RedisTimeSeries 1.4.7 - December 2020

This is a maintenance release for version 1.4.

Headlines:

This release includes support for Redis on Flash (RoF) deployments, improves overall stability, and provides fixes for issues found after the previous release.

Bugfixes:

  • #581 Misaligned allocators usage might crash Redis.
  • #588 ON_DUPLICATE min/max rules not working for negative value.

RedisTimeSeries 1.4.6 - November 2020

This is a maintenance release for version 1.4.

Headlines:

This release improves overall stability and provides fixes for issues found after the previous release.

Minor enhancements:

  • #565 duplicate policy: add SUM option: If a previous sample exists, add the new sample to it so that the updated value is equal to (previous + new). If no previous sample exists, set the updated value equal to the new value. (PR #565
  • #559 Compressed chunk will be be the default global option
  • #559 Added chunkType to TS.INFO

Bugfixes:

  • #528 Out of order insert might crash Redis if there’s an update to an empty downsampled key
  • #561 TS.MRANGE command might crash if there’s an expired key that was deleted in the result set

RedisTimeSeries 1.4 GA (1.4.5 - September 2020)

This is the General Availability release for RedisTimeSeries 1.4.

Highlights:

Ability to backfill time series! You can now add samples to a time series where the time of the sample is older than the newest sample in the series. This enables:

  • Adding out of order of samples to time series.
  • Batch loading of historical samples into an existing series.
  • Updating existing samples (for example for compliance reasons).

This has been the most requested feature for RedisTimeSeries. We look forward to your feedback so we can move to a general availability release soon.

Details:

  • Added functionality:
    • #254 TS.REVRANGE and TS.MREVRANGE [commands] allow for querying in descending order of Timestamps. (https://oss.redislabs.com/redistimeseries/1.4/commands/#tsmrangetsmrevrange)
    • #503 - RDB saves the whole chunk instead of individual samples giving a speed and space improvement when saving or loading an RDB file
    • #502 - The ability to set, at creation time, the data section size of each chunk using flag CHUNK_SIZE. TS.INFO uses chunkSize instead of maxSamplesPerChunk.
    • #437 Allow backfilling of samples and updating of existing samples
      • Works with compressed and uncompressed series.
      • This comes with a performance hit when a sample is written out-of-order. We will publish numbers once we are generally available, but are still considering optimisations.
    • #521 DUPLICATE_POLICY allows to configure on module, series and sample level how to handle duplicate samples. A duplicate sample is a sample for which the series holds already a sample on the same timestamp. Note that the default behaviour is equal to v1.2: BLOCK

Notes: The version inside Redis will be 10405 or 1.4.5 in semantic versioning. Since the version of a module in Redis is numeric, we could not add an GA flag.