-
Notifications
You must be signed in to change notification settings - Fork 14
/
Copy pathCHANGES.txt
107 lines (77 loc) · 5.18 KB
/
CHANGES.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
Splout SQL Change Log
=====================
(DEVELOPMENT 0.3.1-SNAPSHOT)
IMPROVEMENTS
- 2014-12-17 - Deployment cancellation now possible.
- 2014-12-12 - Concurrent deployments support.
- 2014-12-12 - No retrying syntax error queries or timeout queries anymore.
- 2014-12-09 - Existing partition map at Splout Server can be reused. Useful for not being recalculating partition maps always.
- 2014-12-09 - API accepts user defined partition maps
- 2014-11-28 - Timeout when pool of DNodes is full. Avoiding enqueue for always queries when Splout is full.
- 2014-11-12 - qNode connections to DNode retrials are delayed in successive failures.
- 2014-11-12 - All properties in Splout config starting by hazelcast. are passed through Hazelcast as is.
- 2014-09-29 - Accepting responses compression with Gzip and deflate
BUG FIXES
- 2014-12-12 - Bug in registry when starting qNode with already existing DNodes.
- 2014-12-05 - TimeoutThread not working properly due to only covering statement prepare.
- 2014-12-01 - hz.tcp.cluster property only using the first given host
- 2014-11-28 - Solved a race condition in the TimeoutThread
- 2014-11-28 - Thrift pools are not in infinite loop to renewal anymore.
- 2014-11-12 - Upgrading to Hazelcast 3.3.2 that includes many bug fixes.
Splout SQL 0.3.0
IMPROVEMENTS
- 2014-09-01 - Cloudera CDH5 compilation profile added. Integration test for CDH5 added.
- 2014-09-01 - Now default profile is mr2. Fixing mr2 integration test
- 2014-07-31 - Solved a bug that arises when no state was stored in disk
- 2014-07-23 - Increase GET buffer to 65535
- 2014-06-02 - Parallel deployment of partitions in DNode, configure through the property dnode.deploy.parallelism. Default 3.
- 2014-05-20 - Parallel generation of tablespaces with generate command
- 2014-05-20 - Tablespace tables view in the Splout panel
- 2014-04-20 - Sampling size can be configured by setting the property splout.sampling.records.to.sample
in Hadoop configuration.
- 2014-02-14 - Improved FULL_SCAN sampling. Now it uses one job per tablespaces instead of
one per table.
- 2014-02-14 - Renamed RESERVOIR and DEFAULT sampling methods to FULL_SCAN and RANDOM.
The default one is now RESERVOIR as it is the one that gives the best results.
- 2014-02-14 - Sampling is not using memory anymore. That allows to use more samples in
sampling, so that we can have more accurate partition map.
BUG FIXES
- 2014-06-30 - Solved a Bug with generated partitions maps. The code could generate partition maps
with the latest partition empty. Workarround implemented to ensure that all partitions
has some data.
- 2014-04-20 - Solved a Bug that prevents the deployment of tablespaces with 128 or more partitions
- 2013-09-05 - splout-service.sh adapted to run properly with packages from Cloudera
Splout SQL 0.2.5
BUG FIXES
- 2013-08-01 - SQLiteManager wasn't properly closing all connections opened to SQLite files by different threads, which resulted in a waste of disk space after long-running deployments.
Splout SQL 0.2.4
BUG FIXES
- 2013-06-04 - Better handling of temporary folders at TupleSQLite4JavaOutputFormat. Eliminated the collision possibilities.
- 2013-05-24 - Solved small issues with Hive+Splout integration 1) Explicit schema is now ignored and warned instead of crashing 2) Insertion order by works instead of crashing.
Splout SQL 0.2.3
NEW FEATURES
- 2013-04-05 - MapReduce 2 integration. Now YARN is supported. Two versions
are compiled: one compatible with mapreduce 1 (mr1) and the
other compatible with mapreduce 2 (mr2). Use the maven
profile mr2 (-Pmr2) to create binaries for MapReduce 2.
Splout SQL 0.2.2
NEW FEATURES
- 2013-03-15 - Integration with Hive, Cascading and Pig.
IMPROVEMENTS
- 2013-02-06 - #12: Automatic replica balancing to avoid under-replicated partitions (high availability). Disabled by default, experimental new feature.
- 2013-01-17 - Replaced NullableSchema and NullableTuple by the native null support integrated in Pangool
BUG FIXES
- 2013-01-29 - Solved issue #9 now it is possible to use space as separator in SimpleGenerateCMD.
- 2013-01-29 - Solved bug "com.splout.db.hadoop.TableBuilder$TableBuilderException: Invalid compound index: ..., field null not present in specified Schema:"
Splout SQL 0.2.1
IMPROVEMENTS
- 2013-01-09 - Changed Thrift's DNode server to be a THsHaServer and added a connection pool to QNodes for solving issues #5 and #6 as well.
- 2012-12-13 - New SQLite libs recompiled with older debian. Previous ones didn't work under some environments.
- 2012-12-18 - Removed dependency on Jersey which was the responsible for a memory leak.
- 2012-12-20 - Added the possibility of specifying the network interfaces in the Hazelcast facade.
BUG FIXES
- 2012-12-17 - Solved small issue #1 TablespaceGenerator doesn't exit if Reservoir Sampling fails.
- 2012-12-17 - Solved small issue #2 Automatic IP discovery makes the service crash when there is no Internet connection at all.
- 2012-12-20 - Solved issue #4 Hazelcast persistence may be saving only one tablespace version at a time.
Splout SQL 0.1
- Initial Release!