@@ -4,6 +4,71 @@ Releases
4
4
This page documents the releases of Sleeper. Performance figures for each release
5
5
are available [ here] ( docs/13-system-tests.md#performance-benchmarks )
6
6
7
+ ## Version 0.20.0
8
+
9
+ * Note: this release contains breaking changes. It is not possible to upgrade from a previous version of Sleeper
10
+ to version 0.20.0*
11
+
12
+ This contains the following improvements:
13
+
14
+ Tables:
15
+
16
+ - Tables are now internally referenced by a unique ID assigned upon creation. This is in preparation for
17
+ adding the ability to rename tables in the future.
18
+ - Improved support for lots of tables in compaction and ingest status stores by updating the hash key of
19
+ the DynamoDB tables.
20
+ - Table related infrastructure is now shared between all tables. The following resources are now only deployed once:
21
+ - Table data bucket.
22
+ - Table metrics lambda.
23
+ - State store.
24
+ - Table initialisation is no longer performed by CDK.
25
+ - A new client class ` AddTable ` is now responsible for initialising tables.
26
+ - Added configurable timeout property for ` TablePropertiesProvider ` .
27
+
28
+ State store:
29
+
30
+ - The default state store has been updated to the ` S3StateStore ` .
31
+ - The ` minRowKey ` , ` maxRowKey ` , and ` rowKeyTypes ` fields have been removed from the ` FileInfo ` class.
32
+
33
+ Ingest:
34
+
35
+ - Added instance property to allow setting the S3 upload block size.
36
+
37
+ Bulk import:
38
+
39
+ - Added support for overriding spark configuration and platform specification in EMR serverless jobs.
40
+ - Added support for setting the initial capacity of the EMR serverless application.
41
+ - Added support for enabling EMR Studio by using the optional stack ` EmrStudioStack ` .
42
+
43
+ Clients:
44
+
45
+ - The admin client now respects the ` EDITOR ` environment variable when updating properties.
46
+ - Adding an optional stack in the admin client now uploads docker images if the new stack requires one.
47
+
48
+ Query:
49
+
50
+ - Validation failures for queries are now recorded in the ` DynamoDBQueryTracker ` .
51
+ - Added client to view status of query tracker in ` scripts/utility/queryTrackerReport.sh ` .
52
+ - Removed inheritance relationship between ` Query ` and ` LeafPartitionQuery ` .
53
+
54
+ Tests:
55
+
56
+ - Added system tests for using the ` S3StateStore ` .
57
+ - System tests now purge relevant SQS queues if a test fails.
58
+ - Improved performance of ` ingest-runner ` module tests.
59
+ - Added system tests with many tables in one instance.
60
+
61
+ Bugfixes:
62
+
63
+ - Fixed an issue where the python API would not generate unique IDs for each query.
64
+ - Fixed an issue where the instance ID length was not being validated correctly.
65
+ - Fixed an issue where trying to bulk import using EMR serverless to a table using ` S3StateStore ` would
66
+ throw a ` NullPointerException ` .
67
+ - Fixed an issue where sending an ingest job with a null file would not report the job as invalid.
68
+ - Fixed an issue where the role assumed by tasks in the system test data generation cluster exceeded the maximum size.
69
+ - Fixed an issue where the CDK deployment would fail if an ingest source bucket was not set.
70
+ - Fixed a conflict between temporary directory paths used by the CLI.
71
+
7
72
## Version 0.19.0
8
73
9
74
This contains the following improvements:
0 commit comments