A new maintenance release is available for InfluxDB Enterprise.
- InfluxDB Enterprise 1.7.10 — release notes
This maintenance release of InfluxDB Enterprise 1.7.10 includes all of the following fixes contained within InfluxDB 1.7.10 OSS:
- Fix to address corrupt TSM data file renaming failure. When the InfluxDB engine detects a corrupt TSM data file upon startup, the renaming of that file was failing due to a file lock on the data file created by the engine itself.
- Concurrent calls to
tsm1.Engine.Digest()result in a Digest aborted, problem renaming tmp digest error message. Delivered fix makes digest safe for concurrent use.
- Fixed a defect in the TSI index where a filter using the negated equality (!=) operator could result in no matching series being returned for any series stored within the IndexFile portions of the index.
- Fixed compaction logic on infrequent cache snapshots. If the cache snapshot was occurring at a frequency longer than specified by
compact-full-write-cold-duration, a full compaction would be triggered despite newer data which remained in the cache snapshot. This fix ensures that the documented compaction behavior is followed.
- Addressed an issue where multiple blocks for the same series key were having values truncated when they were being read into an empty buffer. The cursor reader code had an optimization that incorrectly assumed the incoming array will be limited to 1,000 values (the maximum block size), but arrays can contain values from multiple matching blocks for the same index entry. Therefore, it’s possible for the array being read to have more than 1,000 values and not fit entirely into the buffer.
- Skip TSM files with block read errors — When InfluxDB finds a bad TSM file during compaction, the error is now logged and the bad file is moved aside. The engine will disregard the file so the next compaction will not hit the same error.
Specific fixes for InfluxDB Enterprise include:
- Added logging for meta nodes when errors related to opening the meta state file occurs. Specific filename and location is also specified as part of the information provided in the log.
- Fixed a defect that occurred for users who were migrating from InfluxDB OSS to InfluxDB Enterprise. The MaxShardGroupID wasn’t being migrated in from OSS which caused the Enterprise edition to create ShardGroupIDs from 0. As a result overlapping, ShardGroupIDs could be imported from the OSS edition.
- Fixed a defect related to removing nodes from the cluster. Shards associated with the node being removed are no longer reassigned to remaining nodes in the cluster. Administrators should rely upon remaining replicas present with the cluster to populate new data nodes that are subsequently added to the cluster.
For our InfluxDB Enterprise customers, log in to the InfluxDB Enterprise portal and download the binaries from there.