This topic describes the release notes for the engine version of Time Series Database (TSDB). For more information about the version description and the download of TSDB SDK, see Version description.
V2.6.13 released on April 30, 2021
Fixed issues
The following issue is fixed: Behavior inconsistency occurs due to uncaught exceptions when you use the IgnoreError mode.
The following issue is fixed: No data is returned for a query when the startTime parameter is set to an excessively early point in time.
V2.6.12 released on April 9, 2021
Design optimization
Audit logs that record data writes are added. This feature negatively affects the write performance. Therefore, to enable this feature, you must submit a ticket. When you use audit logs to perform O&M, you can query the audit log records only in the background.
V2.6.11 released on April 2, 2021
New features
roffset and rlimit are added to
api/mquery
andapi/query/mlast
. The syntax of roffset is compatible with that of tupleOffset. The syntax of rlimit is compatible with that of tupleLimit.
Fixed issues
The issue related to the valid range of timestamps is fixed.
The following issue is fixed: An error is reported when the metric that you want to query does not exist.
V2.6.10 released on February 25, 2021
New features
storage usage and timeline usage metrics for TSDB instances are added. This way, you can create alert rules for these metrics in the CloudMonitor console.
Fixed issues
The following issue is fixed: Internal errors occur when global limits are used.
The following issue is fixed: The IgnoreError mode does not take effect.
V2.6.9 released on January 27, 2021
Design optimization
The performance of latest value queries is optimized.
Fixed issues
The following issue is fixed: The number of timelines cannot be reset immediately after a TRUNCATE operation is performed.
The following issue is fixed: The node status is not reset after an unexpected restart when an asynchronous task is performed.
V2.6.7 released on January 14, 2021
New features
The prefix of a field value can be used as a match condition in multi-value data queries.
Nodes can be scaled up in Standard Edition II, Ultimate Edition I, and Ultimate Edition II.
The blank screen time of slow query logs can be reduced. For more information about slow query logs, see Slow query logs.
Fixed issues
The memory leak issue is fixed for TSQL.
The following issue is fixed: An error occurs if the fields that you want to query do not exist when a multi value-query run.
V2.6.6 released on January 7, 2021
The formatted content in slow query logs can be downloaded.
V2.6.5 released on December 30, 2020
New features
The global limit feature can be applied to the total number of data points that are returned for your query request.
Design optimization
Hints can be pushed down to improve query efficiency.
Fixed issues
The following issue is fixed: The query result is empty due to a null pointer exception in your query.
V2.6.4 released on December 3, 2020
New features
Spatial data can be written and queried. This feature is in public preview. To apply for a free trial, submit a ticket.
Fixed issues
The following issue is fixed: An authentication error occurs due to the created
tsdb
user.The following issue is fixed: Nodes my fail to start when multiple nodes are started in parallel.
V2.6.3 released on October 14, 2020
New features
Query hints can be specified in query criteria. For more information about this feature, see query hint description.
The following two features are supported: time series forecasting and time series forecasting for anomaly detection (abnormaldetect). For more information about the two features, see Time series forecasting description and Time series forecasting for anomaly detection (abnormaldetect) description.
Some errors can be ignored when data is written in batches.
Fixed issues
The following issue is fixed: No data is returned when the query condition contains the specified tag keys and tag values.
V2.6.2 released on August 13, 2020
Byte-array data can be written and queried.
V2.6.1 released on July 30, 2020
New features
Multi-value data queries support the following tuple-level filters: WHERE clauses.
Reverse queries are supported.
Fixed issues
The following issue is fixed: An error occurs when the specified tag values do not exist in multi-value data queries.
The following issue is fixed: When data written to existing timelines is out of order due to cache replacement, the written data cannot be queried.
The following issue is fixed: A memory issue occurs in the engine when the heap memory is inappropriately used.
V2.6.0 released on June 29, 2020
New features
TSQL supports INSERT statements.
Fixed issues
The memory issue caused by an excessively large number of connections in the connection pool is fixed.
The following issue is fixed: The millisecond-level data in downsampling expressions cannot be parsed.
The issue related to the latest value queries for a multi-value data model is fixed.
V2.5.14 released on May 19, 2020
New features
The aggregate functions
rmax
,rmin
,rfirst
, andrlast
that return the actual timestamp versions can be added to downsampling expressions.
Fixed issues
The issue of a memory leak for internal cache is fixed.
The following issue is fixed: Unexpected large queries are run because downsampling expressions are incorrectly parsed.
V2.5.13 released on March 19, 2020
New features
The account management feature is supported. You can manage accounts based on basic access authentication. For more information, see Manage accounts.
Fixed issues
The following issue is fixed: An error occurs when you use wildcards to match tags.
V2.5.12 released on January 16, 2020
New features
Prefix matching and suffix matching can be added as query filters.
Fixed issues
The following issue is fixed: JSON fields are incorrectly used because the
api/query/last
operation does not perform strict authentication on JSON fields.
V2.5.11 released on November 21, 2019
The following issue is fixed: The time granularity in the query result is invalid when the specified time granularity for downsampling is weeks in the calendar form.
The following issue is fixed: The result of the COUNT function is invalid when
preDpValue
is used to filter values in a query.The following issue is fixed: The
delta
andrate
operators are mixed in the multi-value data model.
V2.5.10 released on October 16, 2019
The following issue is fixed: Data of monitoring metrics in TSDB is invalid.
V2.5.9 released on September 3, 2019
New features
The delta operators are supported in queries.
Fixed issues
The following issue is fixed: A tag filter in the
tags
field fails to be parsed, but the same tag filter in thefilters
field is correctly parsed.
V2.5.8 released on August 6, 2019
New features
Median operators are supported in downsampling queries.
Fixed issues
The following issue is fixed: The number of data points returned by
/api/suggest
cannot reach the specified maximum number.
V2.5.7 released on July 19, 2019
New features
TSQL supports the multi-value data model.
New instance specifications are released.
V2.5.6 released on June 27, 2019
New features
The time series preprocessing feature is added.
For
api/query
andapi/mquery
operations, data can be queried based on filter conditions.
V2.5.4 released on June 13, 2019
Fixed issues
The backward compatibility issue for the
Content-Type
field in the header of theapi/query
response body is fixed.
V2.5.3 released on June 10, 2019
API operation changes
When the
metric
,tagkey
, ortagvalue
specified in your query condition does not exist, the corresponding error is no longer reported and an empty result set is returned.
Fixed issues
The following issue is fixed: Query failures occur when
Prometheus
pushes down aggregation operators that TSDB does not support.
V2.5.2 released on May 18, 2019
New features
The aggregation algorithm stddev is supported.
When the result set returned by
api/query/mlast
is in the format specified bytupleFormat
, the result set can be displayed by page.
V2.5.1 released on April 26, 2019
Fixed issues
The following issue is fixed: When the details parameter or the summary parameter is added to your /api/put or /api/mput request, the data point statistics are invalid.
The following issue is fixed: When a single data point is written, data in the JSON format fails to be parsed because the data point contains a space.
The issue of null pointer exceptions for the multi-value data model is fixed.
The following issue is fixed: The number of TSDB timelines keeps increasing and cannot be limited.
The following issue is fixed: When a timeline fails to be deleted, some schema information of your multi-value data is not deleted.
V2.5.0 released on April 13, 2019
TSQL is released. TSQL can make development easier. TSQL can be used to perform a wide range of computing operations on time series data. For example, TSQL can be used to join multiple metrics. In addition, TSQL provides more powerful analysis of multiple metrics for services. For more information, see Quick start for TSQL.
TSDB for Prometheus is released. The remote storage adapter for Prometheus is added to the TSDB engine so that the engine can seamlessly connect to open source self-managed Prometheus services. This way, Kubernetes clusters can be monitored in a quick manner. TSDB for Promethues
V2.4.2 released on March 20, 2019
The api/truncate operation is used to delete dirty data. For more information, see Clear data.
Some stability issues are fixed.
V2.4.1 released on February 28, 2019
The following issue is fixed: The integrity of retrieved data is compromised due to index updates.
V2.4.0 released on February 14, 2019
The multi-value data model for time series data is released. Downsampling and aggregation of data for multiple metrics is supported. For more information, see Developer guide.
V2.3.8 released on January 24, 2019
The tag parameter is used as an optional parameter when data is written. queries and writes without specified tags are supported.
Queries without specified tags are optimized.
Timeline queries are optimized to improve the stability of the queries.
The following issue is fixed: The max field in your /api/dump_meta request does not take effect.
V2.3.3 released on November 27, 2018
Experience optimization
The instances whose specifications range from mlarge to 24xlarge can be scaled up or down. These instances can also be scaled out.
Auto-renewal is supported for TSDB instances. You can enable auto-renewal for TSDB instances that use the subscription billing method.
V2.3.1 released on October 20, 2018
The time parameter can be configured in the query/last operation. This way, the latest business data generated before the specified point in time can be queried. Data timestamps are used to specify points in time.
Milisecond granularity of data reads and writes is supported. The msResolution parameter controls the time granularity of data reads. By default, the value of the msResolution parameter is false. This value specifies that the time granularity for data writes is the same as the time granularity for data reads. If the msResolution parameter is set to true, the time granularity for all the returned data is milliseconds.
The TSDB SDK supports filtering of multi-value data points. If multi-value data points are written using the TSDB SDK, the values of multiple metrics can be combined to filter the data and all of the values for these metrics are returned. For more information, see the topics related to TSDB SDK.
The spatio-temporal feature is released. GPS data can be written, filtered, and analyzed. You can submit a ticket to use the feature for testing.
V2.3.0 released on October 9, 2018
New features
The TSDB engine is upgraded. For 48xlarge and 96xlarge instances, data can be written to an unlimited number of timelines. For instances of other specifications, the supported number of timelines is increased to 10 times the original number of timelines.
V2.2.9 released on September 5, 2018
New features
The time series insight feature is released. You can create time series insight events on the time series insight page of the TSDB console. This allows you to query and analyze time series data in a visualized way.
The data for the demo of time series testing can be imported by using one click. You can perform interactive data queries and analysis without the need to develop code.
V2.2.8 released on August 10, 2018
New features
Data generated at a single point in time can be queried.
Data that uses the millisecond granularity can be read, written, or downsampled.
The following time units of downsampling are added: the month unit and the year unit.
The following two operators are added for data filling policies: near and after.
Fixed issues
The following issue is fixed: Timeline monitoring data and disk monitoring data in the TSDB console is invalid.
The following issue is fixed: Timeline metadata can still be retrieved after the timelines are cleared.
The following issue is fixed: The type of read data is inconsistent with that of written data.
The following issue is fixed: Monitoring data about write rates and disks is invalid.
V2.2.7 released on July 6, 2018
New features
A new version of TSDB SDK is released. The new version of TSDB SDK can be used to write, query, and analyze multi-value data points.
V2.2.4 released on May 24, 2018
New features
Data of the STRING and BOOLEAN data types can be written.
The last operation is provided to obtain the data points generated in the latest period for a timeline.
Fixed issues
The following issue is fixed: The result of deleting timelines is invalid in specific conditions.
V2.2.3 released on April 10, 2018
New features
The O&M monitoring and alerting are added.
TSDB can be connected to Realtime Compute for Apache Flink. Realtime Compute for Apache Flink offered by Alibaba Cloud provides a stream computing engine.
The data visualization tool Grafana can be connected to TSDB.
The billing method of an instance can be changed from pay-as-you-go to subscription in the TSDB console.
The data query results can be returned by page.
The following three data filling policies are added: linear, previous and fixed. linear represents linear interpolation. previous means that a value is interpolated based on the preceding value. fixed means that a fixed value is interpolated.
Data points are filtered by metric value.
Inverted indexes are supported.
Abnormal query requests can be processed. These are requests that timed out and requests in which the specified parameters are invalid.
Optimized features
Access to the pipeline aggregation engine is optimized.
The limit on special characters in tag names and metric names is removed.
Fixed issues
The following issue is fixed: Operations for clearing data and metric deletion do not take effect.
The following issue is fixed: The Instances page in the TSDB console is not opened as expected.
The following issue is fixed: Errors are reported for concurrent queries.
V1.2.0 released on February 7, 2018
Optimized features
The stability of TSDB instances is optimized. The following issue is fixed: Services provided by TSDB instances are unstable when large queries are run.
By default, the downsampled data for an instance is calculated by using the local time zone of the region where the instance is deployed.
Fixed issues
The following issue is fixed: When you query tag values in the TSDB console, the tag values returned are incomplete.
V1.1.2 released on November 30, 2017
New features
The getlastdp operation is provided to query the latest data values of time series.
TSDB is available in the China (Shenzhen) region. This region has the following three zones: Zone A, Zone B and Zone C.
Optimized features
The API operation that is used to delete timelines is optimized.
V1.1.1 released on November 16, 2017
New features
TSDB can be accessed over the Internet or virtual private clouds (VPCs).
V1.1.0 released on November 3, 2017
New features
TSDB Client for Java 1.0.0 is released. It provides a wide range of features, such as standard API operations, non-blocking I/O operations, instance protection, automated delivery of data in batches, and automated connection management.
Timelines can be deleted by calling an API operation or by using the TSDB console. Millions of timelines can be deleted within a few seconds.
Instances are protected. The number of data points in a single request can be specified to ensure instance availability.
TSDB is available in the China (Qingdao) and China (Beijing) regions.
Optimized features
Due to O&M automation, alerts can be automatically sent if instances become unavailable.
Fixed issues
The following issue is fixed: An instance becomes unavailable due to a long query time range.
The following issue is fixed: An instance becomes unavailable when the underlying API operations for asynchronous requests are incompatible.
V1.0.2 released on September 15, 2017
New features
TSDB can be connected to Resource Access Management (RAM) so that Alibaba Cloud accounts, RAM users, and the permissions on instance resource management can be assigned based on your business requirements.
V1.0.1 released on August 28, 2017
New features
TSDB can be connected to CloudMonitor. Alerts can be reported for the instance metrics that are monitored.
Optimized features
The process of creating an instance is optimized. The Instances page shows the progress of an instance creation in real time.
The instance configuration is optimized. The maximum number of tags for a single timeline is increased to 16.
Fixed issues
The following issue is fixed: The name of an instance is not specified. However, after the instance is created, the instance name appears.
The following issue is fixed: A curve for instance monitoring does not appear as expected.
The following issue is fixed: An instance can be created when no VPCs are available, but the instance cannot be used.
V1.0.0 released on 2017 July 31
TSDB V1.0.0 is officially launched. This version can be used for public preview. TSDB V1.0.0 provides the following features:
Time series data that complies with the OpenTSDB protocol can be read and written.
Data can be highly compressed before the data is stored. The storage space of a single data point is less than 2 bytes.
Data management system (DMS)
Data validity: Data validity periods can be customized. Data automatically becomes invalid after the customized data validity period ends. The expired data can be automatically deleted.
Data deletion: Useless data or dirty data can be automatically deleted to release the storage space.
Data query: Data can be randomly queried on the web UI.
Instance monitoring alert
The following three core metrics can be monitored in the TSDB console in real time: write transactions per second (TPS), the total number of stored timelines, and storage space.
Security
By default, TSDB instances can be accessed only over VPCs. This way, instance protection can be enhanced.
The instance access whitelist can be configured to protect your instance and ensure data security.