org.gridgain.grid.dr.cache.receiver

Interface GridDrReceiverCacheConflictContext<K,V>

All Known Implementing Classes:

GridDrReceiverConflictContextImpl

public interface GridDrReceiverCacheConflictContext<K,V>

Data center replication receiver cache conflict context. Whenever conflict which cannot be resolved automatically occurs this entity is passed to GridDrReceiverCacheConflictResolver to perform explicit conflict resolution.

Contains all data necessary for conflict resolution.

Method Summary

Methods

Modifier and Type	Method and Description
void	merge(V mergeVal, long ttl) Force data center replication engine to use neither old, nor new, but some other value passed as argument.
GridDrEntry<K,V>	newEntry() Gets new cache entry.
GridDrEntry<K,V>	oldEntry() Gets old (existing) cache entry.
void	useNew() Force data center replication receiver cache to apply new entry thus overwriting old (existing) entry.
void	useOld() Force data center replication receiver cache to ignore new entry and leave old (existing) entry unchanged.

Method Detail

oldEntry

GridDrEntry<K,V> oldEntry()

Gets old (existing) cache entry.

Returns:

Old (existing) cache entry.

newEntry

GridDrEntry<K,V> newEntry()

Gets new cache entry.

Returns:

New cache entry.

useOld

void useOld()

Force data center replication receiver cache to ignore new entry and leave old (existing) entry unchanged.

useNew

void useNew()

Force data center replication receiver cache to apply new entry thus overwriting old (existing) entry.

Note that updates from remote data centers always have explicit TTL , while local data center updates will only have explicit TTL in case GridCacheEntry.timeToLive(long) was called before update. In the latter case new entry will pick TTL of the old (existing) entry, even if it was set through update from remote data center. it means that depending on concurrent update timings new update might pick unexpected TTL. For example, consider that three updates of the same key are performed: local update with explicit TTL (1) followed by another local update without explicit TTL (2) and one remote update (3). In this case you might expect that update (2) will pick TTL set during update (1). However, in case update (3) occurrs between (1) and (2) and it overwrites (1) during conflict resolution, then update (2) will pick TTL of update (3). To have predictable TTL in such cases you should either always set it explicitly through GridCacheEntry.timeToLive(long) or use merge(Object, long).

merge

void merge(@Nullable
         V mergeVal,
         long ttl)

Force data center replication engine to use neither old, nor new, but some other value passed as argument. In this case old value will be replaced with merge value and update will be considered as local (i.e. GridConfiguration.getDataCenterId() will be used).

Also in case of merge you have to specify new TTL explicitly. For unlimited TTL use 0.

Parameters:

mergeVal - Merge value or null to force remove.

ttl - Time to live in milliseconds.