Docs Menu
Docs Home
/
MongoDB Ops Manager
/

Automation Agent Changelog

On this page

Note

Announcement

The Monitoring and Backup Agent have been merged into the Automation Agent, which will now be known as the MongoDB Agent. Learn more about this change.


Released with Ops Manager 4.0.17 on 2020-02-06

  • Redacts sensitive configuration data in MongoDB Agent log files to improve security.

Released with Ops Manager 4.0.16 on 2019-11-07

  • Adds support for the businessCategory field in extended validation TLS certificates.

  • The Automation Agent is now built using Go 1.13.

Released with Ops Manager 4.0.15 on 2019-09-05

  • Fixes a bug where the MongoDB Agent could panic and delete files from its current working directory.

Released with Ops Manager 4.0.14 on 2019-07-31

  • Fix: Restore downloads performed by Automation Agent now respect the configured sslTrustedMMSServerCertificate Ops Manager parameter.

Note

sslTrustedMMSServerCertificate has been deprecated. Use httpsCAFile instead.

Released with Ops Manager 4.0.13 on 2019-07-04

Released with Ops Manager 4.0.12 on 2019-06-06

  • Fix: Automation Agent now periodically closes all idle HTTP connections.

  • Fix: Allow Automation Agent to connect to mongos through either "short" or "long" hostnames.

  • Fix: Fix failure in the make-init-scripts tool.

  • Fix: Lower the default HTTP Header time out in the Automation Agent from 15 minutes to 30 seconds. This ensures faster failures in the event of Ops Manager service infrastructure changes, such as load balancers moving nodes out of a pool.

  • Fix: When restoring a sharded cluster, remove all sharding metadata from the config.system.sessions collection. If you do not remove the sharding metadata, MongoDB cannot to re-create this collection after the restore is complete.

Released with Ops Manager 4.0.8 on 2019-02-07

  • Fix: Reduce the amount of memory used to generate plans for large sharded clusters.

  • Fix: Rotation of log files when MongoDB log rotation is enabled and the maximum number of uncompressed log files is set to two.

Released with Ops Manager 4.0.7 on 2019-01-10

  • Add support for the net.ssl.certificateSelector configuration option.

  • Fix: Allow the successful change in the storageEngine for standalone binary: mongod with TLS/SSL enabled.

  • Fix: Automation Agent no longer attempts to authenticate to arbiters that are configured to use X.509 for clusterAuthMode.

Released with Ops Manager 4.0.6 on 2018-12-10

Released with Ops Manager 4.0.5 on 2018-11-01

  • Add support for the ssl.FIPSMode parameter.

  • Fix: Memory leak when using Server Pools feature.

Released with Ops Manager 4.0.4 on 2018-10-12

  • Critical Fix: MongoDB 4.0 restores may fail if snapshot downloads for nodes in the replica set finish at significantly different times.

  • Fix: Restore the ability to upgrade from authSchemaVersion 3 to 5 in a sharded cluster.

Released with Ops Manager 4.0.3 on 2018-10-04

  • Fix: Automation Agent can add new MongoDB users even if SCRAM-SHA-1 authentication was enabled for the deployment after enabling SCRAM-SHA-256.

Released with Ops Manager 4.0.2 on 2018-09-06

  • Storage affecting field changes in a replica set could result in data loss if processes were restarted unexpectedly.

  • Automation Agent was unable to properly determine state if the process ID previously used by a MongoDB process was reclaimed by a different process after an unexpected server reboot.

  • Fix: Allow CSRS members to be shut down.

Released with Ops Manager 4.0.1 on 2018-08-02

  • Critical Fix: Set TasksMax=infinity and TasksAccounting=false in systemd scripts for Automation Agent.

  • Fix: Automation Agent handling of enterprise builds for Amazon Linux 2.

Released with Ops Manager 4.0.0 on 2018-06-27

  • Added support for MongoDB 4.0.

  • Added support for SCRAM-SHA-256

  • Added platform support for:

    • zLinux on RHEL 6

    • Debian 9

    • Amazon Linux 2

Important

MongoDB 4.0 Community Edition requires libcurl. Install libcurl before using Ops Manager to install MongoDB 4.0 Community.


Released with Ops Manager 3.6.10 on 2019-01-10

  • Fix: Allow CSRS replica set members to be shut down.

Released with Ops Manager 3.6.9 on 2018-11-01

  • Critical Fix: Storage affecting field changes in a replica set could result in data loss if processes were restarted unexpectedly.

  • Fix: Memory leak when using Server Pools feature.

Released with Ops Manager 3.6.8 on 2018-08-02

  • Critical Fix: Set TasksMax=infinity and TasksAccounting=false in systemd scripts for Automation Agent.

  • Ends support for SLES 11 and Ubuntu 12.04.

  • Automation Agent support for BI Connector 2.5.

  • Compound Text Indexes through Automation.

  • Fix: When collecting logs, ignore errors for files that have disappeared.

  • Replace evergreen Ubuntu 12 testing with Ubuntu 14.

  • Update version of BI Connector tested to 2.5.0.

Released with Ops Manager 3.6.6 on 2018-05-03

  • Fix: Automation Agents should not attempt to rotate BI Connector log files for BI Connectors on other servers.

Released with Ops Manager 3.6.5 on 2018-04-05

  • Fix: Disable MongoDB's TTL monitor thread while applying oplogs during a point in time restore.

  • When performing maintenance on a MongoDB 3.2 shard member for oplog resizing, start the node with --recoverShardingState false.

  • When the Automation Agent performs a resize on a MongoDB 3.2 sharded cluster, disable sharding recovery while the node is started up as a standalone.

Released with Ops Manager 3.6.4 on 2018-03-01

  • Changes made to any storage-affecting options in MongoDB will automatically result in a rolling initial sync of the replica set.

    For single node replica sets and standalones a mongodump / mongorestore will be performed. These options include security.enableEncryption, storage.smallfiles, storage.directoryPerDb and wiredTiger.directoryForIndexes. (The storage.engine parameter has always had this treatment.)

  • Fix: Automation Agent correctly resizes the oplog for MongoDB clusters that use X-509 for cluster authentication.

Note

RELEASE ADVISORY

Ops Manager 3.6.4 fixes an issue in which setting a value for a setParameter field using Automation may not have resulted in appropriate restart of the MongoDB cluster. As a consequence of this fix, clusters in which a setParameter field is specifically set to the default value for the setParameter may experience a rolling restart on upgrade to Ops Manager 3.6.4.

When configuring a setParameter field in the MongoDB configuration via Automation, always perform a rolling restart.

Released with Ops Manager 3.6.3 on 2018-02-01

  • Fix: Automation Agent determination of Goal State was incorrect for multi-server deployments using the ldap.bind.queryPassword parameter. This is now resolved and rolling changes will proceed correctly on these deployments.

  • Fix: Rotation of BI Connector logs by the Automation Agent for timezones with positive GMT offsets.

Released with Ops Manager 3.6.2 on 2018-01-11

  • Allow user to specify sampleRefreshIntervalSecs and sampleSize BI Connector flags

  • Fix: Relax validation when krb5ConfigLocation parameter is specified. This no longer implies that krb5Principal and krb5Keytab are required.

  • Fix: BI Connector Log Rotation config now respects timestamps from UTC hosts.

  • Fix: Improve logic controlling when the Backup Agent uses the Primary as a sync source.

Released with Ops Manager 3.6.1 on 2017-12-19

  • Fix: Prevent race condition when MongoDB version and FCV are updated at the same time.

  • Manage Windows Firewall rules for the BI Connector.

Released with Ops Manager 3.6.0 on 2017-12-05

  • Support for MongoDB 3.6.

  • Support for advanced replica set configuration fields.

  • Support for new Agent API Key model.


Released with Ops Manager 3.4.14 on 2018-05-03

Released with Ops Manager 3.4.13 on 2018-04-05

  • Fix: Automation Agent determination of Goal State was incorrect for multi-server deployments using the ldap.bind.queryPassword parameter. This is now resolved and rolling changes will proceed correctly on these deployments.

Released with Ops Manager 3.4.12 on 2018-02-01

  • Logging enhancements

Released with Ops Manager 3.4.10 on 2017-11-02

Released with Ops Manager 3.4.9 on 2017-10-05

  • Fix failure to collect disk-related hardware statistics on some hardware configurations.

  • When Automation creates a temporary Windows service in order to perform maintenance operations on a mongod, remove the service when the maintenance is completed.

Released with Ops Manager 3.4.7 on 2017-08-03

  • Optimization to reduce the number of checks to see if a process is running.

  • Improve state detection during conversions to config server replica sets.

Released with Ops Manager 3.4.6 on 2017-07-06

  • Fix: During CSRS conversion, use differently named log files for temporary config servers.

  • Fix: During CSRS conversion, only shut down a member when it is in secondary state.

  • Fix: Config file parsing code on Windows did not parse all possible options.

Released with Ops Manager 3.4.5 on 2017-05-18

  • When performing an automated restore to a sharded cluster with different shard names, update the shard identity document.

  • When performing an automated restore, ensure that shard metadata is always updated in the right order.

  • When performing an automated restore, always restore to the default protocol version.

  • Fix: RHEL7 packaging so that Automation Agent starts on server boot.

  • Reduce frequency at which Automation Agent checks managed log files to reduce CPU overhead.

  • Ignore get_mempolicy errors and assume numa not enabled.

Released with Ops Manager 3.4.4 on 2017-03-30

  • Fix problem with shutdown of mongod processes during an automated restores on Windows.

  • Fix for issues using Automation in multi-server deployments using SSL and encrypted PEM key files.

  • Optimization for goal state maintenance of sharded clusters. Automation Agents will execute far fewer commands in steady state.

Released with Ops Manager 3.4.3 on 2017-02-17

  • Fix bug in removal of shards for sharded clusters on MongoDB 3.4.

  • Built with Go 1.7.

  • Support for MacOS Sierra.

Released with Ops Manager 3.4.2 on 2017-01-19

  • Fix: Can install Agent on Windows if the Windows Firewall was disabled.

  • Fix: Can use MONGODB-CR for Agent authentication when LDAP was being used for User authentication.

  • Fix: Issue where Agent would stop sending status after MongoDB reaches its connection limit.

Released with Ops Manager 3.4.1 on 2016-12-27

  • Fix: Can install MongoDB on Power Linux when using Ops Manager in 'Local Mode'.

Released with OpsManager 3.4.0 on 2016-11-29

  • Adds support for automation of MongoDB 3.4 deployments.

  • Adds support for management of Monitoring/Backup Agents on PowerPC-based Linux systems for MongoDB 3.4 or later deployments only.

  • Built using Go 1.6.

  • Adds support for gathering of hardware metrics.

  • When importing a process that uses a password for the PEMKeyFile, no longer requires user to re-enter the PEMKeyFile password.

  • Fix: Can upgrade from MongoDB 2.4 to 2.6 while staying on authSchemaVersion 1.

  • Does not create Windows firewall rules for processes that are started on temporary ports where external access is not required.

  • Uses systemd management on RHEL7 and Ubuntu 16.04.


Released with Ops Manager 2.0.7 on 2016-11-03

  • MongoDB data and log files will have a umask of 027. Requires new package install.

Released with Ops Manager 2.0.6 on 2016-08-18

  • Improve logging on authentication failures.

  • Fix: Can set clusterAuthMode on sharded clusters.

Released with Ops Manager 2.0.5 on 2016-07-14

  • Substantial optimization in state-gathering.

  • Configurable timeout for connections to MongoDB processes.

  • Fix: Problem verifying success when creating text indexes in rolling index builds.

Released with Ops Manager 2.0.4 on 2016-05-20

  • Agent no longer downloads restore data for arbiters.

  • Fix: Some cases where CSRS conversion could get stuck.

  • Fix: Agent can restart a config server if all config servers are down.

  • Fix: validating MongoDB versions when a cluster was on mixed operating systems.

Released with Ops Manager 2.0.3 on 2016-03-24

  • Fix: Can import of arbiter using a different keyfile then existing configuration.

  • Allow specifying a temporary port for use during a CRSR upgrade.

Released with Ops Manager 2.0.2 on 2016-03-01

Released with Ops Manager 2.0.1 on 2016-01-21

  • Stability and performance improvements for restores via automation.

  • Added optimization to prioritize replica set reconfiguration actions over index builds.

  • Improved index building mechanism: index builds are no longer performed in a rolling fashion for 2-node replica sets, but instead are built in the background.

  • Added optimization to not compare unsupported index options when determining whether or not an index already exists.

  • Fix: Can import existing deployments that include arbiters running with authentication.

  • Fix: Rolling storage engine conversion for replica sets to ensure a super majority is always up.

  • Fix: Can create custom roles on sharded clusters running MongoDB 3.2 with config server replica sets.

Released with Ops Manager 2.0.0 on 2015-12-08

  • Added aupport for MongoDB 3.2.0 clusters with config servers as replica sets.

  • Added aupport for automated restores via the Automation Agent.

  • Added aupport for rolling index builds.

  • Added aupport for configuring WiredTiger encrypted storage for MongoDB 3.2.

  • Added aupport for rolling conversion to X-509 member authentication.

  • Improved handling of sharded clusters with members running on both Linux and Windows-based operating systems.

  • Added optimization when starting a new Monitoring or Backup Agent to ensure that the process is running before achieving Goal State.

  • Fix: glibc compatibility issue on RHEL5 and RHEL6.

  • Fix: Automation Agent automatic update failures could cause surge in configuration calls from the Automation Agent.

Released with Ops Manager 1.8.2 on 2015-10-20

  • Fix: Agent from not recognizing RHEL Workstations as RHEL.

Released with Ops Manager 1.8.1 on 2015-08-17

  • Fix: Can manage an existing deploy with user that has "root" privileges.

  • Fix: Storage engine conversions do not get stuck if replica set contained an arbiter.

  • Fix: Can update credentials after failed attempt to manage an existing deployment.

Released with Ops Manager 1.8 on 2015-06-23

  • Added support for managing SSL-enabled deployments.

  • Added support for managing deployment using Kerberos, LDAP, and x.509 Client Certificate authentication.

  • Added support to import an existing mongos with a config file.

  • Added support for importing an existing deployment that contains authenticated arbiters on which the hostname does not resolve locally to the loopback interface.

  • Added the ability to upgrade the authSchemaVersion when auth is not enabled.

  • Added support to change the storage engine for replica sets with more than one data node.

  • Enabled storage engine conversions for single-node replica sets and standalones.

  • Added more detailed logging of when MongoDB, the Monitoring Agent, or the Backup Agent rotate their logs.

  • Added support for distribution-specific MongoDB Community Edition builds.

  • Added up-front validation to ensure that MongoDB processes are running as the same user as the Automation Agent.

  • Added functionality to delete MongoDB binaries on disk that are not used by a managed process.

  • Added optimization where Ops Manager assumes success when starting a forked MongoDB process, rather than waiting for EOF.

  • Improved algorithm for balancing mongod processes across cores.

  • When deleting directories, symlinks are no longer deleted.

  • Fix: Can import credentials for MONGODB-CR users from SCRAM-SHA-1 deployments. See: MMS-2612 for more details.

  • Fix: Can derive the default port for config servers started with the --configsvr option but with no port specified. See: MMS-2489.

  • Fix: Can configure oplog sizes greater than 1TB.

  • Fix: Automation Agent does not interfere with manually-created replica set tags.

  • Ensured that the Automation Agent fails gracefully when an expected user does not exist during an initial import.

Released with Ops Manager 1.6.3 on 2015-06-23

  • Added support for importing an existing deployment that contains authenticated arbiters on which the hostname does not resolve locally to the loopback interface.

  • Fix: Logic used for performing a rolling restart.

  • Fix: with deriving the default port for config servers started with the --configsvr option but with no port specified. See MMS-2489.

Released 2015-04-28

  • Fix: Can update users created on MongoDB 2.4.

  • Fix: No longer have config server repair occur if the third config server was out of sync.

Released 2015-03-26

  • Fix: a rare edge-case that prevented the Automation Agent from successfully enabling authentication.

Released 2015-03-02

Initial release.

Back

MongoDB Agent

On this page