

If data was read using "view snapshots" (e.g. innodb REPEATABLE-READ), use 'since' to avoid the following race condition:

  • a) T1 starts
  • b) T2 updates a row, calls delete(), and commits
  • c) The HOLDOFF_TTL passes, expiring the delete() tombstone
  • d) T1 reads the row and calls set() due to a cache miss
  • e) Stale value is stuck in cache

Setting 'lag' and 'since' help avoids keys getting stuck in stale states.

Be aware that this does not update the process cache for getWithSetCallback() callers. Keys accessed via that method are not generally meant to also be set using this primitive method.

Consider using getWithSetCallback(), which has cache slam avoidance and key versioning features, instead of bare get()/set() calls.

Do not use this method on versioned keys accessed via getWithSetCallback().

Example usage:

$setOpts = Database::getCacheSetOptions( $dbr );
// Fetch the row from the DB
$row = $dbr->selectRow( ... );
$key = $cache->makeKey( 'building', $buildingId );
$cache->set( $key, $row, $cache::TTL_DAY, $setOpts );
set( $key, $value, $exptime=0, $flags=0)
Set an item.
Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
mixed$valueValue to set for the cache key
int$ttlSeconds to live. Special values are:
  • WANObjectCache::TTL_INDEFINITE: Cache forever (default)
  • WANObjectCache::TTL_UNCACHEABLE: Do not cache (if the key exists, it is not deleted)
array$optsOptions map:
  • lag: Highest seconds of replication lag potentially affecting reads used to generate the value. This should not be affected by the duration of transaction "view snapshots" (e.g. innodb REPEATABLE-READ) nor the time elapsed since the first read (though both increase staleness). For reads using view snapshots, only the replication lag during snapshot initialization matters. Use false if replication is stopped/broken on a replica server involved in the reads. Default: 0 seconds
  • since: UNIX timestamp indicative of the highest possible staleness caused by the duration of transaction "view snapshots" (e.g. innodb REPEATABLE-READ) and the time elapsed since the first read. This should not be affected by replication lag. Default: 0 seconds
  • pending: Whether this data is possibly from an uncommitted write transaction. Generally, other threads should not see values from the future and they certainly should not see ones that ended up getting rolled back. Default: false
  • lockTSE: If excessive replication/snapshot lag is detected, then store the value with this TTL and flag it as stale. This is only useful if the reads for this key use getWithSetCallback() with "lockTSE" set. Note that if "staleTTL" is set then it will still add on to this TTL in the excessive lag scenario. Default: WANObjectCache::TSE_NONE
  • staleTTL: Seconds to keep the key around if it is stale. The get()/getMulti() methods return such stale values with a $curTTL of 0, and getWithSetCallback() will call the generation callback in such cases, passing in the old value and its as-of time to the callback. This is useful if adaptiveTTL() is used on the old value's as-of time when it is verified as still being correct. Default: WANObjectCache::STALE_TTL_NONE
  • creating: Optimize for the case where the key does not already exist. Default: false
  • version: Integer version number signifying the format of the value. Default: null
  • walltime: How long the value took to generate in seconds. Default: null @phpcs:ignore Generic.Files.LineLength
Note
Options added in 1.28: staleTTL
Options added in 1.33: creating
Options added in 1.34: version, walltime
Returns
bool Success

Definition at line 785 of file WANObjectCache.php.

References $cache, getCurrentTime(), STALE_TTL_NONE, and TTL_LAGGED.

◆ setLogger()

WANObjectCache::setLogger ( LoggerInterface  $logger)
Parameters
LoggerInterface$logger

Definition at line 381 of file WANObjectCache.php.

References $logger.

Referenced by __construct().

◆ setMockTime()

WANObjectCache::setMockTime ( $time)
Parameters
float | null&$timeMock UNIX timestamp for testing

Definition at line 3145 of file WANObjectCache.php.

◆ touchCheckKey()

WANObjectCache::touchCheckKey (   $key,
  $holdoff = self::HOLDOFF_TTL 
)
final

Increase the last-purge timestamp of a "check" key in all datacenters.

This method should only be called when some heavily referenced data changes in a significant way, such that it is impractical to call delete() on all the cache keys that should be purged. The get*() method calls used to fetch these keys must include the given "check" key in the relevant "check" keys argument/option.

A "check" key essentially represents a last-modified time of an entity. When the key is touched, the timestamp will be updated to the current time. Keys fetched using get*() calls, that include the "check" key, will be seen as purged.

The timestamp of the "check" key is treated as being HOLDOFF_TTL seconds in the future by get*() methods in order to avoid race conditions where keys are updated with stale values (e.g. from a lagged replica DB). A high TTL is set on the "check" key, making it possible to know the timestamp of the last change to the corresponding entities in most cases. This might use more cache space than resetCheckKey().

When a few important keys get a large number of hits, a high cache time is usually desired as well as "lockTSE" logic. The resetCheckKey() method is less appropriate in such cases since the "time since expiry" cannot be inferred, causing any get() after the reset to treat the key as being "hot", resulting in more stale value usage.

Note that "check" keys won't collide with other regular keys.

See also
WANObjectCache::get()
WANObjectCache::getWithSetCallback()
WANObjectCache::resetCheckKey()
Parameters
string$keyCache key made with makeKey()/makeGlobalKey()
int$holdoffHOLDOFF_TTL or HOLDOFF_TTL_NONE constant
Returns
bool True if the item was purged or not found, false on failure

Definition at line 1149 of file WANObjectCache.php.

References getCurrentTime(), and relayVolatilePurges().

◆ useInterimHoldOffCaching()

WANObjectCache::useInterimHoldOffCaching (   $enabled)
final

Enable or disable the use of brief caching for tombstoned keys.

When a key is purged via delete(), there normally is a period where caching is hold-off limited to an extremely short time. This method will disable that caching, forcing the callback to run for any of:

This is useful when both:

  • a) the database used by the callback is known to be up-to-date enough for some particular purpose (e.g. replica DB has applied transaction X)
  • b) the caller needs to exploit that fact, and therefore needs to avoid the use of inherently volatile and possibly stale interim keys
See also
WANObjectCache::delete()
Parameters
bool$enabledWhether to enable interim caching
Since
1.31

Definition at line 2550 of file WANObjectCache.php.

◆ watchErrors()

WANObjectCache::watchErrors ( )

Get a "watch point" token that can be used to get the "last error" to occur after now.

Returns
int A token that the current error event
Since
1.38

Definition at line 2478 of file WANObjectCache.php.

◆ worthRefreshExpiring()

WANObjectCache::worthRefreshExpiring (   $curTTL,
  $logicalTTL,
  $lowTTL 
)
protected

Check if a key is nearing expiration and thus due for randomized regeneration.

If $curTTL is greater than the "low" threshold (e.g. not nearing expiration) then this returns false. If $curTTL <= 0 (e.g. value already expired), then this returns false. Otherwise, the chance of this returning true increases steadily from 0% to 100% as $curTTL moves from the "low" threshold down to 0 seconds.

The logical TTL will be used as the "low" threshold if it is less than $lowTTL.

This method uses deadline-aware randomization in order to handle wide variations of cache access traffic without the need for configuration or expensive state.

Parameters
float$curTTLApproximate TTL left on the key
float$logicalTTLFull logical TTL assigned to the key
float$lowTTLConsider a refresh when $curTTL is less than this; the "low" threshold
Returns
bool

Definition at line 2863 of file WANObjectCache.php.

Referenced by isLotteryRefreshDue().

◆ worthRefreshPopular()

WANObjectCache::worthRefreshPopular (   $asOf,
  $ageNew,
  $timeTillRefresh,
  $now 
)
protected

Check if a key is due for randomized regeneration due to its popularity.

This is used so that popular keys can preemptively refresh themselves for higher consistency (especially in the case of purge loss/delay). Unpopular keys can remain in cache with their high nominal TTL. This means popular keys keep good consistency, whether the data changes frequently or not, and long-tail keys get to stay in cache and get hits too. Similar to worthRefreshExpiring(), randomization is used.

Parameters
float$asOfUNIX timestamp of the value
int$ageNewAge of key when this might recommend refreshing (seconds)
int$timeTillRefreshAge of key when it should be refreshed if popular (seconds)
float$nowThe current UNIX timestamp
Returns
bool

Definition at line 2819 of file WANObjectCache.php.

Referenced by isLotteryRefreshDue().

Member Data Documentation

◆ $asyncHandler

callable null WANObjectCache::$asyncHandler
protected

Function that takes a WAN cache callback and runs it later.

Definition at line 142 of file WANObjectCache.php.

◆ $broadcastRoute

string null WANObjectCache::$broadcastRoute
protected

Routing prefix for operations that should be broadcasted to all data centers.

If null, the there is only one datacenter or a backend proxy broadcasts everything.

Definition at line 151 of file WANObjectCache.php.

◆ $cache

BagOStuff WANObjectCache::$cache
protected

The local datacenter cache.

Definition at line 134 of file WANObjectCache.php.

Referenced by set().

◆ $coalesceScheme

int WANObjectCache::$coalesceScheme
protected

Scheme to use for key coalescing (Hash Tags or Hash Stops)

Definition at line 159 of file WANObjectCache.php.

◆ $epoch

float WANObjectCache::$epoch
protected

Unix timestamp of the oldest possible valid values.

Definition at line 155 of file WANObjectCache.php.

◆ $logger

LoggerInterface WANObjectCache::$logger
protected

Definition at line 138 of file WANObjectCache.php.

Referenced by setLogger().

◆ $processCaches

MapCacheLRU [] WANObjectCache::$processCaches = []
protected

Map of group PHP instance caches.

Definition at line 136 of file WANObjectCache.php.

◆ $secret

string WANObjectCache::$secret
protected

Stable secret used for hashing long strings into key components.

Definition at line 157 of file WANObjectCache.php.

◆ $stats

StatsdDataFactoryInterface WANObjectCache::$stats
protected

Definition at line 140 of file WANObjectCache.php.

◆ $useInterimHoldOffCaching

bool WANObjectCache::$useInterimHoldOffCaching = true
protected

Whether to use "interim" caching while keys are tombstoned.

Definition at line 153 of file WANObjectCache.php.

◆ GRACE_TTL_NONE

const WANObjectCache::GRACE_TTL_NONE = 0

Idiom for set()/getWithSetCallback() meaning "no post-expiration grace period".

Definition at line 202 of file WANObjectCache.php.

◆ HOLDOFF_TTL

const WANObjectCache::HOLDOFF_TTL = self::MAX_COMMIT_DELAY + self::MAX_READ_LAG + 1

Seconds to tombstone keys on delete() and to treat keys as volatile after purges.

Definition at line 184 of file WANObjectCache.php.

Referenced by WANObjectCacheReaper\invoke(), and reap().

◆ HOLDOFF_TTL_NONE

const WANObjectCache::HOLDOFF_TTL_NONE = 0

Idiom for delete()/touchCheckKey() meaning "no hold-off period".

Definition at line 204 of file WANObjectCache.php.

◆ KEY_AS_OF

const WANObjectCache::KEY_AS_OF = 'asOf'

Generation completion timestamp attribute for a key; keep value for b/c (< 1.36)

Definition at line 258 of file WANObjectCache.php.

◆ KEY_CHECK_AS_OF

const WANObjectCache::KEY_CHECK_AS_OF = 'lastCKPurge'

Highest "check" key timestamp for a key; keep value for b/c (< 1.36)

Definition at line 266 of file WANObjectCache.php.

◆ KEY_CUR_TTL

const WANObjectCache::KEY_CUR_TTL = 'curTTL'

Remaining TTL attribute for a key; keep value for b/c (< 1.36)

Definition at line 262 of file WANObjectCache.php.

◆ KEY_TOMB_AS_OF

const WANObjectCache::KEY_TOMB_AS_OF = 'tombAsOf'

Tomstone timestamp attribute for a key; keep value for b/c (< 1.36)

Definition at line 264 of file WANObjectCache.php.

◆ KEY_TTL

const WANObjectCache::KEY_TTL = 'ttl'

Logical TTL attribute for a key.

Definition at line 260 of file WANObjectCache.php.

◆ KEY_VERSION

const WANObjectCache::KEY_VERSION = 'version'

Version number attribute for a key; keep value for b/c (< 1.36)

Definition at line 256 of file WANObjectCache.php.

◆ PASS_BY_REF

const WANObjectCache::PASS_BY_REF = []

Idiom for get()/getMulti() to return extra information by reference.

Definition at line 213 of file WANObjectCache.php.

Referenced by get(), and getMulti().

◆ STALE_TTL_NONE

const WANObjectCache::STALE_TTL_NONE = 0

Idiom for set()/getWithSetCallback() meaning "no post-expiration persistence".

Definition at line 200 of file WANObjectCache.php.

Referenced by set().

◆ TTL_LAGGED

const WANObjectCache::TTL_LAGGED = 30

Max TTL, in seconds, to store keys when a data source has high replication lag.

Definition at line 189 of file WANObjectCache.php.

Referenced by set().


The documentation for this class was generated from the following file: