org.hd.d.pg2k.svrCore
Class DuplicateIDChecker<K>

java.lang.Object
  org.hd.d.pg2k.svrCore.DuplicateIDChecker<K>

All Implemented Interfaces:

MemoryTools.Compactable

public final class DuplicateIDChecker<K>
extends java.lang.Object
implements MemoryTools.Compactable

Mechanism to check for replayed/duplicate messages by ID within a specified time window. This provides an optional fixed upper size; attempts to insert more items than the ceiling capacity will result in old items being removed in age order regardless of age.

Other after meeting any size limit, this always keeps entries for a specified minimum time. Optionally a larger time window may be used if the the system is not starved of memory.

Thread-safe.

A lock can be held on instances of this object to make compound operations atomic.

Does not support the full Map interface.

This aggressively trims dead/expired entries so as to minimise footprint.

DEFAULT_LOAD_FACTOR

public static final float DEFAULT_LOAD_FACTOR

Default load factor.

See Also:

Constant Field Values

lhm

private final java.util.LinkedHashMap<K,java.lang.Long> lhm

Underlying LinkedHashMap on which this is based; never null.

minRetentionMs

private final int minRetentionMs

Maximum age that we will retain IDs for (ms); strictly positive.

preferredRetentionMs

private final int preferredRetentionMs

Optional higher maximum age that we will retain IDs for when there's plenty of memory available (ms); strictly positive. Ignored unless higher than minRetentionMs.

name

private final java.lang.String name

Name of this instance for diagnostics purposes; null if none.

DuplicateIDChecker

private DuplicateIDChecker(int minRetentionMs,
                           int preferredRetentionMs,
                           int maxCapacity,
                           float loadFactor,
                           java.lang.String name)

Create an instance with the given parameters.

Parameters:

minRetentionMs - the minimum time that we will keep an ID for (ms); strictly positive

preferredRetentionMs - preferred time to retain entries if there is lots of free memory and ineffective/ignored unless larger than minRetentionMs; strictly positive

maxCapacity - the maximum capacity; strictly positive

loadFactor - the load factor of the underlying hash table; in the range 0.0f to 1.0f exclusive (typically ~0.7f)

create

public static <K> DuplicateIDChecker<K> create(int minRetentionMs,
                                               java.lang.String name)

Create an instance with the given parameters. The standard default is used for the load factor, and the map size is effectively unlimited.

Parameters:

minRetentionMs - the minimum time that we will keep an ID for (ms); strictly positive

name - name of this instance for diagnostics purposes; null if none

create

public static <K> DuplicateIDChecker<K> create(int minRetentionMs,
                                               int preferredRetentionMs,
                                               java.lang.String name)

Create an instance with the given parameters. The standard default is used for the load factor, and the map size is effectively unlimited.

Parameters:

minRetentionMs - the minimum time that we will keep an ID for (ms); strictly positive

preferredRetentionMs - preferred time to retain entries if there is plenty of free memory and ineffective/ignored unless larger than minRetentionMs; strictly positive

name - name of this instance for diagnostics purposes; null if none

create

public static <K> DuplicateIDChecker<K> create(int minRetentionMs,
                                               int preferredRetentionMs,
                                               int maxCapacity,
                                               float loadFactor,
                                               java.lang.String name)

Create an instance with the given parameters.

Parameters:

minRetentionMs - the minimum time that we will keep an ID for (ms); strictly positive

preferredRetentionMs - preferred time to retain entries if there is plenty of free memory and ineffective/ignored unless larger than minRetentionMs; strictly positive

maxCapacity - the maximum capacity; strictly positive

loadFactor - the load factor of the underlying hash table; in the range 0.0f to 1.0f exclusive (typically ~0.7f)

name - name of this instance for diagnostics purposes; null if none

compact

public void compact()

Try to trim all dead entries to reduce memory footprint. Will attempt to minimise memory footprint.

Specified by:

compact in interface MemoryTools.Compactable

_compact

public void _compact()

Try to trim all dead entries to reduce memory footprint. Will attempt to minimise memory footprint.

Must only be called while the instance lock is held.

containsKey

public boolean containsKey(K key)

Returns true if we have a mapping from the supplied key (message ID), false otherwise. This is equivalent to checking if this is a duplicate message.

This ignores expired values and does not add anything to the collection.

add

public java.lang.Long add(K key)

Record an ID, returning the insertion time of the previous insertion of this ID if any.

remove

public java.lang.Long remove(K key)

Remove an ID and return the insertion time of the removed ID, if any, else null.

ageOfOldestEntry

public java.lang.Long ageOfOldestEntry()

Returns the timestamp of the oldest entry currently in this Map, or null if none.

clear

public void clear()

Clear the structure.

size

public int size()

Return the number of non-expired IDs held; non-negative.

toString

public java.lang.String toString()

Provide a human-readable summary of status. Useful for debugging/tuning, for example.

Overrides:

toString in class java.lang.Object

hasExpired

private boolean hasExpired(java.util.Map.Entry<K,java.lang.Long> entry,
                           long thresholdTime)

Returns true when an entry has expired and can be removed.

Parameters:

entry - full entry including the timestamp; never null

thresholdTime - expiry threshold time to use, an older entry should be expired

thresholdTime

private long thresholdTime()

Compute threshold time for entry expiry. Lifetime drops as free space falls beneath our target.

This is made proportional so that transients are less likely to discard a slow-to-regain acceptance window.

hasExpired

private boolean hasExpired(java.util.Map.Entry<K,java.lang.Long> entry)

Returns true when an entry has expired and can be removed. Computes the expiry threshold itself, so is suitable for one-off expiry attempts.

Parameters:

entry - full entry including the timestamp; never null