4.4. Shard Management

4.4.1. Introduction

This document discusses how sharding works in CouchDB along with how to safely add, move, remove, and create placement rules for shards and shard replicas.

A shard) is a horizontal partition of data in a database. Partitioning data into shards and distributing copies of each shard (called “shard replicas” or just “replicas”) to different nodes in a cluster gives the data greater durability against node loss. CouchDB clusters automatically shard databases and distribute the subsets of documents that compose each shard among nodes. Modifying cluster membership and sharding behavior must be done manually.

4.4.1.1. Shards and Replicas

How many shards and replicas each database has can be set at the global level, or on a per-database basis. The relevant parameters are q and n.

q is the number of database shards to maintain. n is the number of copies of each document to distribute. The default value for n is 3, and for q is 2. With q=2, the database is split into 2 shards. With n=3, the cluster distributes three replicas of each shard. Altogether, that’s 6 shard replicas for a single database.

In a 3-node cluster with q=8, each node would receive 8 shards. In a 4-node cluster, each node would receive 6 shards. We recommend in the general case that the number of nodes in your cluster should be a multiple of n, so that shards are distributed evenly.

CouchDB nodes have a etc/default.ini file with a section named cluster which looks like this:

  1. [cluster]
  2. q=2
  3. n=3

These settings specify the default sharding parameters for newly created databases. These can be overridden in the etc/local.ini file by copying the text above, and replacing the values with your new defaults. The values can also be set on a per-database basis by specifying the q and n query parameters when the database is created. For example:

  1. $ curl -X PUT "$COUCH_URL:5984/database-name?q=4&n=2"

This creates a database that is split into 4 shards and 2 replicas, yielding 8 shard replicas distributed throughout the cluster.

4.4.1.2. Quorum

Depending on the size of the cluster, the number of shards per database, and the number of shard replicas, not every node may have access to every shard, but every node knows where all the replicas of each shard can be found through CouchDB’s internal shard map.

Each request that comes in to a CouchDB cluster is handled by any one random coordinating node. This coordinating node proxies the request to the other nodes that have the relevant data, which may or may not include itself. The coordinating node sends a response to the client once a quorum) of database nodes have responded; 2, by default. The default required size of a quorum is equal to r=w=((n+1)/2) where r refers to the size of a read quorum, w refers to the size of a write quorum, and n refers to the number of replicas of each shard. In a default cluster where n is 3, ((n+1)/2) would be 2.

The size of the required quorum can be configured at request time by setting the r parameter for document and view reads, and the w parameter for document writes. For example, here is a request that directs the coordinating node to send a response once at least two nodes have responded:

  1. $ curl "$COUCH_URL:5984/{db}/{doc}?r=2"

Here is a similar example for writing a document:

  1. $ curl -X PUT "$COUCH_URL:5984/{db}/{doc}?w=2" -d '{...}'

Setting r or w to be equal to n (the number of replicas) means you will only receive a response once all nodes with relevant shards have responded or timed out, and as such this approach does not guarantee ACIDic consistency. Setting r or w to 1 means you will receive a response after only one relevant node has responded.

4.4.2. Examining database shards

There are a few API endpoints that help you understand how a database is sharded. Let’s start by making a new database on a cluster, and putting a couple of documents into it:

  1. $ curl -X PUT $COUCH_URL:5984/mydb
  2. {"ok":true}
  3. $ curl -X PUT $COUCH_URL:5984/mydb/joan -d '{"loves":"cats"}'
  4. {"ok":true,"id":"joan","rev":"1-cc240d66a894a7ee7ad3160e69f9051f"}
  5. $ curl -X PUT $COUCH_URL:5984/mydb/robert -d '{"loves":"dogs"}'
  6. {"ok":true,"id":"robert","rev":"1-4032b428c7574a85bc04f1f271be446e"}

First, the top level /db endpoint will tell you what the sharding parameters are for your database:

  1. $ curl -s $COUCH_URL:5984/db | jq .
  2. {
  3.   "db_name": "mydb",
  4. ...
  5.   "cluster": {
  6.     "q": 8,
  7.     "n": 3,
  8.     "w": 2,
  9.     "r": 2
  10.   },
  11. ...
  12. }

So we know this database was created with 8 shards (q=8), and each shard has 3 replicas (n=3) for a total of 24 shard replicas across the nodes in the cluster.

Now, let’s see how those shard replicas are placed on the cluster with the /db/_shards endpoint:

  1. $ curl -s $COUCH_URL:5984/mydb/_shards | jq .
  2. {
  3.   "shards": {
  4.     "00000000-1fffffff": [
  5.       "node1@127.0.0.1",
  6.       "node2@127.0.0.1",
  7.       "node4@127.0.0.1"
  8.     ],
  9.     "20000000-3fffffff": [
  10.       "node1@127.0.0.1",
  11.       "node2@127.0.0.1",
  12.       "node3@127.0.0.1"
  13.     ],
  14.     "40000000-5fffffff": [
  15.       "node2@127.0.0.1",
  16.       "node3@127.0.0.1",
  17.       "node4@127.0.0.1"
  18.     ],
  19.     "60000000-7fffffff": [
  20.       "node1@127.0.0.1",
  21.       "node3@127.0.0.1",
  22.       "node4@127.0.0.1"
  23.     ],
  24.     "80000000-9fffffff": [
  25.       "node1@127.0.0.1",
  26.       "node2@127.0.0.1",
  27.       "node4@127.0.0.1"
  28.     ],
  29.     "a0000000-bfffffff": [
  30.       "node1@127.0.0.1",
  31.       "node2@127.0.0.1",
  32.       "node3@127.0.0.1"
  33.     ],
  34.     "c0000000-dfffffff": [
  35.       "node2@127.0.0.1",
  36.       "node3@127.0.0.1",
  37.       "node4@127.0.0.1"
  38.     ],
  39.     "e0000000-ffffffff": [
  40.       "node1@127.0.0.1",
  41.       "node3@127.0.0.1",
  42.       "node4@127.0.0.1"
  43.     ]
  44.   }
  45. }

Now we see that there are actually 4 nodes in this cluster, and CouchDB has spread those 24 shard replicas evenly across all 4 nodes.

We can also see exactly which shard contains a given document with the /db/_shards/doc endpoint:

  1. $ curl -s $COUCH_URL:5984/mydb/_shards/joan | jq .
  2. {
  3.   "range": "e0000000-ffffffff",
  4.   "nodes": [
  5.     "node1@127.0.0.1",
  6.     "node3@127.0.0.1",
  7.     "node4@127.0.0.1"
  8.   ]
  9. }
  10. $ curl -s $COUCH_URL:5984/mydb/_shards/robert | jq .
  11. {
  12.   "range": "60000000-7fffffff",
  13.   "nodes": [
  14.     "node1@127.0.0.1",
  15.     "node3@127.0.0.1",
  16.     "node4@127.0.0.1"
  17.   ]
  18. }

CouchDB shows us the specific shard into which each of the two sample documents is mapped.

4.4.3. Moving a shard

When moving shards or performing other shard manipulations on the cluster, it is advisable to stop all resharding jobs on the cluster. See Stopping Resharding Jobs for more details.

This section describes how to manually place and replace shards. These activities are critical steps when you determine your cluster is too big or too small, and want to resize it successfully, or you have noticed from server metrics that database/shard layout is non-optimal and you have some “hot spots” that need resolving.

Consider a three-node cluster with q=8 and n=3. Each database has 24 shards, distributed across the three nodes. If you add a fourth node to the cluster, CouchDB will not redistribute existing database shards to it. This leads to unbalanced load, as the new node will only host shards for databases created after it joined the cluster. To balance the distribution of shards from existing databases, they must be moved manually.

Moving shards between nodes in a cluster involves the following steps:

  1. Ensure the target node has joined the cluster.
  2. Copy the shard(s) and any secondary index shard(s) onto the target node.
  3. Set the target node to maintenance mode.
  4. Update cluster metadata to reflect the new target shard(s).
  5. Monitor internal replication to ensure up-to-date shard(s).
  6. Clear the target node’s maintenance mode.
  7. Update cluster metadata again to remove the source shard(s)
  8. Remove the shard file(s) and secondary index file(s) from the source node.

4.4.3.1. Copying shard files

Shard files live in the data/shards directory of your CouchDB install. Within those subdirectories are the shard files themselves. For instance, for a q=8 database called abc, here is its database shard files:

  1. data/shards/00000000-1fffffff/abc.1529362187.couch
  2. data/shards/20000000-3fffffff/abc.1529362187.couch
  3. data/shards/40000000-5fffffff/abc.1529362187.couch
  4. data/shards/60000000-7fffffff/abc.1529362187.couch
  5. data/shards/80000000-9fffffff/abc.1529362187.couch
  6. data/shards/a0000000-bfffffff/abc.1529362187.couch
  7. data/shards/c0000000-dfffffff/abc.1529362187.couch
  8. data/shards/e0000000-ffffffff/abc.1529362187.couch

Secondary indexes (including JavaScript views, Erlang views and Mango indexes) are also sharded, and their shards should be moved to save the new node the effort of rebuilding the view. View shards live in data/.shards. For example:

  1. data/.shards
  2. data/.shards/e0000000-ffffffff/_replicator.1518451591_design
  3. data/.shards/e0000000-ffffffff/_replicator.1518451591_design/mrview
  4. data/.shards/e0000000-ffffffff/_replicator.1518451591_design/mrview/3e823c2a4383ac0c18d4e574135a5b08.view
  5. data/.shards/c0000000-dfffffff
  6. data/.shards/c0000000-dfffffff/_replicator.1518451591_design
  7. data/.shards/c0000000-dfffffff/_replicator.1518451591_design/mrview
  8. data/.shards/c0000000-dfffffff/_replicator.1518451591_design/mrview/3e823c2a4383ac0c18d4e574135a5b08.view
  9. ...

Since they are files, you can use cp, rsync, scp or other file-copying command to copy them from one node to another. For example:

  1. # one one machine
  2. $ mkdir -p data/.shards/{range}
  3. $ mkdir -p data/shards/{range}
  4. # on the other
  5. $ scp {couch-dir}/data/.shards/{range}/{database}.{datecode}* \
  6.   {node}:{couch-dir}/data/.shards/{range}/
  7. $ scp {couch-dir}/data/shards/{range}/{database}.{datecode}.couch \
  8.   {node}:{couch-dir}/data/shards/{range}/

4.4.3.2. Set the target node to true maintenance mode

Before telling CouchDB about these new shards on the node, the node must be put into maintenance mode. Maintenance mode instructs CouchDB to return a 404 Not Found response on the /_up endpoint, and ensures it does not participate in normal interactive clustered requests for its shards. A properly configured load balancer that uses GET /_up to check the health of nodes will detect this 404 and remove the node from circulation, preventing requests from being sent to that node. For example, to configure HAProxy to use the /_up endpoint, use:

  1. http-check disable-on-404
  2. option httpchk GET /_up

If you do not set maintenance mode, or the load balancer ignores this maintenance mode status, after the next step is performed the cluster may return incorrect responses when consulting the node in question. You don’t want this! In the next steps, we will ensure that this shard is up-to-date before allowing it to participate in end-user requests.

To enable maintenance mode:

  1. $ curl -X PUT -H "Content-type: application/json" \
  2.     $COUCH_URL:5984/_node/{node-name}/_config/couchdb/maintenance_mode \
  3.     -d "\"true\""

Then, verify that the node is in maintenance mode by performing a GET /_up on that node’s individual endpoint:

  1. $ curl -v $COUCH_URL/_up
  2. …
  3. < HTTP/1.1 404 Object Not Found
  4. …
  5. {"status":"maintenance_mode"}

Finally, check that your load balancer has removed the node from the pool of available backend nodes.

4.4.3.3. Updating cluster metadata to reflect the new target shard(s)

Now we need to tell CouchDB that the target node (which must already be joined to the cluster) should be hosting shard replicas for a given database.

To update the cluster metadata, use the special /_dbs database, which is an internal CouchDB database that maps databases to shards and nodes. This database is automatically replicated between nodes. It is accessible only through the special /_node/_local/_dbs endpoint.

First, retrieve the database’s current metadata:

  1. $ curl http://localhost/_node/_local/_dbs/{name}
  2. {
  3.   "_id": "{name}",
  4.   "_rev": "1-e13fb7e79af3b3107ed62925058bfa3a",
  5.   "shard_suffix": [46, 49, 53, 51, 48, 50, 51, 50, 53, 50, 54],
  6.   "changelog": [
  7.     ["add", "00000000-1fffffff", "node1@xxx.xxx.xxx.xxx"],
  8.     ["add", "00000000-1fffffff", "node2@xxx.xxx.xxx.xxx"],
  9.     ["add", "00000000-1fffffff", "node3@xxx.xxx.xxx.xxx"],
  10.     …
  11.   ],
  12.   "by_node": {
  13.     "node1@xxx.xxx.xxx.xxx": [
  14.       "00000000-1fffffff",
  15.       …
  16.     ],
  17.     …
  18.   },
  19.   "by_range": {
  20.     "00000000-1fffffff": [
  21.       "node1@xxx.xxx.xxx.xxx",
  22.       "node2@xxx.xxx.xxx.xxx",
  23.       "node3@xxx.xxx.xxx.xxx"
  24.     ],
  25.     …
  26.   }
  27. }

Here is a brief anatomy of that document:

  • _id: The name of the database.
  • _rev: The current revision of the metadata.
  • shard_suffix: A timestamp of the database’s creation, marked as seconds after the Unix epoch mapped to the codepoints for ASCII numerals.
  • changelog: History of the database’s shards.
  • by_node: List of shards on each node.
  • by_range: On which nodes each shard is.

To reflect the shard move in the metadata, there are three steps:

  1. Add appropriate changelog entries.
  2. Update the by_node entries.
  3. Update the by_range entries.

As of this writing, this process must be done manually.

To add a shard to a node, add entries like this to the database metadata’s changelog attribute:

  1. ["add", "{range}", "{node-name}"]

The {range} is the specific shard range for the shard. The {node-name} should match the name and address of the node as displayed in GET /_membership on the cluster.

Once you have figured out the new changelog entries, you will need to update the by_node and by_range to reflect who is storing what shards. The data in the changelog entries and these attributes must match. If they do not, the database may become corrupted.

Continuing our example, here is an updated version of the metadata above that adds shards to an additional node called node4:

  1. {
  2.   "_id": "{name}",
  3.   "_rev": "1-e13fb7e79af3b3107ed62925058bfa3a",
  4.   "shard_suffix": [46, 49, 53, 51, 48, 50, 51, 50, 53, 50, 54],
  5.   "changelog": [
  6.     ["add", "00000000-1fffffff", "node1@xxx.xxx.xxx.xxx"],
  7.     ["add", "00000000-1fffffff", "node2@xxx.xxx.xxx.xxx"],
  8.     ["add", "00000000-1fffffff", "node3@xxx.xxx.xxx.xxx"],
  9.     ...
  10.     ["add", "00000000-1fffffff", "node4@xxx.xxx.xxx.xxx"]
  11.   ],
  12.   "by_node": {
  13.     "node1@xxx.xxx.xxx.xxx": [
  14.       "00000000-1fffffff",
  15.       ...
  16.     ],
  17.     ...
  18.     "node4@xxx.xxx.xxx.xxx": [
  19.       "00000000-1fffffff"
  20.     ]
  21.   },
  22.   "by_range": {
  23.     "00000000-1fffffff": [
  24.       "node1@xxx.xxx.xxx.xxx",
  25.       "node2@xxx.xxx.xxx.xxx",
  26.       "node3@xxx.xxx.xxx.xxx",
  27.       "node4@xxx.xxx.xxx.xxx"
  28.     ],
  29.     ...
  30.   }
  31. }

Now you can PUT this new metadata:

  1. $ curl -X PUT http://localhost/_node/_local/_dbs/{name} -d '{...}'

4.4.3.4. Forcing synchronization of the shard(s)

New in version 2.4.0.

Whether you pre-copied shards to your new node or not, you can force CouchDB to synchronize all replicas of all shards in a database with the /db/_sync_shards endpoint:

  1. $ curl -X POST $COUCH_URL:5984/{db}/_sync_shards
  2. {"ok":true}

This starts the synchronization process. Note that this will put additional load onto your cluster, which may affect performance.

It is also possible to force synchronization on a per-shard basis by writing to a document that is stored within that shard.

4.4.3.5. Monitor internal replication to ensure up-to-date shard(s)

After you complete the previous step, CouchDB will have started synchronizing the shards. You can observe this happening by monitoring the /_node/{node-name}/_system endpoint, which includes the internal_replication_jobs metric.

Once this metric has returned to the baseline from before you started the shard sync, or is 0, the shard replica is ready to serve data and we can bring the node out of maintenance mode.

4.4.3.6. Clear the target node’s maintenance mode

You can now let the node start servicing data requests by putting "false" to the maintenance mode configuration endpoint, just as in step 2.

Verify that the node is not in maintenance mode by performing a GET /_up on that node’s individual endpoint.

Finally, check that your load balancer has returned the node to the pool of available backend nodes.

4.4.3.7. Update cluster metadata again to remove the source shard

Now, remove the source shard from the shard map the same way that you added the new target shard to the shard map in step 2. Be sure to add the ["remove", {range}, {source-shard}] entry to the end of the changelog as well as modifying both the by_node and by_range sections of the database metadata document.

4.4.3.8. Remove the shard and secondary index files from the source node

Finally, you can remove the source shard replica by deleting its file from the command line on the source host, along with any view shard replicas:

  1. $ rm {couch-dir}/data/shards/{range}/{db}.{datecode}.couch
  2. $ rm -r {couch-dir}/data/.shards/{range}/{db}.{datecode}*

Congratulations! You have moved a database shard replica. By adding and removing database shard replicas in this way, you can change the cluster’s shard layout, also known as a shard map.

4.4.4. Specifying database placement

You can configure CouchDB to put shard replicas on certain nodes at database creation time using placement rules.

First, each node must be labeled with a zone attribute. This defines which zone each node is in. You do this by editing the node’s document in the special /_nodes database, which is accessed through the special node-local API endpoint at /_node/_local/_nodes/{node-name}. Add a key value pair of the form:

  1. "zone": "{zone-name}"

Do this for all of the nodes in your cluster. For example:

  1. $ curl -X PUT http://localhost/_node/_local/_nodes/{node-name} \
  2.     -d '{ \
  3.         "_id": "{node-name}",
  4.         "_rev": "{rev}",
  5.         "zone": "{zone-name}"
  6.         }'

In the local config file (local.ini) of each node, define a consistent cluster-wide setting like:

  1. [cluster]
  2. placement = {zone-name-1}:2,{zone-name-2}:1

In this example, CouchDB will ensure that two replicas for a shard will be hosted on nodes with the zone attribute set to {zone-name-1} and one replica will be hosted on a new with the zone attribute set to {zone-name-2}.

This approach is flexible, since you can also specify zones on a per- database basis by specifying the placement setting as a query parameter when the database is created, using the same syntax as the ini file:

  1. curl -X PUT $COUCH_URL:5984/{db}?zone={zone}

The placement argument may also be specified. Note that this will override the logic that determines the number of created replicas!

Note that you can also use this system to ensure certain nodes in the cluster do not host any replicas for newly created databases, by giving them a zone attribute that does not appear in the [cluster] placement string.

4.4.5. Splitting Shards

The /_reshard is an HTTP API for shard manipulation. Currently it only supports shard splitting. To perform shard merging, refer to the manual process outlined in the Merging Shards section.

The main way to interact with /_reshard is to create resharding jobs, monitor those jobs, wait until they complete, remove them, post new jobs, and so on. What follows are a few steps one might take to use this API to split shards.

At first, it’s a good idea to call GET /_reshard to see a summary of resharding on the cluster.

  1. $ curl -s $COUCH_URL:5984/_reshard | jq .
  2. {
  3.   "state": "running",
  4.   "state_reason": null,
  5.   "completed": 3,
  6.   "failed": 0,
  7.   "running": 0,
  8.   "stopped": 0,
  9.   "total": 3
  10. }

Two important things to pay attention to are the total number of jobs and the state.

The state field indicates the state of resharding on the cluster. Normally it would be running, however, another user could have disabled resharding temporarily. Then, the state would be stopped and hopefully, there would be a reason or a comment in the value of the state_reason field. See Stopping Resharding Jobs for more details.

The total number of jobs is important to keep an eye on because there is a maximum number of resharding jobs per node, and creating new jobs after the limit has been reached will result in an error. Before staring new jobs it’s a good idea to remove already completed jobs. See reshard configuration section for the default value of max_jobs parameter and how to adjust if needed.

For example, if the jobs have completed, to remove all the jobs run:

  1. $ curl -s $COUCH_URL:5984/_reshard/jobs | jq -r '.jobs[].id' |\
  2.   while read -r jobid; do\
  3.       curl -s -XDELETE $COUCH_URL:5984/_reshard/jobs/$jobid\
  4.   done

Then it’s a good idea to see what the db shard map looks like.

  1. $ curl -s $COUCH_URL:5984/db1/_shards | jq '.'
  2. {
  3.   "shards": {
  4.     "00000000-7fffffff": [
  5.       "node1@127.0.0.1",
  6.       "node2@127.0.0.1",
  7.       "node3@127.0.0.1"
  8.     ],
  9.     "80000000-ffffffff": [
  10.       "node1@127.0.0.1",
  11.       "node2@127.0.0.1",
  12.       "node3@127.0.0.1"
  13.     ]
  14.   }
  15. }

In this example we’ll split all the copies of the 00000000-7fffffff range. The API allows a combination of parameters such as: splitting all the ranges on all the nodes, all the ranges on just one node, or one particular range on one particular node. These are specified via the db, node and range job parameters.

To split all the copies of 00000000-7fffffff we issue a request like this:

  1. $ curl -s -H "Content-type: application/json" -XPOST $COUCH_URL:5984/_reshard/jobs \
  2.   -d '{"type": "split", "db":"db1", "range":"00000000-7fffffff"}' | jq '.'
  3. [
  4.   {
  5.     "ok": true,
  6.     "id": "001-ef512cfb502a1c6079fe17e9dfd5d6a2befcc694a146de468b1ba5339ba1d134",
  7.     "node": "node1@127.0.0.1",
  8.     "shard": "shards/00000000-7fffffff/db1.1554242778"
  9.   },
  10.   {
  11.     "ok": true,
  12.     "id": "001-cec63704a7b33c6da8263211db9a5c74a1cb585d1b1a24eb946483e2075739ca",
  13.     "node": "node2@127.0.0.1",
  14.     "shard": "shards/00000000-7fffffff/db1.1554242778"
  15.   },
  16.   {
  17.     "ok": true,
  18.     "id": "001-fc72090c006d9b059d4acd99e3be9bb73e986d60ca3edede3cb74cc01ccd1456",
  19.     "node": "node3@127.0.0.1",
  20.     "shard": "shards/00000000-7fffffff/db1.1554242778"
  21.   }
  22. ]

The request returned three jobs, one job for each of the three copies.

To check progress of these jobs use GET /_reshard/jobs or GET /_reshard/jobs/{jobid}.

Eventually, these jobs should complete and the shard map should look like this:

  1. $ curl -s $COUCH_URL:5984/db1/_shards | jq '.'
  2. {
  3.   "shards": {
  4.     "00000000-3fffffff": [
  5.       "node1@127.0.0.1",
  6.       "node2@127.0.0.1",
  7.       "node3@127.0.0.1"
  8.     ],
  9.     "40000000-7fffffff": [
  10.       "node1@127.0.0.1",
  11.       "node2@127.0.0.1",
  12.       "node3@127.0.0.1"
  13.     ],
  14.     "80000000-ffffffff": [
  15.       "node1@127.0.0.1",
  16.       "node2@127.0.0.1",
  17.       "node3@127.0.0.1"
  18.     ]
  19.   }
  20. }

4.4.6. Stopping Resharding Jobs

Resharding at the cluster level could be stopped and then restarted. This can be helpful to allow external tools which manipulate the shard map to avoid interfering with resharding jobs. To stop all resharding jobs on a cluster issue a PUT to /_reshard/state endpoint with the "state": "stopped" key and value. You can also specify an optional note or reason for stopping.

For example:

  1. $ curl -s -H "Content-type: application/json" \
  2.   -XPUT $COUCH_URL:5984/_reshard/state \
  3.   -d '{"state": "stopped", "reason":"Moving some shards"}'
  4. {"ok": true}

This state will then be reflected in the global summary:

  1. $ curl -s $COUCH_URL:5984/_reshard | jq .
  2. {
  3.   "state": "stopped",
  4.   "state_reason": "Moving some shards",
  5.   "completed": 74,
  6.   "failed": 0,
  7.   "running": 0,
  8.   "stopped": 0,
  9.   "total": 74
  10. }

To restart, issue a PUT request like above with running as the state. That should resume all the shard splitting jobs since their last checkpoint.

See the API reference for more details: /_reshard.

4.4.7. Merging Shards

The q value for a database can be set when the database is created or it can be increased later by splitting some of the shards Splitting Shards. In order to decrease q and merge some shards together, the database must be regenerated. Here are the steps:

  1. If there are running shard splitting jobs on the cluster, stop them via the HTTP API Stopping Resharding Jobs.
  2. Create a temporary database with the desired shard settings, by specifying the q value as a query parameter during the PUT operation.
  3. Stop clients accessing the database.
  4. Replicate the primary database to the temporary one. Multiple replications may be required if the primary database is under active use.
  5. Delete the primary database. Make sure nobody is using it!
  6. Recreate the primary database with the desired shard settings.
  7. Clients can now access the database again.
  8. Replicate the temporary back to the primary.
  9. Delete the temporary database.

Once all steps have completed, the database can be used again. The cluster will create and distribute its shards according to placement rules automatically.

Downtime can be avoided in production if the client application(s) can be instructed to use the new database instead of the old one, and a cut- over is performed during a very brief outage window.