Release Notes
Announcements
Removing a Shard
Scenarios
When service traffic changes cause a decrease in instance data and load, you can reduce the shard quantity to release certain database resources and adjust the cluster scale to match the current load. This helps you adjust configurations as needed and avoid resource waste.
Note:
Removing shards will trigger data migration, which may exert a long-term impact on instance load. It is recommended to prioritize the evaluation of directly downgrading mongod node specifications to achieve cost reduction. For specific operations, see Adjusting Mongod Node Specification.
The shard removal feature is currently in public beta. If you need to use this feature, please submit a ticket to apply.
Impacts and Constraints
Before reducing the shard quantity, please carefully read the following key information to ensure that the operation meets expectations and does not impact service stability.
Dimension | Shard Removal |
Billing Impact | Yearly/monthly subscription: The remaining balance of the original specifications and the price of the new specifications will be calculated and the difference will be refunded. At the next renewal cycle, charges will be based on the new specifications. Pay-as-you-go: Charges for the original specifications will be settled, and billing restarts based on the new specifications. For the detailed billing information, see Configuration Adjustment Billing. |
Version Requirement | General Edition: TencentDB for MongoDB 4.0 and later. Cloud Disk Edition: Shard removal is not supported. |
Service Impact | During the operation, no service interruption will occur. Database read and write operations are normal and both automatic and manual backups are supported. During the operation, you are strictly prohibited from performing any operations that may interfere with data migration, including but not limited to: Data rollback and instance termination Configuration adjustment and primary-secondary switch Connection address modification Operations on the shards to be removed |
Limits and Constraints | Removal order: To-be-removed shards are selected by creation time in descending order. Space requirement: After removal, the total storage space of the remaining shards must be greater than or equal to 120% of the used space of the cluster. Otherwise, removal is not allowed. Removal quantity: The number of shards to be removed each time cannot exceed half of the original shard quantity of the cluster. After removal, the number of shards must be greater than or equal to 2. Note: Due to kernel limitations, you can only remove one shard at a time in TencentDB for MongoDB 4.0. Balancer requirement: Before removal, enable the balancer and set the balancing window to control the impact of data migration on services. Time consumption: Shard removal will trigger data redistribution. This process may be slow, depending on the data volume, balancer settings, and other factors. |
Pre-Operation Checklist
No. | Check Item | Check Criteria | Specific Instruction and Operational Guideline |
1 | Instance status | The instance status is Running. | On the Instance List page, check that the target instance is running normally. |
2 | Shard quantity | Before removal, the shard quantity must be greater than 2. | This is a prerequisite. If the current number of shards equals 2, removal cannot be performed. Removing shards would result in fewer than 2 shards, which does not meet cluster architecture requirements. |
3 | Instance expiration time | The expiration time of a yearly/monthly subscription instance must be greater than or equal to 14 days. | In the Configuration Information section on the Instance Details page, view Expiration Time to ensure that the instance has sufficient remaining service time. |
4 | Storage space after removal | Total storage space of the remaining shards > Current used space of the cluster x 120% | 1. In the Specifications section on the Instance Details page, view the total storage space and used space. 2. Calculate the total storage space of the remaining shards, as shown in the example below. Current total space: 150 GB (3 shards x 50 GB) Shards to be removed: 1 shard Space after removal: 100 GB Currently used: 1.5 GB Minimum required space: 1.5 GB x 120% = 1.8 GB Verification: Because 100 GB is greater than 1.8 GB, the check passed. |
5 | Balancer settings | Ensure that the balancer is enabled. It is recommended to set the balancing window. | 1. On the Parameter Settings page in the console, set the openBalance.window parameter to true to enable the balancer. 2. Set off-peak hours as the balancing window by configuring the balance.window parameter to minimize the impact of data migration on services. For specific operations, see Database Parameter Adjustment. |
6 | Primary shard check | Confirm that the shard to be removed is not the primary shard of any unsharded collection. | 1. Connect to the instance using the MongoDB Shell, query the config database by executing getSiblingDB("config").databases.find(), and identify the databases where a to-be-removed shard serves as the primary shard. Note: The shard name format is cmgo-xxxxxxxx_n. xxxxxxxx is the unique instance identifier (8 characters) and n is the shard sequence number starting from 0. For example, a 5-shard instance would have shard names ranging from cmgo-xxxxxxxx_0 to cmgo-xxxxxxxx_4. If you need to remove the last shard, you should specify cmgo-xxxxxxxx_4.
2. For each identified database, use the getCollectionNames() method to obtain a complete list of all collections.
3. Check the sharding status of each collection by executing db.collection.stats().sharded. If the return value is false (indicating unsharded), you must execute the movePrimary command to move the primary shard of the database containing this collection to another shard. For details, see movePrimary (database command). Note: In TencentDB for MongoDB 4.0 and 4.2, after movePrimary is executed, it is necessary to refresh the routing information. Otherwise, read and write operations may be blocked. The refresh method is as follows: Version 4.0: After executing movePrimary, restart all mongos nodes or execute the flushRouterConfig command on all mongos nodes. Version 4.2: After executing movePrimary, restart all mongos and mongod nodes or execute the flushRouterConfig command on all mongos and mongod nodes.
See Appendix: Example of Verifying That the Shard to Be Removed Is Not the Primary Shard of Any Unsharded Collection for operational examples. |
Removing a Shard
1. Log in to the TencentDB for MongoDB console.
2. In the MongoDB drop-down list on the left sidebar, select Shard Instance.
3. Above the Instance List page on the right, select the target region.
4. In the instance list, find the target instance.
5. Click Instance ID to go to the Instance Details page and click the Node Management tab.
6. In the Node Operation drop-down list, select Delete Shard.
7. In the Delete Shard window, learn about the restrictions and conditions for removing shards and refer to the table below to specify the number of shards to be removed.
Parameter Category | Parameter | Example | Description |
Basic Instance Information | Instance Name | test-4dot2-XXX | Name of the instance for which the shard quantity is to be reduced. |
| Expiration Time | 2025-04-24 19:23:43 | Service expiration time of a yearly/monthly subscription instance. |
| Instance Architecture | A sharded cluster instance that has 3 shards, each with 4 nodes, among which there is 1 read-only node | Cluster architecture of the instance: sharded cluster instance. Shard quantity: shard quantity before removal. Nodes per shard: number of nodes per shard and number of read-only nodes. |
| Node Specification | 2 cores, 4 GB memory, 250 GB storage, and 10 nodes in total | Single shard node specification: including CPU cores, memory, and storage capacity. Total number of cluster nodes: total number of nodes in the instance before removal (Number of shards x Nodes per shard = Total number of cluster nodes). |
Configuration Downgrade | Number of Shards to Be Deleted | 2 | Specify the number of shards to be removed for the instance. |
| | | After specifying the number of shards to be removed, click Check Shards to Be Removed to check whether the conditions for removing shards are met. Check the number of remaining shards: Ensure that the number of remaining shards is not less than 2 after removal. Check the storage space: Ensure that the total storage space of the remaining shards after removal is greater than 120% of the currently used space in the cluster. Check the primary shard information: Check whether the shards to be removed contain any primary shards of databases. If yes, you must manually move them to other shards before proceeding. |
Cost Information | Fees | 0.92 USD/hour | Pay-as-you-go: The hourly billing rate after shard removal for the instance will be displayed. Yearly/monthly subscription: The refund amount after the specifications are downgraded will be displayed. |
8. After confirming the details, click OK. If the instance uses yearly/monthly subscription billing, you must pay the price difference between the original and new specifications on the Confirm Product Information page to successfully complete the configuration adjustment.
Note:
If the data volume on the shard node is too large, the balancing window is too small, or there is a jumbo chunk, it will directly cause the shard removal task to take a long time, or even the task will be blocked and fail. For specific troubleshooting and resolution methods, see Shard Removal Task: Guide for Confirming the Progress and Troubleshooting Issues.
9. During task execution, if you need to stop the removal, click Terminate on the Task Management page. After termination, the shard quantity will revert to the original level.
Related APIs
API | Description |
Adjusts the cloud database instance configurations | |
Queries the sales specifications of cloud databases |
Appendix: Example of Verifying That the Shard to Be Removed Is Not the Primary Shard of Any Unsharded Collection
1. Query the config database by executing
getSiblingDB("config").databases.find(), and identify the databases whose primary shard is the shard to be removed. The following example shows how to query the databases where shard cmgo-xxxxxxxx_3 serves as the primary shard.mongos> db.getSiblingDB("config").databases.find({primary: "cmgo-xxxxxxxx_3"}){ "_id" : "bulk_db_db_3", "primary" : "cmgo-xxxxxxxx_3", "partitioned" : false, "version" : { "uuid" : UUID("xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxx"), "timestamp" : Timestamp(1762855003, 5), "lastMod" : 1 } }{ "_id" : "bulk_db_db_4", "primary" : "cmgo-xxxxxxxx_3", "partitioned" : false, "version" : { "uuid" : UUID("xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"), "timestamp" : Timestamp(1762855035, 2), "lastMod" : 1 } }{ "_id" : "bulk_db_db_10", "primary" : "cmgo-xxxxxxxx_3", "partitioned" : false, "version" : { "uuid" : UUID("xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"), "timestamp" : Timestamp(1762855333, 3), "lastMod" : 1 } }{ "_id" : "bulk_db_db_14", "primary" : "cmgo-xxxxxxxx_3", "partitioned" : false, "version" : { "uuid" : UUID("xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"), "timestamp" : Timestamp(1762855462, 30), "lastMod" : 1 } }
2. For each identified database, use the getCollectionNames() method to obtain a complete list of all collections. The following example shows how to query all collections in the database named bulk_db_db_77.
mongos> db.getSiblingDB("bulk_db_db_77").getCollectionNames()["collection_1","collection_10","collection_11","collection_12","collection_13","collection_14","collection_15","collection_16","collection_17","collection_18","collection_19","collection_2","collection_20","collection_3","collection_4","collection_5","collection_6","collection_7","collection_8","collection_9"]
3. Check the sharding status of each collection by executing db.collection.stats().sharded. The following example shows how to query the sharding status of the bulk_db_db_77.collection1 collection. If "sharded" : false is returned, you must use the movePrimary command to move the primary shard of the database containing this collection to another shard.
mongos> db.getSiblingDB("bulk_db_db_77").collection1.stats(){"sharded" : false,"primary" : "cmgo-xxxxxxxx_3","ns" : "bulk_db_db_77.collection1","count" : 0,"size" : 0,"storageSize" : 0,"totalIndexSize" : 0,"totalSize" : 0,"indexSizes" : {},"avgObjSize" : 0,"maxSize" : NumberLong(0),"nindexes" : 0,"scaleFactor" : 1,"nchunks" : 1,"shards" : {"cmgo-xxxxxxxx_3" : {"ns" : "bulk_db_db_77.collection1","size" : 0,"count" : 0,"numOrphanDocs" : 0,"storageSize" : 0,"totalSize" : 0,"nindexes" : 0,"totalIndexSize" : 0,"indexDetails" : {},"indexSizes" : {},"scaleFactor" : 1,"ok" : 1,"$clusterTime" : {"clusterTime" : Timestamp(1763094536, 1),"signature" : {"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),"keyId" : NumberLong(0)}},"$configTime" : Timestamp(1763094535, 2),"$topologyTime" : Timestamp(1762853939, 5),"operationTime" : Timestamp(1763094536, 1)}},"ok" : 1,"$clusterTime" : {"clusterTime" : Timestamp(1763094536, 1),"signature" : {"hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),"keyId" : NumberLong(0)}},"operationTime" : Timestamp(1763094536, 1)}mongos> db.getSiblingDB("bulk_db_db_77").collection1.stats().shardedfalse
Help and Support
Was this page helpful?
You can also Contact sales or Submit a Ticket for help.
Help us improve! Rate your documentation experience in 5 mins.
Feedback