Docs Home → MongoDB Cloud Manager
Edit a Deployment's Configuration
On this page
You can modify a deployment's configuration and topology, including its MongoDB versions, storage engines, and numbers of hosts or shards. You can make modifications at all levels of a deployment's topology from a top-level sharded cluster or replica set to lower levels, such as a replica set within a sharded cluster, or an individual process within a replica set. You can also modify standalone processes.
Considerations
Apply Changes to Cluster or Member
If you make configuration changes to an individual MongoDB process within a cluster, any future changes to the cluster no longer apply to the child process.
Example
If you turn off journaling for a replica set member and then later change the journal commit interval for the replica set, the change does not apply to the member.
MongoDB Version
To choose which versions of MongoDB are available to Cloud Manager, see Add a Custom MongoDB Build.
Check the following documents for any considerations or compatibility issues before changing a deployment's MongoDB version:
The documentation for your driver.
Plan the version change during a predefined maintenance window.
Change the MongoDB version on a staging environment before changing a production environment. Your staging environment should mirror your production environment. This can help avoid compatibility issues that may result in downtime for your production deployment.
Follow the MongoDB release notes when performing manual upgrades of replica sets and sharded clusters.
Note
Downgrading Limitations
You cannot downgrade a MongoDB deployment:
From version 5.0 to any version before 4.4.0
From version 4.4 to any version before 4.2.6
From version 4.2 to any version 4.0.12 (for Windows) or 4.0.7 (for Linux)
From version 4.0 to any version before 3.6.23
Backup Considerations for MongoDB Versions
To learn more about backup considerations, see Backup Considerations.
If you choose to upgrade to MongoDB 4.2 with
"featureCompatibilityVersion" : 4.2
, Cloud Manager displays a modal that
asks you to agree to the special license that MongoDB, Inc.
grants to use MongoDB Enterprise for backups.
Storage Engine
Important
MongoDB removed support for the MMAPv1 storage engine in MongoDB 4.2. If you edit your deployment's configuration to change your storage engine to WiredTiger Storage Engine, Cloud Manager restarts the MongoDB processes.
If you run or upgrade to MongoDB 3.0 or later and modify the MongoDB storage engine, Cloud Manager shuts down and restarts the MongoDB process. For a multi-member replica set, Cloud Manager performs a rolling inital sync of each member.
Cloud Manager creates backup directories during the migration from one storage engine to the other if the host has adequate disk space. If disk space is insufficient, no backups are taken. Cloud Manager does not delete the backup directories once the migration is complete. You can keep or delete the previous backup directories. The backup directories are located in the mongod's data directory.
Example
If the data directory was /data/process
, the backup would be
/data/process.bak.UNIQUENAME
. The UNIQUENAME
is a random
string that Cloud Manager generates.
Before you can change the storage engine for a standalone instance or replica set, you must give the Automation write access to the MongoDB data directory's parent directory. The agent creates a temporary backup of the data in the parent directory when updating the storage engine. Storage engine changes on standalone instances also require adequate disk space to perform a full /mongodump and /mongorestore. This disk space is then restored to the instance after the storage engine configuration change. Cloud Manager does not delete the backup directories.
You cannot change the storage engine on a config server. For more information on storage engines and the available options, see Storage in the MongoDB manual.
Fixed Properties
You cannot modify the following settings after a deployment has been created:
You can modify the following deployment settings:
log path
at the process level
Deployment Topology
You can make modifications at all levels of a deployment's topology, including child processes.
To modify the topology or processes, use this tutorial or one of the more specific tutorials:
Project-Level Modifications
Some modifications that affect a deployment occur at the project level. The following changes affect every MongoDB process in the project. For these changes, use the specified tutorials:
To enable TLS for the deployment, see Enable TLS for a Deployment.
To enable authentication for the deployment, see Enable Authentication for a Cloud Manager Project.
To add or modify MongoDB users and roles for the deployment, see Manage MongoDB Users.
Multiple Modifications
You can combine multiple modifications into one deployment.
Example
You could make all the following modifications before clicking the Review Changes button:
Add the latest stable version of MongoDB to the Add a Custom Build.
Enable TLS for the deployment's MongoDB processes.
Add a new sharded cluster running the latest stable version of MongoDB from above.
When you click Review Changes, the review displays all the changes on one screen for you to confirm before deploying.
Force Reconfigure
For Replica Sets and Sharded Clusters Only
The MongoDB Agent can force a replica set to accept a new configuration
when you set the Force Reconfigure Replication Setting to
Yes
. Only force a reconfiguration to recover a replica set from a
state in which a minority of its members are available.
Warning
Forcing a replica set reconfiguration might lead to a rollback of majority-committed writes.
Proceed with caution. Contact MongoDB Support if you have questions about the potential impacts of this operation.
Tip
See also:
Reconfigure a Replica Set with Unavailable Members in the MongoDB Manual.
Removing a Shard
For Sharded Clusters Only
When you remove a shard, any unsharded databases in that shard are moved to a remaining shard using the movePrimary command.
All sharded collections remain online and available during the shard
removal process. However, read and write operations sent to unsharded
collections during the movePrimary
operation can result in
unexpected behavior, including failure of the migration or loss of
data.
We recommend moving the primary shard for any databases containing unsharded collections before removing the shard.
To learn more about removing shards, see Remove Shards from an Existing Sharded Cluster.
Removing Multiple Replica Set Members
You can remove or migrate multiple replica set members at once, but a majority of the voting members must remain. If you need to remove more voting members, remove them one at a time.
Example
Example 1
You have a four-node replica set. All nodes are voting members. You can remove only one node, which preserves the majority of three out of four voting nodes. You can remove another node from the remaining three-node replica set afterward. This preserves the majority of the remaining voting nodes.
Example
Example 2
You have a four-node replica set. Three nodes are voting members and one node is a non-voting member. You can remove one voting member and one non-voting member at the same time. This preserves the majority of two out of three voting nodes.
To learn more about voting, see Replica Set High Availability and Replica Set Elections.
Prerequisites
Your deployment must be running a version of the Automation that is compatible with Cloud Manager. If your deployment is not running a compatible version of the agent, Cloud Manager displays a banner prompting you to update your agents.
Procedure
Select the type of deployment you want to edit: