Warp10 Platform Versions Save

The Most Advanced Time Series Platform

3.2.0

2 months ago

This release is mainly a bugfix release. We highly recommend you migrate to this version if using version 17 of the Java Virtual Machine or if you are running a distributed version of Warp 10 with Directory sharding.

The release includes the following changes:

  • Fixed an issue related to closing Directory iterators when accessing multiple shards which could lead to NullPointerException being thrown.

  • Fixed an issue in UNBUCKETIZE.CALENDAR when applied to GTS with empty buckets.

  • Fixed issue which prevented the Warp10OutputFormat to function properly within Spark due to a mismatch in versions of Guava.

  • Changed order of config files, they are now loaded in the order of the numberic prefix appearing before the first -.

  • Added missing support for quietAfter and activeAfter parameters when doing a dryrun delete in a distributed Warp 10 instance.

  • Added support for internal statement wrapping.

  • Changed behavior when the use of version 17 of the Java Virtual Machine is detected. Warp 10 will now stop and emit a warning message indicating that the configuration key labelsid.slowimpl should be set to true. This is needed to ensure an issue with the Just In Time compiler for that version of the JVM is not triggered within the code used to compute series id.

You can read a more detailed version of this release note on the SenX blog.

3.1.2

5 months ago

Emergency bugfix release addressing an issue with BIGDECIMAL handling.

Use this release of 3.1.x instead of the previous 3.1.1 and 3.1.0.

3.1.1

6 months ago

NOT FOR PRODUCTION USE

This is mainly a bugfix release. Release 3.1.0 had an issue which prevented it to run correctly in a distributed setup, as a consequence 3.1.0 MUST NOT be used for distributed instances. Release 3.1.1 fixes that issue.

Apart from that bugfix release 3.1.1 improves the build mechanism.

3.1.0

6 months ago

NOT FOR PRODUCTION USE

  • Added support for Argon2 key derivation
  • Added configuration keys to control directory queries made to multiple shards
  • Updated the InfluxDB plugin with faster performance when ingesting ILP data, added function ILP->
  • Fixed a Thrift bug which prevented deserializing blobs over 100MB
  • Improved performance of sort operations on GTS which contain duplicate data points, improvement can reach 5000x in some cases
  • Added configuration (labelsid.slowimpl) to select slow implementation of labelsId to circumvent a JDK17 JIT bug which is still occurring on some setups
  • Fixed Gzip OS in header
  • Added archive backend
  • Optimized binary operations on GTS
  • Added filler.value
  • Added support for BigDecimal
  • Fixed issue when setting WARP10_IDENT

3.0.3

8 months ago
  • Bumped Hadoop dependencies to 3.3.6
  • Fixed an issue with the clearing of metrics via SENSISION.SET
  • Modified JSON-> so it produces LinkedHashMap
  • Fixed an issue in warp10.sh which affected tokengen and run if using custom JVM options

3.0.2

9 months ago

This patch release addresses the following issues:

  • Tenant prefix was not correctly added to delete messages
  • Macro Authentication Plugins could fail with NPE
  • Macro Authentication Plugins are now loaded prior to Ingress/Egress being available preventing their bypass at startup

The following changes were added:

  • FSTORE no longer creates CRC files
  • Tenant prefixes can now be specified explicitely for Directory/Egress/Store instead of tenant names therefore simplifying restore operations
  • Warp 10 metrics can now be retrieved in OpenMetrics format

3.0.1

9 months ago

This is mainly a bugfix release addressing several issues discovered in edge cases post 3.0.0 release.

  • Fixed issue #1276 related to BUCKETIZE operations in the presence of duplicate ticks
  • Fixed several FoundationDB issues triggered by the use of multiple FDB clients
  • Fixed an issue present since 2016 which could lead to the impossibility to fetch data when fetching from a large number of GTS with a consecutive number of them without data in the selected time range
  • Fixed incorrect handling of gskip/gcount in some standalone /api/v0/fetch operations
  • Added checks to prevent writing to FoundationDB when no tenant is defined in Warp 10 but FDB requires one

3.0.0

10 months ago

Release 3.0.0

3.0.0-beta4

11 months ago

This beta corrects a side effect of the switch to FoundationDB. As FDB limits the size of records it can store a limit was introduced in Warp 10 to ensure metadata (class + labels + attributes) do not exceed a threshold which ensures they can be stored in FDB. This limit was enforced unconditionally even when not using FDB. This beta release introduces new configuration keys to relax the check and only emit a warning.

The version of the FoundationDB Java bindings was also set to 7.1.33 which is the latest stable version.

The README file was refactored.

The utility warp10.sh now includes an option to initialize the LevelDB directory.

3.0.0-beta3

11 months ago

This beta3 version fixes an issue that triggers a nasty JVM bug present at least in JDK 17 to 20. The bug could lead to incorrect series ID which could corrupt the metadata store of Warp 10.

The JDK bug is related to the JIT compiler and the inlining process.

Any user of Warp 10 3.x using a JDK >= 17 is urged to switch to this beta.