Release Announcement: InfluxDB OSS 1.8.2 and InfluxDB Enterprise 1.8.2

By Tim Hall / Product
Aug 24, 2020

Navigate to:

A new maintenance release for InfluxDB OSS is available now.  Additionally, a new maintenance release for InfluxDB Enterprise is also available. 

This is the first maintenance release of InfluxDB Enterprise on the 1.8 line and builds upon both the InfluxDB OSS 1.8.1 and 1.8.2 releases. As a part of our normal quality checks, InfluxDB Enterprise 1.8.1 did not pass, and as a result, we did not release it. 

InfluxDB 1.8.2 reverts a single change which resulted in some community users, and during our testing of InfluxDB Enterprise, reporting a significantly higher memory utilization.  

For those InfluxDB Enterprise customers who may have missed the InfluxDB OSS 1.8.1 release announcement, highlights include:

  • Optional Feature:
    • Allow configuration for users to add custom HTTP response headers to InfluxDB. This allows users to supply headers to comply with internal security policies. This is achieved by first adding the http-headers configuration option. http-headers accepts a list of key value pairs containing headers and their values.
  • Performance Improvements:
    • InfluxQL query planner now plans each field in parallel to improve planning time. Previously a query with multiple fields (e.g. SELECT SUM(x), COUNT(x)...) would plan each field one after another.
    • Improved performance of DELETE/DROP by batching tombstone writes.
  • Fixes and Other Improvements: 
    • Flux: Addressed regression issue where calling buckets() would cause a panic.
    • TSI: Addressed a couple of edge cases which resulted in segfaults.
    • TSI: Addressed deadlock when DROP and DELETE were issued through separate sessions.
    • Storage: Improved error handling when creating snapshots.

For the InfluxDB Enterprise 1.8.2 release, a number of Enterprise-specific issues and improvements have been delivered. These include:

  • Hinted Handoff Improvements:
    • A new configuration option has been added to the [hinted-handoff] section called max-pending-writes, which defaults to 1024. See [max-pending-writes] for more detail.
    • A new configuration parameter has been added to the [cluster] section called allow-out-of-order-writes, which defaults to false to match the existing behavior.  There are some important operational considerations to review before turning this on. But the result is that this can reduce the time required to drain the hinted handoff queue and increase throughput during recovery.
    • Additionally, in certain scenarios, the hinted-handoff queue would fail to drain. Upon node startup, the queue segment files are now verified and truncated if any are corrupted. 
    • Some additional logging has been added when a node starts writing to the hinted handoff queue as well.
    • An update has been made to the hinted handoff queue to ensure that various entries to segment files occur atomically. Prior to this change, entries were written to disk in three separate writes (len, data, offset). If the process stopped in the middle of any of those writes, the hinted handoff segment file was left in an invalid state.
  • influxd-ctl CLI Improvements:
    • There is a new verbose option for show-shards which provides additional details on the shards themselves. 
  • Fixes and Other Improvements: 
    • Eliminate orphaned, temporary directories and — in the event such a directory is created — provide useful log information regarding the reason.
    • LDAP Configuration: GroupSearchBaseDNs, SearchFilter, GroupMembershipSearchFilter, and GroupSearchFilter values in the LDAP section of the configuration file are now all escaped.
    • Resolve a cluster read service issue that caused a panic. Previously, if no tags keys or values were read, the cluster read service returned a nil cursor. Now, an empty cursor is returned. 

Again, the binaries for the latest open source release can be found on our downloads page. 

For our InfluxDB Enterprise customers, log in to the InfluxDB Enterprise portal and download the binaries from there.  If you need need more help?  Contact support.

For our community, if you find issues or have questions, please join our InfluxDB Community Slack, post them in our InfluxDB Github Repo or our Community Site and we will take a look.