Release Notes for MongoDB 3.2

3.2.22 - Dec 28, 2018

• SERVER-38425 Oplog Visibility Query is a collection scan in 3.2.21

• 3.2.22 Changelog

• All JIRA issues closed in 3.2.22

3.2.21 - Sep 7, 2018

• SERVER-32999: Platform Support: remove Debian 7.

• SERVER-36010 Change log messages for Windows stacktraces to use error() or severe() rather than log().

• SERVER-30724 Initial sync might miss ops that were in flight when it started.

• 3.2.21 Changelog

• All JIRA issues closed in 3.2.21

3.2.20 - May 10, 2018

• SERVER-32241: applyOps reports success even when a nested applyOps fails.

• SERVER-32923: Platform Support: remove SLES11 builds.

• SERVER-31535: Platform Support: remove Ubuntu 12.04 builds.

• 3.2.20 Changelog

• All JIRA issues closed in 3.2.20

3.2.19 - Feb 6, 2018

• SERVER-30189: Reduce calls to allocator for large $in expressions

• SERVER-30853: Update Mongod init scripts to set unlimited locked memory

• SERVER-32001: Unindexing a key in a partial unique index may cause server crash

• SERVER-32048: Updates using a numeric path component may cause index entries not to be created

• 3.2.19 Changelog

• All JIRA issues closed in 3.2.19

3.2.18 - Nov 29, 2017

• SERVER-29287: Upgrade pcre to 8.41

• SERVER-31101: WT table not dropped after collection is dropped due to long-running OperationContext

• SERVER-31149: Enable recovery progress messages

• 3.2.18 Changelog

• All JIRA issues closed in 3.2.18

3.2.17 - Sept 28, 2017

Issues fixed:

• SERVER-30636: RangeDeleter assert failed because of replication lag

• SERVER-30943: Segmentation fault on attempt to access an invalidated BSON Object in JS scope

• WT-3327: Checkpoints can hang if time runs backward

• 3.2.17 Changelog

• All JIRA issues closed in 3.2.17

3.2.16 - Jul 27, 2017

Issues fixed:

• SERVER-26952: Cache SCRAM-SHA-1 ClientKey

• SERVER-28578: When doing an upgrade from PV0 to PV1, we should not return from the reconfigure until the lastVote document is written.

• SERVER-29568: Enable configuration of OpenSSL cipher suite via setParameter

• 3.2.16 Changelog

• All JIRA issues closed in 3.2.16

3.2.15 - Jul 5, 2017

Issues fixed:

• SERVER-29237: Add ShardingTaskExecutorPoolMaxConnecting flag for mongos.

• SERVER-29850: Access violation due to a bug in internal page splitting in WiredTiger.

• WT-3362: Cursor opens should never block for the duration of a checkpoint.

• 3.2.15 Changelog

• All JIRA issues closed in 3.2.15

3.2.14 -Jun 13, 2017

Issues fixed:

• SERVER-26452: renameCollection should handle write conflicts.

• SERVER-29365: No-op applyOps does not wait for majority writeConcern before returning.

• SERVER-28014: When ssl mode is preferSSL, should log connections that do not use SSL.

• 3.2.14 Changelog

• All JIRA issues closed in 3.2.14

3.2.13 -- May 1, 2017

Issues fixed:

• SERVER-3181: Add nameOnly option to listDatabases to only get db names, not size info.

• SERVER-21818: Capture system metrics in FTDC

• SERVER-25318: Limit total memory utilization for bulk index builds

• 3.2.13 Changelog

• All JIRA issues closed in 3.2.13

3.2.12 -- Feb 1, 2017

Issues fixed:

• SERVER-27125 Arbiters in pv1 should vote no in elections if they can see a healthy primary of equal or greater priority to the candidate.

• SERVER-25012: createIndex blocks for duration of checkpoint while holding locks.

• SERVER-25865: $group operation is slow on Windows.

• 3.2.12 Changelog

• All JIRA issues closed in 3.2.12

3.2.11 -- Nov 18, 2016

Issues fixed:

• SERVER-26182: In a mixed version replica set with v3.2 and v3.0 members, a sync of a v3.2 member from a v3.0 member can result in a query response that exceed 16 MB BSON size limit and crash the v3.2 member.

• SERVER-24662: Upgrade to Perl compatible regular expressions (PCRE 8.39).

• SERVER-26652 Invalid definitions in systemd configuration for Debian.

• SERVER-24386: killCursor operation during aggregation pipeline can cause segmentation fault.

• 3.2.11 Changelog

• All JIRA issues closed in 3.2.11

3.2.10 -- Sep 30, 2016

Issues fixed:

• SERVER-25974: When WT cache utilization is full, application threads can stall for an extended period of time.

• SERVER-20306: MongoDB with WiredTiger may experience excessive memory fragmentation, leading to swapping and/or out-of-memory errors.

• SERVER-16801: When updating a field of one numerical type to the same number of different numerical type, update operation considers it as a noop.

• 3.2.10 Changelog

• All JIRA issues closed in 3.2.10

3.2.9 -- Aug 16, 2016

Issues fixed:

• SERVER-7285: Support systemd in future compatible Ubuntu distributions.

• WT-2798: Create operations on WiredTiger when journaling is disabled may lead to inconsistent data in the event of a crash.

• SERVER-17856: On mongod instances, users can run currentOp and killOp on own operations.

• 3.2.9 Changelog

• All JIRA issues closed in 3.2.9

3.2.8 -- Jul 12, 2016

Issues fixed:

• SERVER-18329: Debian 8(Jessie) packages.

• SERVER-24580: Improve performance when WiredTiger cache is full.

• WT-2696: Race condition on unclean shutdown may miss log records with large updates.

• WT-2706: Race condition on log file switch can cause missing log records.

• 3.2.8 Changelog

• All JIRA issues closed in 3.2.8

3.2.7 -- Jun 7, 2016

Issues fixed:

• SERVER-24054: JS segmentation fault on load of certain NaNs.

• SERVER-24058: Connection pool asio doesn't honor setup timeouts.

• SERVER-24117: Mongo binaries ELF stack has become executable.

• 3.2.7 Changelog

• All JIRA issues closed in 3.2.7

3.2.6 -- Apr 28, 2016

• First production release of the in-memory storage engine.

Issues fixed:

• SERVER-22970: Background index build may result in extra index key entries that do not correspond to indexed documents.

• SERVER-22043: mongo shell method count() ignores read preference.

• 3.2.6 Changelog

• All JIRA issues closed in 3.2.6

3.2.5 -- Apr 14, 2016

Issues fixed:

• SERVER-22964: IX GlobalLock held while waiting for WiredTiger cache eviction.

• SERVER-22831: Low query rate with heavy cache pressure and an idle collection.

• SERVER-21681: Include index size stats for in-memory storage engine.

• 3.2.5 Changelog

• All JIRA issues closed in 3.2.5

3.2.4 -- Mar 8, 2016

Issues fixed:

• SERVER-22495, SERVER-22728: Update optime when running with journaling disabled.

• SERVER-22269: Have read concern majority reflect journaled state on the primary

• SERVER-22683: replication.enableMajorityReadConcern ignores actual setting value and instead implies true.

• SERVER-22043: count() method in the mongo shell does not apply the read preferences.

• 3.2.4 Changelog

• All JIRA issues closed in 3.2.4

3.2.3 -- Feb 17, 2016

Issues fixed:

• SERVER-22261: For MMAPv1 journaling, the "last sequence number" file (lsn file) may be ahead of what is synced to the data files.

• SERVER-22167: In some cases, insert operations fails to add the _id field to large documents.

• SERVER-22456: Increased timeout for querying oplog to 1 minute.

• 3.2.3 Changelog

• All JIRA issues closed in 3.2.3

3.2.2 -- Feb 16, 2016

Replaced by MongoDB 3.2.3. Users wishing to run MongoDB 3.2 should skip 3.2.2 and upgrade directly to 3.2.3.

3.2.1 -- Jan 12, 2016

Issues fixed:

• SERVER-21868: During a regular shutdown of a replica set, secondaries may mark certain replicated but yet to be applied operations as successfully applied.

• SERVER-20262: Improve insert workload performance with WiredTiger on Windows.

• SERVER-21366: Long-running transactions during chunk moves.

• 3.2.1 Changelog

• All JIRA issues closed in 3.2.1

WiredTiger Default Cache Size

Starting in MongoDB 3.2, the WiredTiger internal cache, by default, will use the larger of either:

• 60% of RAM minus 1 GB, or

• 1 GB.

For more information, see WiredTiger and Memory Use.

WiredTiger Journal Write Frequency

MongoDB 3.2 configures WiredTiger to write to the journal files at every 50 milliseconds. This is in addition to the existing journal write intervals and conditions. For more information, see Journaling Process.

New Aggregation Stages

{
   $lookup:
     {
       from: <collection to join>,
       localField: <fieldA>,
       foreignField: <fieldB>,
       as: <output array field>
     }
}

New Accumulators for $group Stage

New Aggregation Arithmetic Operators

New Aggregation Array Operators

{
  $concatArrays: [ <array1>, <array2>, ... ]
}

{
  $filter:
    {
      input: <array>,
      as: <string>,
      cond: <expression>
    }
}

Accumulator Expression Availability

Starting in version 3.2, the following accumulator expressions, previously only available in the $group stage, are now also available in the $project stage:

• $avg

• $min

• $max

• $sum

• $stdDevPop

• $stdDevSamp

When used as part of the $project stage, these accumulator expressions can accept either:

• A single argument: <accumulator> : <arg>

• Multiple arguments: <accumulator> : [ <arg1>, <arg2>, ... ]

General Enhancements

• In MongoDB 3.2, $project stage supports using the square brackets [] to directly create new array fields. For an example, see Project New Array Fields.

• MongoDB 3.2 introduces the minDistance option for the $geoNear stage.

• $unwind stage no longer errors on non-array operand. If the operand does not resolve to an array but is not missing, null, or an empty array, $unwind treats the operand as a single element array.

  $unwind stage can:

  • include the array index of the array element in the output by specifying a new option includeArrayIndex in the stage specification.

  • output those documents where the array field is missing, null or an empty array by specifying a new option preserveNullAndEmptyArrays in the stage specification.

  To support these new features, $unwind can now take an alternate syntax. See $unwind for details.

Optimization

Indexes can cover aggregation operations.

MongoDB improves the overall performance of the pipeline in large sharded clusters.

If the pipeline starts with an exact $match on a shard key, the entire pipeline runs on the matching shard only. Previously, the pipeline would have been split, and the work of merging it would have to be done on the primary shard.

For aggregation operations that run on multiple shards, if the operations do not require running on the database's primary shard, these operations can route the results to any shard to merge the results and avoid overloading the primary shard for that database. Aggregation operations that require running on the database's primary shard are the $out stage and $lookup stage.

Compatibility

For compatibility changes, see Aggregation Compatibility Changes.

text Index Version 3

MongoDB 3.2 introduces a version 3 of the text index. Key features of the new version of the index are:

• Improved case insensitivity.

• Diacritic insensitivity.

• Additional delimiters for tokenization.

Starting in MongoDB 3.2, version 3 is the default version for new text indexes.

$text Operator Enhancements

The $text operator adds support for:

• case sensitive text search with the new $caseSensitive option, and

• diacritic sensitive text search with the new $diacriticSensitive option.

For more information and examples, see the $text operator reference sections Case Insensitivity and Diacritic Insensitivity.

Support for Additional Languages

Starting in 3.2, MongoDB Enterprise provides support for the following languages: Arabic, Farsi (specifically Dari and Iranian Persian dialects), Urdu, Simplified Chinese, and Traditional Chinese.

For details, see Text Search with Basis Technology Rosette Linguistics Platform.

inMemory Storage Engine

MongoDB Enterprise 3.2 provides an in-memory storage engine. Other than some metadata, the in-memory storage engine does not maintain any on-disk data. By avoiding disk I/O, the in-memory storage engine allows for more predictable latency of database operations.

To select this storage engine, specify

• inMemory for the --storageEngine option or the storage.engine setting.

• --dbpath. Although the in-memory storage engine does not write data to the filesystem, it maintains in the --dbpath small metadata files and diagnostic data as well temporary files for building large indexes.

The inMemory storage engine uses document-level locking. For more details, see In-Memory Storage Engine.

ephemeralForTest Storage Engine

MongoDB 3.2 provides a new for-test storage engine. Other than some metadata, the for-test storage engine does not maintain any on-disk data, removing the need to clean up between test runs. The for-test storage engine is unsupported.

To select this storage engine, specify

• ephemeralForTest for the --storageEngine option or the storage.engine setting.

• --dbpath. Although the for-test storage engine does not write data to the filesystem, it maintains small metadata files in the --dbpath.

The ephemeralForTest storage engine uses collection-level locking.

Bit Test Query Operators

MongoDB 3.2 provides new query operators to test bit values:

• $bitsAllSet

• $bitsAllClear

• $bitsAnySet

• $bitsAnyClear

SpiderMonkey JavaScript Engine

MongoDB 3.2 uses SpiderMonkey as the JavaScript engine for the mongo shell and mongod server. SpiderMonkey provides support for additional platforms and has an improved memory management model.

This change affects all JavaScript behavior including the commands mapReduce, group, and the query operator $where; however, this change should be completely transparent to the user.

mongo Shell and CRUD API

To provide consistency with the MongoDB drivers' CRUD (Create/Read/Update/Delete) API, the mongo shell introduces additional CRUD methods that are consistent with the drivers' CRUD API:

WiredTiger and fsyncLock

Starting in MongoDB 3.2, the WiredTiger storage engine supports the fsync command with the lock option or the mongo shell method db.fsyncLock(). That is, for the WiredTiger storage engine, these operations can guarantee that the data files do not change, ensuring consistency for the purposes of creating backups.

Platform Support

Starting in 3.2, 32-bit binaries are deprecated and will be unavailable in future releases.

MongoDB 3.2 also deprecates support for Red Hat Enterprise Linux 5.

$type Operator Accepts String Aliases

$type operator accepts string aliases for the BSON types in addition to the numbers corresponding to the BSON types.

explain() Support for distinct() Operation

db.collection.explain() adds support for db.collection.distinct() method. For more information, see db.collection.explain().

explain Support for findAndModify and distinct Commands

explain adds support for the distinct and findAndModify commands. For more information, see explain.

Deprecation of the HTTP Interface

Starting in 3.2, MongoDB deprecates its HTTP interface.

keysExamined Count Includes the Last Scanned Key

For explain operations run in executionStats or allPlansExecution mode, the explain output contains the keysExamined statistic, representing the number of index keys examined during index scans.

Prior to 3.2, keysExamined count in some queries did not include the last scanned key. As of 3.2 this error has been corrected. For more information, see keysExamined.

The diagnostic logs and the system profiler report on this statistic.

Geospatial Optimization

MongoDB 3.2 introduces version 3 of 2dsphere indexes, which index GeoJSON geometries at a finer gradation. The new version improves performance of 2dsphere index queries over smaller regions. In addition, for both 2d indexes and 2dsphere indexes, the performance of geoNear queries has been improved for dense datasets.

Diagnostic Data Capture

To facilitate analysis of the MongoDB server behavior by MongoDB engineers, MongoDB 3.2 introduces a diagnostic data collection mechanism for logging server statistics to diagnostic files at periodic intervals. By default, the mechanism captures data at 1 second intervals. To modify the interval, see diagnosticDataCollectionPeriodMillis.

MongoDB creates a diagnostic.data directory under the mongod instance's --dbpath or storage.dbPath. The diagnostic data is stored in files under this directory.

The maximum size of the diagnostic files is configurable with the diagnosticDataCollectionFileSizeMB, and the maximum size of the diagnostic.data directory is configurable with diagnosticDataCollectionDirectorySizeMB.

The default values for the capture interval and the maximum sizes are chosen to provide useful data to MongoDB engineers with minimal impact on performance and storage size. Typically, these values will only need modifications as requested by MongoDB engineers for specific diagnostic purposes.

Write Concern

• For replica sets using protocolVersion: 1, secondaries acknowledge write operations after the secondary members have written to their respective on-disk journals, regardless of the j option.

• For replica sets using protocolVersion: 1, w: "majority" implies j: true.

• With j: true, MongoDB returns only after the requested number of members, including the primary, have written to the journal. Previously j: true write concern in a replica set only requires the primary to write to the journal, regardless of the w: <value> write concern.

journalCommitInterval for WiredTiger

MongoDB 3.2 adds support for specifying the journal commit interval for the WiredTiger storage engine. See --journalCommitInterval option. In previous versions, the option is applicable to MMAPv1 storage engine only.

For the corresponding configuration file setting, MongoDB 3.2 adds the storage.journal.commitIntervalMs setting and deprecates storage.mmapv1.journal.commitIntervalMs. The deprecated storage.mmapv1.journal.commitIntervalMs setting acts as an alias to the new storage.journal.commitIntervalMs setting.