- Release Notes >
- Release Notes for MongoDB 3.6
Release Notes for MongoDB 3.6¶
On this page
- Minor Releases
- Security
- Aggregation
- Array Update Operator Enhancements
- 3.6 Compatible Drivers
- Change Streams
- Client Sessions
- Server Sessions
- JSON Schema
- Replica Sets
- Sharded Clusters
- General Enhancements
- Changes Affecting Compatibility
- Upgrade Procedures
- Download
- Known Issues in 3.6.3
- Known Issues in 3.6.2
- Known Issues in 3.6.1
- Known Issues in 3.6.0
- Report an Issue
Minor Releases¶
3.6.20 - Sep 14, 2020¶
Issues Fixed:
- SERVER-40317: $facet execution has no limit on how much memory it can consume
- SERVER-41600: IndexCatalog::refreshEntry should invalidate the index from the CollectionInfoCache
- SERVER-43233: Add ability to request only specific attribute(s) for the LDAP groups
- SERVER-48709: signing key generator thread on config server not waken up as expected
- SERVER-48993: explodeForSort can produce incorrect query plan
- All JIRA issues closed in 3.6.20
- 3.6.20 Changelog
3.6.19 - Jul 23, 2020¶
Issues Fixed:
- SERVER-46487 The mongos routing for scatter/gather ops can have unbounded latency
- SERVER-45610 Some reads work while system is RECOVERING
- SERVER-47695 Write commands run by threads that can survive rollback can fail operationTime invariant in ServiceEntryPoint
- SERVER-47994 Fix for numerical overflow in GeoHash
- SERVER-47686 Upgrade static OpenSSL to 1.1.1g
- All JIRA issues closed in 3.6.19
- 3.6.19 Changelog
3.6.18 - May 1, 2020¶
Issues fixed:
- SERVER-46466: Race with findAndModify retryable write and session migration
- SERVER-45935: [3.6] WT error handler incorrectly prints “An unsupported journal format detected”
- SERVER-45418: DocumentSourceCursor batching memory accounting does not account for empty documents, leads to unbounded memory use for count-like aggregates
- SERVER-45363: Issue with mongodb text indexes and weights when using wildcard specifier
- SERVER-35050: Don’t abort collection clone due to negative document count
- All JIRA issues closed in 3.6.18
- 3.6.18 Changelog
Note
Fixed issues include those that resolve the following Common Vulnerabilities and Exposure (CVE):
- CVE-2020-7921 (See SERVER-45472)
3.6.17 - Jan 27, 2020¶
Issues fixed:
- SERVER-42565: Aggregations and find commands sort missing fields differently
- SERVER-44733: Change stream should throw ChangeStreamFatalError if a single shard cannot be targeted for updateLookup
- SERVER-44174: $push and $addToSet should restrict memory usage
- SERVER-37772: Platform Support: Add Community & Enterprise RHEL 8 x64
- SERVER-45396: fix the “me” field in isMaster responses when using splithorizon
- WT-5042: Reduce configuration parsing overhead from checkpoints
- All JIRA issues closed in 3.6.17
- 3.6.17 Changelog
3.6.16 - Dec 6, 2019¶
Issues fixed:
- SERVER-38002: Upgrade Pcre to at least 8.42
- SERVER-35463: Mark listCommands as pre-auth
- SERVER-44584: Rewriting updates as modifications incorrectly considers logging state
- All JIRA issues closed in 3.6.16
- 3.6.16 Changelog
3.6.15 - Nov 6, 2019¶
Issues fixed:
- SERVER-39004: Introduce a quota mechanism for the overflow file
- SERVER-38994: Step down on SIGTERM
- SERVER-40355: rs.config that contains an _id greater than the number of nodes will crash
- SERVER-43074: Do not use a global variable to encode ‘multikeyPath’ information when writing out catalog documents
- WT-4869: Stop adding cache pressure when eviction is falling behind
- All JIRA issues closed in 3.6.15
- 3.6.15 Changelog
3.6.14 - Aug 26, 2019¶
Issues fixed:
- SERVER-36443: Long-running queries should not cause a build-up of unused ChunkManager objects
- SERVER-38945: SSL performance regression
- SERVER-40134: Distinct command against a view can return incorrect results when the distinct path is multikey
- SERVER-42055: Only acquire a collection IX lock to write the lastVote document
- SERVER-42603: Recent service file change may cause cyclic dependencies
- SERVER-42089: Platform Support: Remove Enterprise RHEL 6.7 zSeries
- WT-4803: Implement file_max configuration for Cache Overflow mechanism
- All JIRA issues closed in 3.6.14
- 3.6.14 Changelog
Note
Fixed issues include those that resolve the following Common Vulnerabilities and Exposures (CVEs):
- CVE-2019-2389 (See SERVER-40563)
- CVE-2019-2390 (See SERVER-42233)
3.6.13 - Jun 10, 2019¶
Issues fixed:
- SERVER-16571: Use Actual Memory Constraint vs. Total System Memory When They Differ
- SERVER-38984: Attach IDs to users
- SERVER-41213: Unique background index builds may produce inconsistent keys
- SERVER-37765: Platform Support: Remove Ubuntu 14.04
- SERVER-37774: Platform Support: Remove Enterprise Ubuntu 16.04 PPCLE
- All JIRA issues closed in 3.6.13
- 3.6.13 Changelog
Note
Fixed issues include those that resolve the following Common Vulnerabilities and Exposures (CVEs):
- CVE-2019-2386 (See SERVER-38984)
3.6.12 - April 8, 2019¶
Issues fixed:
- SERVER-35219: Regain MongoDB balancer performance with sessions
- SERVER-39847: Migrating session info can trigger fassert when destination shard has transaction history truncated by oplog
- SERVER-37255: replSetReconfig with concurrent election can trigger invariant
- SERVER-39394: Views collation check should traverse through nested $lookup pipelines
- SERVER-39487: Aggregation operator $sqrt output for NumberDecimal is the same as $exp
- SERVER-37722: Quiet mode does not suppress connection end events
- All JIRA issues closed in 3.6.12
- 3.6.12 Changelog
3.6.11 - Mar 1, 2019¶
Issues fixed:
- SERVER-32146: Log slow oplog entry application
- SERVER-9043: Link flushRouterConfig on the config server and shards
- SERVER-37624: Sessions never expire when you change value of localLogicalSessionTimeoutMinutes
- SERVER-38344: Early release of distributed database locks during initial collection sharding results in migration/split failures
- SERVER-39166: $graphLookup should force a pipeline to split in sharded cluster
- All JIRA issues closed in 3.6.11
- 3.6.11 Changelog
3.6.10 - Jan 22, 2019¶
Issues fixed:
- SERVER-29825: Do not allow rename from unreplicated to replicated DB or vice-versa
- SERVER-35936: MongoDB Community installer fails to install Compass
- SERVER-36977: Initial mongod.log is created using umask vs mode 600
- SERVER-37080: Implement
tuneable batch size for chunk migrations
- SERVER-37616: Implement
tuneable batch size for the rangedeleter
- SERVER-37182: Different values when referencing whole object vs. a field of that object after $arrayToObject
- SERVER-37200: $match stage following $listSessions not working against mongos
- All JIRA issues closed in 3.6.10
- 3.6.10 Changelog
3.6.9 - Nov 16, 2018¶
Issues fixed:
- SERVER-32943: Query during background index build can lead to suboptimal cached plan
- SERVER-35657: Do not delay journal flushes when operations are waiting for oplog visibility
- SERVER-37058: Update with numeric field names inside an array can cause validation to fail
- SERVER-37132: Negation of $in with regex can incorrectly plan from the cache, leading to missing query results
- All JIRA issues closed in 3.6.9
- 3.6.9 Changelog
3.6.8 - Sep 19, 2018¶
Issues fixed:
- SERVER-34204: Tailable cursor fails on getMore against a sharded cluster.
- SERVER-34846: Covered index with collated field returns incorrect result when collation not involved in match or sort.
- SERVER-37002: Dropping a collection with long index names via rename fails under MMAPv1.
- SERVER-36149: Fix privileges on setFCV virtual namespace.
- All JIRA issues closed in 3.6.8
- 3.6.8 Changelog
3.6.7 - Aug 25, 2018¶
Issues fixed:
- SERVER-27725: Use batch insert when migrating chunks
- SERVER-36070: Aggregation with $out results in error when Auditing is enabled
- TOOLS-2075: mongoreplay always replays to secondary
- All JIRA issues closed in 3.6.7
- 3.6.7 Changelog
3.6.6 - Jul 10, 2018¶
Issues fixed:
- SERVER-33538:
mapReduce
“replace” on a sharded output collection can lead to UUIDCatalog inconsistencies. - SERVER-32999: Platform Support: Remove Debian 7 builds.
- SERVER-5461 Add
syncSourceHost
field toreplSetGetStatus
output. - All JIRA issues closed in 3.6.6
- 3.6.6 Changelog
3.6.5 - May 29, 2018¶
Issues fixed:
- SERVER-34746: Segmentation fault when shard is started with
--shardsvr
before being added to a shard. - SERVER-34423: collMod interruption may cause invariant failure.
- SERVER-34390: MongoDB binaries for macOS support TLS 1.2.
- SERVER-29463: Platform Support: add Debian 9 “stretch”.
- SERVER-34399: $changeStream with invalid resume token crashes the server.
- SERVER-29301: Upgrade MozJS to ESR 45.9.0.
- All JIRA issues closed in 3.6.5
- 3.6.5 Changelog
3.6.4 - Apr 13, 2018¶
Issues fixed:
- SERVER-28670: Add sharding metadata refresh metrics section to serverStatus.
- SERVER-32677: Segmentation fault converting ReplicaSet to Replicated Shard Cluster.
- SERVER-33763: 3.6 drivers fail to communicate with 3.6 sharded clusters running at FCV 3.4.
- SERVER-32923: Remove SLES11 support.
- SERVER-31535: Remove Ubuntu 12.04 support.
- SERVER-32498:
currentOp
output no longer returnsthreadId
field. - All JIRA issues closed in 3.6.4
- 3.6.4 Changelog
3.6.3 - February 23, 2018¶
Issues fixed:
- SERVER-32441: 3.6
mongod
crash on find with index and nested$and
/$or
. - SERVER-32606: Tailing oplog on secondary fails with CappedPositionLost
- SERVER-32631: specifying
--bind_ip
localhost results in error “address already in use” - All JIRA issues closed in 3.6.3
- 3.6.3 Changelog
3.6.2 - January 10, 2018¶
Issues fixed:
- SERVER-31267: CollectionCloner fails if collection is dropped between getMore calls.
- SERVER-31625: The contents of
Unknown macro: {USER}
needs to be escaped when querying for the groups using LDAP server. - SERVER-31684: QueryPlanKilled (operation exceeded time limit) in $changeStream with updateLookup.
- SERVER-31982: Shard does not call config commit chunk migration command with majority writeConcern nor checks for writeConcern errors..
- SERVER-32246: PID file permission on v3.6 make it not monitorable using pid file.
- SERVER-32255: UUIDs may be absent from shard secondary local collections.
- SERVER-32282: Aggregation text search returns text score even if it wasn’t requested when targeting multiple shards in a sharded cluster.
- SERVER-32396
mongo
shell failed to connect with 3.6 connection string SRV - SERVER-32430: DocumentSourceSort sorts array documents incorrectly if there is a non-simple collation.
- SERVER-32529: Requiring replSet for shards breaks Queryable Backup.
- All JIRA issues closed in 3.6.2
- 3.6.2 Changelog
3.6.1 - Dec 26, 2017¶
Issues fixed:
- SERVER-30768: Primary queries using maxTimeMS cause temporary shard write unavailability if ExceededTimeLimit.
- SERVER-31225: The
mongod
process forks before listening for connections. - SERVER-31885: changeStream cursor is not returned on a
mongos
when the database does not exist. - SERVER-32085: changeStream reports incorrect documentKey for unsharded collections that become sharded.
- SERVER-32046: Arrays of certain NumberDecimals can trigger an invariant failure.
- SERVER-32048: Updates using a numeric path component may cause index entries not to be created.
- All JIRA issues closed in 3.6.1
- 3.6.1 Changelog
Security¶
Default Bind to Localhost¶
Starting with MongoDB 3.6, MongoDB binaries, mongod
and
mongos
, bind to localhost
by default.
From MongoDB versions 2.6 to 3.4, only the binaries from the
official MongoDB RPM (Red Hat, CentOS, Fedora Linux, and derivatives)
and DEB (Debian, Ubuntu, and derivatives) packages would bind to
localhost
by default. To learn more about this change, see
Localhost Binding Compatibility Changes.
MONGODB-CR
Deprecation¶
As of MongoDB 3.6, MONGODB-CR
authentication mechanism is
deprecated. If you have not upgraded your MONGODB-CR
authentication
schema to SCRAM, see Upgrade to SCRAM.
Authentication Restrictions¶
To restrict database user connections to specified IP addresses, added
authenticationRestrictions
parameter to the following:
Commands | Methods |
---|---|
createUser |
db.createUser() |
updateUser |
db.updateUser() |
createRole |
db.createRole() |
updateRole |
db.updateRole() |
Additional Security Enhancements¶
Added the
opensslCipherConfig
parameter to control the OpenSSL ciphers when using TLS/SSL encryption.If authentication is turned on, you can only issue a
getMore
against cursors you created.Added the
convertToCapped
action torestore
role.See also
Aggregation¶
The following features are available starting in MongoDB 3.6.
More Expressive $lookup
¶
$lookup
adds support for specifying multiple join
conditions as well as uncorrelated sub-queries by allowing variable
specification and pipeline execution on the joined collection.
For details, see the $lookup
syntax for
Join Conditions and Uncorrelated Sub-queries.
New Aggregation Stages¶
MongoDB 3.6 adds the following new aggregation pipeline stages:
Operator | Description |
---|---|
$currentOp |
Returns a stream of documents containing information
on active and/or dormant operations on a mongod
instance. Uses the new aggregation helper
db.aggregate() . |
$listSessions |
Lists server sessions in the system.sessions collection in the
config database. Uses the new aggregation helper
db.aggregate() . |
$listLocalSessions |
Lists server sessions cached in memory by the server. |
New Aggregation Operators¶
Operator | Description |
---|---|
$arrayToObject |
Converts an array of key value pairs to a document. |
$objectToArray |
Converts a document to an array of documents representing key-value pairs. |
$mergeObjects |
Combines multiple documents into a single document. |
$dateFromString |
Converts a date/time string to a date object. |
$dateFromParts |
Constructs a BSON Date object given the date’s constituent parts. |
$dateToParts |
Returns a document containing the constituent parts of a date. |
New Aggregation Helper¶
MongoDB 3.6 adds a helper, db.aggregate()
, to perform
aggregations that do not rely on an underlying collection, such
as those that start with $currentOp
or
$listLocalSessions
.
New Options¶
aggregate
command and the
db.collection.aggregate()
method support the following new options:
hint
option to specify the index to use.Note
The
hint
does not apply to$lookup
and$graphLookup
stages.comment
option to help trace the operation through the database profiler, currentOp, and logs.
Support for Time Zones¶
MongoDB 3.6 adds support for time zones to aggregation date operators.
See also
Array Update Operator Enhancements¶
arrayFilters
¶
The following commands and methods can accept an arrayFilters
parameter to specify which elements to modify in an array field:
Commands | Methods |
---|---|
findAndModify |
|
update |
|
|
Multi-Element Array Updates¶
MongoDB 3.6 adds the following new positional operators for update operations on arrays and nested arrays:
- The all positional
$[]
operator updates all elements in an array. - The filtered positional
$[<identifier>]
operator updates all elements in an array that match the arrayFilters criteria.
3.6 Compatible Drivers¶
The following drivers are feature compatible with MongoDB 3.6:
Java 3.6+ Python 3.6+ C 1.9+ |
C# 2.5+ Node 3.0+ Ruby 2.5+ |
Perl 2.0+ PHPC 1.4+ Scala 2.2+ |
DNS-constructed Seedlist mongodb+srv
¶
In addition to the standard connection format, the 3.6 drivers support a DNS-constructed seedlist. For more information, see DNS Seed List Connection Format.
Change Streams¶
MongoDB 3.6 supports opening change streams against replica sets and sharded clusters with replica set shards.
Change streams allow applications to access real-time data changes without the complexity and risk of tailing the oplog. Applications can use change streams to subscribe to all data changes on a collection and respond to those changes.
You can open a change stream from any 3.6-series driver using the
db.collection.watch()
method. See the documentation for
your preferred driver for complete instructions on usage.
See Change Streams for more information.
Important
To use change streams, featureCompatibilityVersion
must be set to
“3.6”. For more information, see View FeatureCompatibilityVersion and
setFeatureCompatibilityVersion
.
Client Sessions¶
Causal Consistency¶
To provide causal consistency, MongoDB 3.6 enables causal consistency in client sessions. A causally consistent client session denotes that the associated sequence of read and acknowledged write operations have a causal relationship that is reflected by their ordering. Client applications must ensure that only one thread at a time executes these operations in a client session.
Applications can start a client session and associate operations with a specific session. Applications must ensure that only one thread at a time executes these operations in a client session.
Important
To use client sessions:
- Clients require MongoDB drivers updated for MongoDB 3.6.
featureCompatibilityVersion
must be “3.6”. For more information, see View FeatureCompatibilityVersion andsetFeatureCompatibilityVersion
.
Retryable Writes¶
Important
For retryable writes:
- Clients require MongoDB drivers updated for MongoDB 3.6.
featureCompatibilityVersion
must be “3.6”. For more information, see View FeatureCompatibilityVersion andsetFeatureCompatibilityVersion
.
Starting in MongoDB 3.6, certain acknowledged write operations on replica sets and sharded clusters are “retryable” to provide handling of transient network errors or replica set elections.
With retryable writes, MongoDB drivers automatically retries these
operations upon encountering network errors or encountering a
replica set failover during which
time the replica set has no primary. To enable retryable writes for the
3.6 drivers, see retryWrites
.
As the retry attempt is made only once, the retryable feature can help address transient network errors but not persistent network errors.
For more information on retryable writes, see Retryable Writes.
mongo
Shell Changes¶
MongoDB 3.6 adds the new command-line option --retryWrites
to
the mongo
shell. The option enables
Retryable Writes in the
mongo
shell.
The following new methods have been added to the Mongo()
connection
object in the mongo
shell:
Mongo.isCausalConsistency()
Mongo.startSession()
- Various
Session
methods - Various
SessionOptions
methods
Server Sessions¶
MongoDB’s server sessions, or logical sessions, are the underlying framework used by client sessions to support Causal Consistency and retryable writes.
Important
Applications use client sessions to interface with server sessions.
Server sessions are available for standalone mongod
instances, replica sets, and sharded clusters.
Server Session Commands¶
Starting in 3.6, MongoDB drivers associate all operations with a server session, with the exception of unacknowledged writes. If the deployment enforces authentication/authorization, server sessions are associated with the authenticated users.
The following commands can be used to list, manage, and kill server sessions throughout MongoDB clusters:
Commands | Descriptions |
---|---|
endSessions |
Expires specified server sessions. |
killAllSessions |
Kills all server sessions. |
killAllSessionsByPattern |
Kills all server sessions that match the specified pattern. |
killSessions |
Kills specified server sessions. |
refreshSessions |
Refreshes idle server sessions. |
startSession |
Starts a new server session. |
Parameters¶
The following new parameters are available for server sessions:
logicalSessionRefreshMinutes
localLogicalSessionTimeoutMinutes
maxAcceptableLogicalClockDriftSecs
Aggregation Stages¶
To support server sessions, MongoDB 3.6 adds the following new aggregation pipeline stages:
Operator | Description |
---|---|
$listSessions |
Lists the server sessions in the system.sessions collection in the
config database. |
$listLocalSessions |
Lists the server sessions cached in memory by the server. Uses
the new aggregation helper db.aggregate() . |
General¶
serverStatus
returns information on the number of
logicalSessionRecordCache.
Command Options¶
Starting in 3.6, MongoDB drivers associate all operations with a server session, with the exception of unacknowledged writes. The following options are available for all commands to support association with a server session:
Important
The mongo
shell and the drivers assign these options
to the commands in the session.
Option | Type | Description |
---|---|---|
lsid |
Document | The document that specifies the unique id of the session
associated with the command. If the txnNumber is
specified, the lsid is required. |
txnNumber |
64-bit integer | A strictly increasing non-negative number that uniquely identifies the command in the command’s session. If specified, the command must also include the |
For the delete
, insert
, and update
commands that take an array of statements, the following option is also
available:
Important
Do not manually set stmtIds
. MongoDB sets the stmtIds
to be strictly increasing non-negative numbers.
Option | Type | Description |
---|---|---|
stmtIds |
Array of 32-bit integers | Array of numbers that uniquely identify their respective write operations within the write command. |
JSON Schema¶
MongoDB 3.6 adds the $jsonSchema
operator to support
document validation using JSON
Schema. For details, see $jsonSchema
.
To use $jsonSchema
, featureCompatibilityVersion
must be
set to “3.6”.
See also
Replica Sets¶
- Deprecate replica set protocol version 0 (
pv0
). For more information on the replica set protocol versions, see Replica Set Protocol Version. - Added the
replSetResizeOplog
command to dynamically resize a replica set member’s oplog. Available for instances running the WiredTiger storage engine. - Added the catchUpTakeoverDelayMillis configuration option, dictating the amount of time a node waits to run for election after determining that it is ahead of the current primary.
- For replica sets that use the protocol version 1 (
pv1
), arbiters will vote no in elections if they detect a healthy primary of equal or greater priority to the candidate. For more information on replica set protocol versions, see Replica Set Protocol Version. - Added the
oplogInitialFindMaxSeconds
parameter to adjust how long a member of a replica set should wait for itsfind
command to finish during data synchronization. - Added the
waitForSecondaryBeforeNoopWriteMS
parameter to specify how long a secondary must wait if theafterClusterTime
is greater than the last applied time from the oplog. - Added support for running the following during a replica set member’s
initial sync:
renameCollection
convertToCapped
$out
stages in aggregation pipelines- Map-reduce jobs that output to a new collection
Sharded Clusters¶
Starting in 3.6, shards must be replica sets. To upgrade your sharded cluster to version 3.6, the shard servers must be running as a replica set.
To convert an existing shard standalone instance to a shard replica set, see Convert a Shard Standalone to a Shard Replica Set.
Starting in 3.6, all members of a shard replica set, not just the primary, maintain the metadata regarding chunk metadata. This prevents reads from the secondaries from returning orphaned data when not using
"available"
read concern.Added
ShardingTaskExecutorPoolMaxConnecting
parameter formongos
to control the rate at whichmongos
adds connections to amongod
instance.Added
orphanCleanupDelaySecs
that determines the minimum delay before a migrated chunk is deleted from the source shard.The
config.system.sessions
collection in theconfig
database may now be sharded.
General Enhancements¶
MongoDB Compass Packaging¶
The MongoDB Server download is packaged with a platform-specific installation script for MongoDB Compass Community Edition. This script installs MongoDB Compass as part of the MongoDB Server installation process.
Collection Identifier¶
Collections have an immutable unique identifier. The featureCompatibilityVersion
must be set
to "3.6"
.
See also
New Query Operators¶
MongoDB 3.6 adds the following new query operators:
- The new
$jsonSchema
operator matches documents that validate against the given JSON Schema. To use$jsonSchema
,featureCompatibilityVersion
must be set to “3.6”. - The
$expr
allows the use of aggregation expressions within the query language.
See also
Removed Operators¶
MongoDB 3.6 removes the deprecated $pushAll
update operator. For more
information, see Remove $pushAll Update Operator.
Indexes¶
- An index can cover a query on fields within nested documents.
- Multikey indexes can cover queries over non-array keys if the index tracks which fields make it multikey.
- When creating an index, you cannot specify
*
as the name of the index. See Indexes Named *.
Commands¶
- Added the following options to the
listDatabases
command:nameOnly
to return only the database names (which does not require database locks) instead of returning both the database names and size information (which does require database locks).filter
to return databases that match the specified match criteria on the outputs.
- Modified behavior of the
validate
command and thedb.collection.validate()
method such that only forfull
validation does the WiredTiger storage engine force a checkpoint, flush all in-memory data to disk, then verify the on-disk data. See also validate Operation. - The
<database>.system.profile
entry forupdate
anddelete
contains the entire update/delete document applied to the named collection. dropDatabase
waits until all collections drops in the database have propagated to a majority of the replica set members.- For commands run on replica sets and sharded clusters, the response
documents include the
operationTime
and$clusterTime
. Seedb.runCommand
anddb.adminCommand
. - Deprecated getPrevError command.
Wire Protocol and Compression¶
- MongoDB 3.6 introduces a new wire protocol opcode called OP_MSG. This opcode’s message format is extensible and designed to subsume the functionality of other opcodes.
- MongoDB 3.6 adds support for the
zlib
compressor for--networkMessageCompressors
option (ornet.compression.compressors
setting if using the configuration file).--networkMessageCompressors
option (ornet.compression.compressors
setting) enables network compression for communication amongmongod
,mongos
,mongo
shell, and drivers that support theOP_COMPRESSED
message format. mongod
andmongos
enable network compression by default withsnappy
as the compressor. For more information on network compression, seenet.compression.compressors
.
Read Concern¶
New
"available"
read concern is available. For unsharded collections (including collections in a standalone deployment or a replica set deployment),"local"
and"available"
read concerns behave identically. For sharded clusters,"available"
provides greater tolerance for partitions but may return orphan documents if the shard is undergoing chunk migrations.See also
Starting in MongoDB 3.6, MongoDB enables support for “majority” read concern by default.
For MongoDB 3.6.1 - 3.6.x, you can disable read concern “majority” to prevent the storage cache pressure from immobilizing a deployment with a primary-secondary-arbiter (PSA) architecture. Disabling “majority” read concern also disables support for change streams
For more information, see Disable Read Concern Majority.
FTDC¶
MongoDB 3.6 adds support for Diagnostics Capture (also known as FTDC
) in
mongos
. [1] In previous versions, the feature is available for
mongod
only. See Diagnostic Parameters.
Note
FTDC is enabled by default.
[1] | MongoDB 3.4.14+ also adds mongos support for FTDC. |
Additional Enhancements¶
MongoDB 3.6 includes the following enhancements:
- Added support for specifying full Unix domain socket paths with
--bind_ip
. mongod
now offers a--timeZoneInfo
option. Use this option to specify the path to your system time zone database. The default configuration file included with Linux and macOS packages sets this to/usr/share/zoneinfo
.- Date operations now accept consistent date ranges across all supported
operating systems. You may safely work with years in the range
0
through9999
. - The new
honorSystemUmask
startup option formongod
causes new files created by MongoDB to have the read/write permissions specified by the umask of the user who runs themongod
process. Only available on Linux and macOS systems. - The
maxWriteBatchSize
limit of a database, which indicates the maximum number of write operations permitted in a write batch, raises from1,000
to100,000
. - The
planCacheListPlans
database command produces the same output as thePlanCache.getPlansByQuery()
shell method. The output from both operations now includes a timestamp for when the plans were generated. - The new
KeysRotationIntervalSec
server parameter specifies the number of seconds for which an HMAC signing key is valid before rotating to the next one. - The
find
command’s optionoplogReplay
now supports the$eq
operator.
Changes Affecting Compatibility¶
Some changes can affect compatibility and may require user actions. For a detailed list of compatibility changes, see Compatibility Changes in MongoDB 3.6.
Upgrade Procedures¶
Feature Compatibility Version 3.4
To upgrade, the 3.4 instances must have
featureCompatibilityVersion
set to 3.4. To check the version:
For specific details on verifying and setting the
featureCompatibilityVersion
as well as information on other
prerequisites/considerations for upgrades, refer to the individual
upgrade instructions:
If you need guidance on upgrading to 3.6, MongoDB offers major version upgrade services to help ensure a smooth transition without interruption to your MongoDB application.
Known Issues in 3.6.3¶
- WT-3724:
- MongoDB 3.6 is not tested on APFS, the new filesystem in macOS 10.13 and may encounter errors.
- Operations that build in-memory structures are not causally consistent; however, only some operations return errors when associated with causally consistent sessions. See Causal Consistency Limitations.
Known Issues in 3.6.2¶
- SERVER-31760:
$expr
does not use indexes for equality match against field, including when used as part of a$lookup
aggregation stage with foreign pipeline.
- WT-3724:
- MongoDB 3.6 is not tested on APFS, the new filesystem in macOS 10.13 and may encounter errors.
- Operations that build in-memory structures are not causally consistent; however, only some operations return errors when associated with causally consistent sessions. See Causal Consistency Limitations.
Known Issues in 3.6.1¶
- SERVER-31760:
$expr
does not use indexes for equality match against field, including when used as part of a$lookup
aggregation stage with foreign pipeline.
- WT-3724:
- MongoDB 3.6 is not tested on APFS, the new filesystem in macOS 10.13 and may encounter errors.
- Operations that build in-memory structures are not causally consistent; however, only some operations return errors when associated with causally consistent sessions. See Causal Consistency Limitations.
Known Issues in 3.6.0¶
- SERVER-31760:
$expr
does not use indexes for equality match against field, including when used as part of a$lookup
aggregation stage with foreign pipeline.
- TOOLS-1827:
- When using a URI with SRV, fetched TXT records will be ignored if no query parameters are specified in the command line URI. To get around this, explicitly specify the TLS/SSL setting used to communicate with the database (either ssl=true or ssl=false) in the URI query string.
- WT-3724:
- MongoDB 3.6 is not tested on APFS, the new filesystem in macOS 10.13 and may encounter errors.
- Operations that build in-memory structures are not causally consistent; however, only some operations return errors when associated with causally consistent sessions. See Causal Consistency Limitations.
See also
Report an Issue¶
To report an issue, see https://github.com/mongodb/mongo/wiki/Submit-Bug-Reports for instructions on how to file a JIRA ticket for the MongoDB server or one of the related projects.