bigtable-client-1.1.1

NOTE: Do not use this release. Use 1.1.2 instead.

• Attempted to fix RefreshingOAuth2CredentialsInterceptor due to a customer issue. Users in datacenters outside of GCP may have experienced problems related to the same issue.
• App Profiles are now supported in the import / export process
• Parent project refactoring to clarify management of HBase 2.0 artifacts.

bigtable-client-1.1.0

• Upgrading grpc to 1.9.0
• Removing the deprecated bigtable-hbase-dataflow artifact. Use bigtable-hbase-beam instead.
• Adding initial (and partial) support for HBase 2.0's new APIs, including the async APIs.
• Changed the default settings of the client to improve throughput of BulkMutator and BufferedMutator
• Added a new Filters object to the core library that simplifies creation of Cloud Bigtable RowFilter objects
• Added a new HBase Filter that wraps a Cloud Bigtable RowFilter
• Fixed the implementation SingleColumnValueFilter to be more efficient.
• Removing the bigtable-protos jar in favor of com.google.api.grpc:proto-google-cloud-bigtable-v2 and com.google.api.grpc:proto-google-cloud-bigtable-admin-v2

bigtable-client-1.0.0

• This release has minor bug fixes on top of bigtable-client-1.0.0-pre4, and is intended to symbolize the combined efforts of bigtable-client-1.0.0-pre1 - bigtable-client-1.0.0-pre4.
• The one notable feature is integration with Stackdriver tracing. Here is an example that leverages the new tracing features.

bigtable-client-1.0.0-pre4

The next release is likely to be 1.0.0 which will be based on the 1.0.0-pre4 version. No new features will be in 1.0.0, but there may be some minor bug fixes.

• Deprecated our Dataflow v1 artifacts
• Removed the bigtable-hbase-1.0, bigtable-hbase-1.1, bigtable-hbase-1.2 and bigtable-hbase-1.3 artifacts. bigtable-hbase-1.x should be used in most cases, and bigtable-hbase-1.x-hadoop should be used in hadoop environments
• Upgraded to grpc 1.7.0
  Created a Beam based import/export tools. The Dataflow version is deprecated.
• Minor Bug fixes: fixed a NPEs when canceling reads: fixed an issue when writing rows with large cell counts (100,000) in Buffered Mutator.

bigtable-client-1.0.0-pre3

• Added an Apache Beam-compatible version of the Cloud Dataflow connector. See the Cloud Dataflow release notes for information about migrating from the Dataflow 1.x SDK to the Dataflow 2.x (Beam-compatible) SDK.

Note: The previous version of the Cloud Dataflow connector is deprecated and will be removed in the next release.
Fixed a thread leak for scans that are only partially read.

• Enhanced the Cloud Dataflow connector's throttling feature to account for jobs that write to multiple instances.
• Added an option to treat HBase namespace operations as no-ops that log a warning, rather than prohibited operations that throw an exception. (Cloud Bigtable does not support namespaces.) To enable this feature, set the option google.bigtable.namespace.warnings to true, either in your hbase-site.xml file or programmatically.

bigtable-client-1.0.0-pre2

• The netty-tcnative-boringssl-static dependency is now included as a transitive dependency in the Cloud Bigtable Maven artifacts. You should remove this dependency from your pom.xml file.
• The method Table#checkAndPut now performs comparisons in the correct order.
• Using a ValueFilter with an empty string comparison now works correctly.
• Using an empty FilterList no longer triggers a full table scan.
• Optimized the method Table#exists, which checks whether a row exists.
• When possible, the client now retries failed requests for table administration operations.
• Added an experimental Maven plugin that runs the Cloud Bigtable emulator.
• Upgraded to gRPC 1.5.0.

bigtable-client-1.0.0-pre1

This version includes the following changes:

• The HBase client for Java now provides three Maven artifacts:
  • bigtable-hbase-1.x: Use this artifact for standalone applications where you control your dependencies.
  • bigtable-hbase-1.x-hadoop: Use this artifact for Hadoop environments.
  • bigtable-hbase-1.x-shaded: Use this artifact for environments other than Hadoop that require older versions of the HBase client for Java's dependencies, such as protobuf and Guava.
• You must update your application to use one of these new artifacts. In addition, if your configuration settings include a value for hbase.client.connection.impl, you must change the value to com.google.cloud.bigtable.hbase1_x.BigtableConnection.
• You no longer need to include an hbase-client artifact in your Maven project.
  Improved the client-side throttling mechanism to reduce the likelihood of Cloud Dataflow jobs overloading a Cloud Bigtable cluster. To enable this feature, set the option google.bigtable.buffered.mutator.throttling.enable to "true", either in your hbase-site.xml file or programmatically. Enabling this option is now recommended for all Cloud Dataflow jobs that write to Cloud Bigtable.

bigtable-client-0.9.7.1

• Fixed a bug in the Cloud Dataflow connector that caused some records to be read multiple times, or not at all, at the end of a long read job. This issue was introduced in version 0.9.4.
• Fixed an issue with long-running OAuth requests that caused gRPC to throw "Not started" exceptions. This issue was introduced in version 0.9.6.2.
• Added support for HBase's FamilyFilter, which you can use to filter based on the column family.

bigtable-client-0.9.7

• Added an experimental client-side throttling mechanism to reduce the likelihood of Cloud Dataflow jobs overloading a Cloud Bigtable cluster. To enable this feature, set the option google.bigtable.buffered.mutator.throttling.enable to true, either in your hbase-site.xml file or programmatically.
• Added the ability to configure authentication by passing in a JSON service account key as a string. To use this feature, set the option google.bigtable.auth.json.value to the text of your JSON service account key, either in your hbase-site.xml file or programmatically.
• Fixed an issue that prevented Table.checkAnd*() methods from working correctly with comparators other than EQUAL and NOT_EQUAL.
• Improved error handling of BufferedMutator operations under heavy load.
• Upgraded to gRPC 1.3.0.

bigtable-client-0.9.6.2

• You must now use version 1.1.33.Fork26 of the netty-tcnative-boringssl-static library.
• Added HBase 1.3.x compatibility. HBase 1.3.x is now the default version for our core integration. Most applications will continue to work with older versions of HBase, with the exception of the HBase shell.
• When an OAuth token is revoked, causing the client to receive an UNAUTHORIZED response, the client now retrieves a new OAuth token and retries the request.
• Improved the performance of Table.get(List<Get> gets) by requesting multiple concurrent batches instead of a single large batch of requests.
• Added a cache for the underlying Netty/gRPC channel pool as a performance enhancement for systems that open many connections, such as Cloud Dataflow or Hadoop connectors. To enable this feature, set the option google.bigtable.use.cached.data.channel.pool to true, either in your hbase-site.xml file or programmatically.
• Added support for HBase's MultiRowRangeFilter, and added a subclass of Scan, BigtableExtendedScan, to enable a scan with an arbitrary set of row keys and range.
• PrefixFilter used to perform a full table scan. Now, under most conditions, it only scans for the rows in the PrefixFilter. One exception is if you are using a complex FilterList that contains a PrefixFilter with the operator FilterList.Operator.MUST_PASS_ALL; in this case, the filter performs a full table scan. Use PrefixFilter with BigtableExtendedScan to optimize performance in this case.
• Fixed an issue that could cause a ValueFilter with CompareFilter.CompareOp.EQUAL to fail against the Cloud Bigtable emulator.
• The client now uses Maven's google-common-protos artifact for protobuf objects rather than having a copy in bigtable-protos.
• Upgraded to gRPC 1.2.0.