leventov
|
01f15fb9b1
|
Fix msync JNA implementations and add SimplePersistedMapOverflowTest
|
2016-02-05 16:35:46 +07:00 |
|
leventov
|
299cd8341c
|
HCOLL-397 msync before pointing to newly created tier bulk in the header
|
2016-01-31 21:16:42 +07:00 |
|
leventov
|
8891626bb2
|
Don't force writing the memory to disk in VanillaChronicleHash.close(), release the virtual memory instead
|
2016-01-31 19:10:18 +07:00 |
|
leventov
|
0d93fb1d22
|
Rename VanillaChronicleHash.tierHashLookupEntrySize field to tierHashLookupSlotSize, update the memory layout specification, and key hashing and checksum algorithms specification
|
2016-01-30 19:20:50 +07:00 |
|
leventov
|
040a8d69b6
|
Simplify entry checksum computation for not replicated ChronicleSet
|
2016-01-30 17:10:33 +07:00 |
|
leventov
|
69337322ee
|
Simplify VanillaChronicleHash.roundUpMapHeaderSize() because rational behind former complexity (considering buffering in Map header serializer) is 1) gone (because BinaryWire is used instead of built-in serialization 2) was incorrect anyway, because rounding up doesn't prevent problems with buffering
|
2016-01-21 15:13:29 +03:00 |
|
leventov
|
0ee9b64978
|
Rename VanillaChronicleHash.segmentSize() to tierSize()
|
2016-01-20 21:56:36 +03:00 |
|
leventov
|
72ba0e4d57
|
Move HashSplitting.Splitting.forSegments() method to HashSplitting interface, because interfaces could contain static methods since Java 8
|
2016-01-19 18:36:28 +03:00 |
|
leventov
|
b06c64a47c
|
Don't serialize HashSplitting$ForPowerOf2Segments.mask field
|
2016-01-19 18:34:09 +03:00 |
|
leventov
|
bb22985d9c
|
Remove entriesPerSegment field from VanillaChronicleHash
|
2016-01-19 14:17:38 +03:00 |
|
leventov
|
8055a8b1ae
|
Make private BigSegmentHeader's methods static
|
2016-01-18 17:39:55 +03:00 |
|
leventov
|
0ead764edf
|
BigSegmentHeader: try deregister wait on try-upgrade-update-to-write procedure, if it sees unexpected state (update lock is not held)
|
2016-01-18 17:34:14 +03:00 |
|
leventov
|
7e32081111
|
Add missing firstFreeTierIndex reset in globalMutableState, in SegmentsRecovery
|
2016-01-17 17:44:23 +03:00 |
|
leventov
|
56e5e93388
|
Hide the remaining parts of the replication API from the public eye
|
2016-01-17 17:08:13 +03:00 |
|
leventov
|
d8a8a3c319
|
Add ChronicleHashBuilder.createOrRecoverPeristedTo() method
|
2016-01-17 16:03:55 +03:00 |
|
leventov
|
dd4c2f67d8
|
Remove replication frontend implementation in favor of Chronicle Engine
|
2016-01-15 22:19:52 +03:00 |
|
leventov
|
28eeeb70e7
|
Remove the complexity and API-transparency of implementing and supporting built-in serializability for marshallers and core implementation classes, as this is replaced with Wire Marshallable
|
2016-01-14 16:07:22 +03:00 |
|
leventov
|
511fc7729d
|
Fix a bug in SegmentsRecovery
|
2016-01-12 11:53:36 +03:00 |
|
leventov
|
98be95a4a0
|
Fix a bug in ListMarshaller
|
2016-01-12 11:20:44 +03:00 |
|
leventov
|
262ad51a4b
|
Fix bugs in recovery
|
2016-01-11 19:13:50 +03:00 |
|
leventov
|
b168612d90
|
Add ListMarshaller and SetMarshaller factory methods
|
2016-01-11 15:22:42 +03:00 |
|
leventov
|
c75965020e
|
Add CharSequenceBytesReader and prohibit Readers and Writers of CharSequences and Strings to read and write nulls
|
2016-01-11 15:21:38 +03:00 |
|
leventov
|
75252c9fdf
|
Simplify CharSequenceSizedReader
|
2016-01-11 15:19:15 +03:00 |
|
leventov
|
85939d6335
|
Use shortcut ChronicleMapBuilder API in Issue58
|
2016-01-10 20:51:01 +03:00 |
|
leventov
|
ac68523ef8
|
Make SegmentHeader stage to be always init before Locks stage, to avoid duplicate init of the latter
|
2016-01-10 20:01:13 +03:00 |
|
leventov
|
fe7c7787e8
|
Enlarge InterProcessLock.lock() timeout from 2 to 60 seconds, make lockInterruptibly() and tryLock(long, TimeUnit) actually interruptable
|
2016-01-09 19:08:18 +03:00 |
|
leventov
|
021d77e973
|
HCOLL-423 throw IllegalStateException if averageKeySize() or averageKey() is called when size is statically known
|
2016-01-09 19:08:17 +03:00 |
|
leventov
|
65afe387e3
|
Fix grammar and style of the spec
|
2016-01-09 19:08:16 +03:00 |
|
leventov
|
c7cab84be4
|
Add missing reset in SegmentsRecovery.zeroOutFirstSegmentTierCountersArea()
|
2016-01-06 04:07:47 +03:00 |
|
leventov
|
34e942e0ff
|
HCOLL-417 recognize modification iterators appearance cross-process
|
2016-01-06 03:57:30 +03:00 |
|
leventov
|
5191381fc5
|
Move reserved space for global mutable state between existing global mutable state and modification iteration bit sets of first tiers
|
2016-01-06 02:04:40 +03:00 |
|
leventov
|
6ee78ba926
|
HCOLL-378 #36 Fix bugs in recovery
|
2016-01-06 00:39:36 +03:00 |
|
leventov
|
ebee1d36c3
|
Fix a heuristic in VanillaChronicleHash.computeNumberOfTiersInBulk()
|
2016-01-06 00:37:37 +03:00 |
|
leventov
|
97e4c12de1
|
HCOLL-378 #36 Add ChronicleHashBuilder.recoverPersistedTo() and ChronicleHashInstanceBuilder.recover() functionality
|
2016-01-04 23:24:39 +03:00 |
|
leventov
|
b4c97ced36
|
Conditional update of ReplicatedGlobalMutableState.modificationIteratorInitAt
|
2016-01-03 04:01:17 +03:00 |
|
leventov
|
7847675f6f
|
Synchronize ModificationIterator acquisition with globalMutableStateLock, add a bit set identifying already initialized modification bit sets, not to clear them twice in ModificationIterator constructor
|
2016-01-03 01:27:24 +03:00 |
|
leventov
|
9b782c4ec2
|
Remove prevEntry parameter from CompactOffHeapLinearHashTable.clearEntry(), writeEntry() and writeEntryVolatile() methods, as unneeded
|
2016-01-02 16:12:01 +03:00 |
|
leventov
|
5287296fc7
|
Improve changes management in MapRemoteOperations and SetRemoteOperations
|
2016-01-01 21:22:44 +03:00 |
|
leventov
|
e20ea011e3
|
Add ReplicableEntry.dropChangeFor(remoteIdentifier)
|
2016-01-01 21:21:34 +03:00 |
|
leventov
|
d68f151b3a
|
Should acquire updateLock, not readLock, before iterating each segment
|
2015-12-31 18:47:24 +03:00 |
|
leventov
|
04adb466a2
|
Add ReplicableEntry.raiseChangedFor(), raiseChangedForAllExcept(), improve MapRemoteOperations
|
2015-12-30 22:33:42 +03:00 |
|
leventov
|
2c1721e2e3
|
Do propagate accepted replication events, which originated from the node we are currently on, because it might be recovery
|
2015-12-30 21:15:59 +03:00 |
|
leventov
|
b2cd5af700
|
Rework modification iteration and bootstrapping
|
2015-12-30 20:38:55 +03:00 |
|
leventov
|
19fadf5628
|
Remove stateless clients traces in TcpReplicator
|
2015-12-28 19:03:30 +03:00 |
|
leventov
|
1221f11097
|
Make TimeProvider non-extensible and non-configurable; Fix a bug in
ReplicatedQueryAlloc.CleanupAction
|
2015-12-27 21:32:26 +03:00 |
|
leventov
|
123fb177c6
|
Make NanosecondPrecisionSystemTimeProvider monotonic
|
2015-12-27 01:53:57 +03:00 |
|
leventov
|
2928d18d11
|
Fix the prev commit 02b2c90 for ChronicleSet
|
2015-12-26 22:27:35 +03:00 |
|
leventov
|
02b2c90ad0
|
Cover the last corner case in DefaultEventualConsistencyStrategy
|
2015-12-26 22:17:27 +03:00 |
|
leventov
|
c15c568473
|
Remove Replica.EntryCallback.onBeforeEntry() and Replica.EntryCallback.onAfterEntry() methods as unused
|
2015-12-26 18:27:50 +03:00 |
|
leventov
|
1743f1c694
|
Replication updates with local identifier shouldn't use the same timestamp, because this allows two different updates which replace each other on different nodes, and don't converge. Extend default TimeProvider precision to nanoseconds, to exclude possibility of 'microsecond overflow' on frequent changes to the same entry
|
2015-12-26 17:28:35 +03:00 |
|