Collection Management Commands
A collection is a single logical index that uses a single Solr configuration file (solrconfig.xml
) and a single index schema.
CREATE: Create a Collection
/admin/collections?action=CREATE&name=name
CREATE Parameters
The CREATE action allows the following parameters:
name
- The name of the collection to be created. This parameter is required.
router.name
The router name that will be used. The router defines how documents will be distributed among the shards. Possible values are
implicit
orcompositeId
, which is the default.The
implicit
router does not automatically route documents to different shards. Whichever shard you indicate on the indexing request (or within each document) will be used as the destination for those documents.The
compositeId
router hashes the value in the uniqueKey field and looks up that hash in the collection’s clusterstate to determine which shard will receive the document, with the additional ability to manually direct the routing.When using the
implicit
router, theshards
parameter is required. When using thecompositeId
router, thenumShards
parameter is required.For more information, see also the section Document Routing.
numShards
- The number of shards to be created as part of the collection. This is a required parameter when the
router.name
iscompositeId
. shards
- A comma separated list of shard names, e.g.,
shard-x,shard-y,shard-z
. This is a required parameter when therouter.name
isimplicit
. replicationFactor
The number of replicas to be created for each shard. The default is
1
.This will create a NRT type of replica. If you want another type of replica, see the
tlogReplicas
andpullReplica
parameters below. See the section Types of Replicas for more information about replica types.nrtReplicas
- The number of NRT (Near-Real-Time) replicas to create for this collection. This type of replica maintains a transaction log and updates its index locally. If you want all of your replicas to be of this type, you can simply use
replicationFactor
instead. tlogReplicas
- The number of TLOG replicas to create for this collection. This type of replica maintains a transaction log but only updates its index via replication from a leader. See the section Types of Replicas for more information about replica types.
pullReplicas
- The number of PULL replicas to create for this collection. This type of replica does not maintain a transaction log and only updates its index via replication from a leader. This type is not eligible to become a leader and should not be the only type of replicas in the collection. See the section Types of Replicas for more information about replica types.
maxShardsPerNode
When creating collections, the shards and/or replicas are spread across all available (i.e., live) nodes, and two replicas of the same shard will never be on the same node.
If a node is not live when the CREATE action is called, it will not get any parts of the new collection, which could lead to too many replicas being created on a single live node. Defining
maxShardsPerNode
sets a limit on the number of replicas the CREATE action will spread to each node.If the entire collection can not be fit into the live nodes, no collection will be created at all. The default
maxShardsPerNode
value is1
. A value of-1
means unlimited. If apolicy
is also specified then the stricter ofmaxShardsPerNode
and policy rules apply.createNodeSet
Allows defining the nodes to spread the new collection across. The format is a comma-separated list of node_names, such as
localhost:8983_solr,localhost:8984_solr,localhost:8985_solr
.If not provided, the CREATE operation will create shard-replicas spread across all live Solr nodes.
Alternatively, use the special value of
EMPTY
to initially create no shard-replica within the new collection and then later use the ADDREPLICA operation to add shard-replicas when and where required.createNodeSet.shuffle
Controls whether or not the shard-replicas created for this collection will be assigned to the nodes specified by the
createNodeSet
in a sequential manner, or if the list of nodes should be shuffled prior to creating individual replicas.A
false
value makes the results of a collection creation predictable and gives more exact control over the location of the individual shard-replicas, buttrue
can be a better choice for ensuring replicas are distributed evenly across nodes. The default istrue
.This parameter is ignored if
createNodeSet
is not also specified.collection.configName
Defines the name of the configuration (which must already be stored in ZooKeeper) to use for this collection.
If not provided, Solr will use the configuration of
_default
configset to create a new (and mutable) configset named<collectionName>.AUTOCREATED
and will use it for the new collection. When such a collection is deleted, its autocreated configset will be deleted by default when it is not in use by any other collection.router.field
If this parameter is specified, the router will look at the value of the field in an input document to compute the hash and identify a shard instead of looking at the
uniqueKey
field. If the field specified is null in the document, the document will be rejected.Please note that RealTime Get or retrieval by document ID would also require the parameter
_route_
(orshard.keys
) to avoid a distributed search.property.name=value
- Set core property name to value. See the section Defining core.properties for details on supported properties and values.
autoAddReplicas
When set to
true
, enables automatic addition of replicas when the number of active replicas falls below the value set forreplicationFactor
. This may occur if a replica goes down, for example. The default isfalse
, which means new replicas will not be added.While this parameter is provided as part of Solr’s set of features to provide autoscaling of clusters, it is available even when you have not implemented any other part of autoscaling (such as a policy). See the section SolrCloud Autoscaling Automatically Adding Replicas for more details about this option and how it can be used.
async
- Request ID to track this action which will be processed asynchronously.
rule
- Replica placement rules. See the section Rule-based Replica Placement for details.
snitch
- Details of the snitch provider. See the section Rule-based Replica Placement for details.
policy
- Name of the collection-level policy. See Defining Collection-Specific Policies for details.
waitForFinalState
- If
true
, the request will complete only when all affected replicas become active. The default isfalse
, which means that the API will return the status of the single action, which may be before the new replica is online and active. withCollection
- The name of the collection with which all replicas of this collection must be co-located. The collection must already exist and must have a single shard named
shard1
. See Colocating collections for more details. alias
- Starting with version 8.1 when a collection is created additionally an alias can be created that points to this collection. This parameter allows specifying the name of this alias, effectively combining this operation with CREATEALIAS
Collections are first created in read-write mode but can be put in readOnly
mode using the MODIFYCOLLECTION action.
CREATE Response
The response will include the status of the request and the new core names. If the status is anything other than "success", an error message will explain why the request failed.
Examples using CREATE
Input
http://localhost:8983/solr/admin/collections?action=CREATE&name=newCollection&numShards=2&replicationFactor=1&wt=xml
Output
<response>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">3764</int>
</lst>
<lst name="success">
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">3450</int>
</lst>
<str name="core">newCollection_shard1_replica1</str>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">3597</int>
</lst>
<str name="core">newCollection_shard2_replica1</str>
</lst>
</lst>
</response>
RELOAD: Reload a Collection
/admin/collections?action=RELOAD&name=name
The RELOAD action is used when you have changed a configuration in ZooKeeper.
RELOAD Parameters
name
- The name of the collection to reload. This parameter is required.
async
- Request ID to track this action which will be processed asynchronously.
RELOAD Response
The response will include the status of the request and the cores that were reloaded. If the status is anything other than "success", an error message will explain why the request failed.
Examples using RELOAD
Input
http://localhost:8983/solr/admin/collections?action=RELOAD&name=newCollection&wt=xml
Output
<response>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1551</int>
</lst>
<lst name="success">
<lst name="10.0.1.6:8983_solr">
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">761</int>
</lst>
</lst>
<lst name="10.0.1.4:8983_solr">
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1527</int>
</lst>
</lst>
</lst>
</response>
MODIFYCOLLECTION: Modify Attributes of a Collection
/admin/collections?action=MODIFYCOLLECTION&collection=<collection-name>&<attribute-name>=<attribute-value>&<another-attribute-name>=<another-value>&<yet_another_attribute_name>=
It’s possible to edit multiple attributes at a time. Changing these values only updates the z-node on ZooKeeper, they do not change the topology of the collection. For instance, increasing replicationFactor
will not automatically add more replicas to the collection but will allow more ADDREPLICA commands to succeed.
An attribute can be deleted by passing an empty value. For example, yet_another_attribute_name=
(with no value) will delete the yet_another_attribute_name
parameter from the collection.
MODIFYCOLLECTION Parameters
collection
- The name of the collection to be modified. This parameter is required.
attribute=value
- Key-value pairs of attribute names and attribute values.
At least one attribute
parameter is required.
The attributes that can be modified are:
maxShardsPerNode
replicationFactor
autoAddReplicas
collection.configName
rule
snitch
policy
withCollection
readOnly
other custom properties that use a
property.
prefix
See the CREATE action section above for details on these attributes.
Read-Only Mode
Setting the readOnly
attribute to true
puts the collection in read-only mode,
in which any index update requests are rejected. Other collection-level actions (e.g., adding /
removing / moving replicas) are still available in this mode.
The transition from the (default) read-write to read-only mode consists of the following steps:
the
readOnly
flag is changed in collection state,any new update requests are rejected with 403 FORBIDDEN error code (ongoing long-running requests are aborted, too),
a forced commit is performed to flush and commit any in-flight updates.
This may potentially take a long time if there are still major segment merges running in the background. |
a collection RELOAD action is executed.
Removing the readOnly
property or setting it to false enables the
processing of updates and reloads the collection.
LIST: List Collections
Fetch the names of the collections in the cluster.
/admin/collections?action=LIST
Examples using LIST
Input
http://localhost:8983/solr/admin/collections?action=LIST
Output
{
"responseHeader":{
"status":0,
"QTime":2011},
"collections":["collection1",
"example1",
"example2"]}
RENAME: Rename a Collection
/admin/collections?action=RENAME&name=existingName&target=targetName
Renaming a collection sets up a standard alias that points to the underlying collection, so that the same (unmodified) collection can now be referred to in query, index and admin operations using the new name.
This command does NOT actually rename the underlying Solr collection - it sets up a new one-to-one alias using the new name, or renames the existing alias so that it uses the new name, while still referring to the same underlying Solr collection. However, from the user’s point of view the collection can now be accessed using the new name, and the new name can be also referred to in other aliases.
The following limitations apply:
the existing name must be either a SolrCloud collection or a standard alias referring to a single collection. Aliases that refer to more than 1 collection are not supported.
the existing name must not be a Routed Alias.
the target name must not be an existing alias.
RENAME Command Parameters
name
- Name of the existing SolrCloud collection or an alias that refers to exactly one collection and is not a Routed Alias.
target
- Target name of the collection. This will be the new alias that refers to the underlying SolrCloud collection. The original name (or alias) of the collection will be replaced also in the existing aliases so that they also refer to the new name. Target name must not be an existing alias.
Examples using RENAME
Assuming there are two actual SolrCloud collections named collection1
and collection2
,
and the following aliases already exist:
col1 -> collection1
: this resolves tocollection1
.col2 -> collection2
: this resolves tocollection2
.simpleAlias -> col1
: this resolves tocollection1
.compoundAlias -> col1,col2
: this resolves tocollection1,collection2
The RENAME of col1
to foo
will change the aliases to the following:
foo -> collection1
: this resolves tocollection1
.col2 -> collection2
: this resolves tocollection2
.simpleAlias -> foo
: this resolves tocollection1
.compoundAlias -> foo,col2
: this resolves tocollection1,collection2
.
If we then rename collection1
(which is an actual collection name) to collection2
(which is also
an actual collection name) the following aliases will exist now:
foo -> collection2
: this resolves tocollection2
.col2 -> collection2
: this resolves tocollection2
.simpleAlias -> foo
: this resolves tocollection2
.compoundAlias -> foo,col2
: this would resolve now tocollection2,collection2
so it’s reduced to simplycollection2
.collection1
->collection2
: this newly created alias effectively hidescollection1
from regular query and update commands, which are directed now tocollection2
.
DELETE: Delete a Collection
/admin/collections?action=DELETE&name=collection
DELETE Parameters
name
- The name of the collection to delete. This parameter is required.
async
- Request ID to track this action which will be processed asynchronously.
DELETE Response
The response will include the status of the request and the cores that were deleted. If the status is anything other than "success", an error message will explain why the request failed.
Examples using DELETE
Input
Delete the collection named "newCollection".
http://localhost:8983/solr/admin/collections?action=DELETE&name=newCollection&wt=xml
Output
<response>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">603</int>
</lst>
<lst name="success">
<lst name="10.0.1.6:8983_solr">
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">19</int>
</lst>
</lst>
<lst name="10.0.1.4:8983_solr">
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">67</int>
</lst>
</lst>
</lst>
</response>
COLLECTIONPROP: Collection Properties
Add, edit or delete a collection property.
/admin/collections?action=COLLECTIONPROP&name=collectionName&propertyName=propertyName&propertyValue=propertyValue
COLLECTIONPROP Parameters
name
- The name of the collection for which the property would be set.
propertyName
- The name of the property.
propertyValue
- The value of the property. When not provided, the property is deleted.
COLLECTIONPROP Response
The response will include the status of the request and the properties that were updated or removed. If the status is anything other than "0", an error message will explain why the request failed.
Examples using COLLECTIONPROP
Input
http://localhost:8983/solr/admin/collections?action=COLLECTIONPROP&name=coll&propertyName=foo&propertyValue=bar&wt=xml
Output
<response>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">0</int>
</lst>
</response>
MIGRATE: Migrate Documents to Another Collection
/admin/collections?action=MIGRATE&collection=name&split.key=key1!&target.collection=target_collection&forward.timeout=60
The MIGRATE command is used to migrate all documents having a given routing key to another collection. The source collection will continue to have the same data as-is but it will start re-routing write requests to the target collection for the number of seconds specified by the forward.timeout
parameter. It is the responsibility of the user to switch to the target collection for reads and writes after the MIGRATE action completes.
The routing key specified by the split.key
parameter may span multiple shards on both the source and the target collections. The migration is performed shard-by-shard in a single thread. One or more temporary collections may be created by this command during the ‘migrate’ process but they are cleaned up at the end automatically.
This is a long running operation and therefore using the async
parameter is highly recommended. If the async
parameter is not specified then the operation is synchronous by default and keeping a large read timeout on the invocation is advised. Even with a large read timeout, the request may still timeout but that doesn’t necessarily mean that the operation has failed. Users should check logs, cluster state, source and target collections before invoking the operation again.
This command works only with collections using the compositeId router. The target collection must not receive any writes during the time the MIGRATE command is running otherwise some writes may be lost.
Please note that the MIGRATE API does not perform any de-duplication on the documents so if the target collection contains documents with the same uniqueKey as the documents being migrated then the target collection will end up with duplicate documents.
MIGRATE Parameters
collection
- The name of the source collection from which documents will be split. This parameter is required.
target.collection
- The name of the target collection to which documents will be migrated. This parameter is required.
split.key
- The routing key prefix. For example, if the uniqueKey of a document is "a!123", then you would use
split.key=a!
. This parameter is required. forward.timeout
- The timeout, in seconds, until which write requests made to the source collection for the given
split.key
will be forwarded to the target shard. The default is 60 seconds. property.name=value
- Set core property name to value. See the section Defining core.properties for details on supported properties and values.
async
- Request ID to track this action which will be processed asynchronously.
MIGRATE Response
The response will include the status of the request.
Examples using MIGRATE
Input
http://localhost:8983/solr/admin/collections?action=MIGRATE&collection=test1&split.key=a!&target.collection=test2&wt=xml
Output
<response>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">19014</int>
</lst>
<lst name="success">
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1</int>
</lst>
<str name="core">test2_shard1_0_replica1</str>
<str name="status">BUFFERING</str>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">2479</int>
</lst>
<str name="core">split_shard1_0_temp_shard1_0_shard1_replica1</str>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1002</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">21</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1655</int>
</lst>
<str name="core">split_shard1_0_temp_shard1_0_shard1_replica2</str>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">4006</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">17</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1</int>
</lst>
<str name="core">test2_shard1_0_replica1</str>
<str name="status">EMPTY_BUFFER</str>
</lst>
<lst name="192.168.43.52:8983_solr">
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">31</int>
</lst>
</lst>
<lst name="192.168.43.52:8983_solr">
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">31</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1</int>
</lst>
<str name="core">test2_shard1_1_replica1</str>
<str name="status">BUFFERING</str>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1742</int>
</lst>
<str name="core">split_shard1_1_temp_shard1_1_shard1_replica1</str>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1002</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">15</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1917</int>
</lst>
<str name="core">split_shard1_1_temp_shard1_1_shard1_replica2</str>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">5007</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">8</int>
</lst>
</lst>
<lst>
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">1</int>
</lst>
<str name="core">test2_shard1_1_replica1</str>
<str name="status">EMPTY_BUFFER</str>
</lst>
<lst name="192.168.43.52:8983_solr">
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">30</int>
</lst>
</lst>
<lst name="192.168.43.52:8983_solr">
<lst name="responseHeader">
<int name="status">0</int>
<int name="QTime">30</int>
</lst>
</lst>
</lst>
</response>
REINDEXCOLLECTION: Re-Index a Collection
/admin/collections?action=REINDEXCOLLECTION&name=name
The REINDEXCOLLECTION command reindexes a collection using existing data from the source collection.
Reindexing is potentially a lossy operation - some of the existing indexed data that is not available as stored fields may be lost, so users should use this command with caution, evaluating the potential impact by using different source and target collection names first, and preserving the source collection until the evaluation is complete. |
The target collection must not exist (and may not be an alias). If the target collection name is the same as the source collection then first a unique sequential name will be generated for the target collection, and then after reindexing is done an alias will be created that points from the source name to the actual sequentially-named target collection.
When reindexing is started the source collection is put in read-only mode to ensure that all source documents are properly processed.
Using optional parameters a different index schema, collection shape (number of shards and replicas) or routing parameters can be requested for the target collection.
Reindexing is executed as a streaming expression daemon, which runs on one of the
source collection’s replicas. It is usually a time-consuming operation so it’s recommended to execute
it as an asynchronous request in order to avoid request time outs. Only one reindexing operation may
execute concurrently for a given source collection. Long-running, erroneous or crashed reindexing
operations may be terminated by using the abort
option, which also removes partial results.
REINDEXCOLLECTION Parameters
name
- Source collection name, may be an alias. This parameter is required.
cmd
- Optional command. Default command is
start
. Currently supported commands are:start
- default, starts processing if not already running,abort
- aborts an already running reindexing (or clears a left-over status after a crash), and deletes partial results,status
- returns detailed status of a running reindexing command.
target
- Target collection name, optional. If not specified a unique name will be generated and after all documents have been copied an alias will be created that points from the source collection name to the unique sequentially-named collection, effectively "hiding" the original source collection from regular update and search operations.
q
- Optional query to select documents for reindexing. Default value is
*:*
. fl
- Optional list of fields to reindex. Default value is
*
. rows
- Documents are transferred in batches. Depending on the average size of the document large batch sizes may cause memory issues. Default value is 100.
configName
collection.configName
- Optional name of the configset for the target collection. Default is the same as the source collection.
There’s a number of optional parameters that determine the target collection layout. If they
are not specified in the request then their values are copied from the source collection.
The following parameters are currently supported (described in details in the CREATE collection section):
numShards
, replicationFactor
, nrtReplicas
, tlogReplicas
, pullReplicas
, maxShardsPerNode
,
autoAddReplicas
, shards
, policy
, createNodeSet
, createNodeSet.shuffle
, router.*
.
removeSource
- Optional boolean. If true then after the processing is successfully finished the source collection will be deleted.
async
- Optional request ID to track this action which will be processed asynchronously.
When the reindexing process has completed the target collection is marked using
property.rx: "finished"
, and the source collection state is updated to become read-write.
On any errors the command will delete any temporary and target collections and also reset the
state of the source collection’s read-only flag.
Examples using REINDEXCOLLECTION
Input
http://localhost:8983/solr/admin/collections?action=REINDEXCOLLECTION&name=newCollection&numShards=3&configName=conf2&q=id:aa*&fl=id,string_s
This request specifies a different schema for the target collection, copies only some of the fields, selects only the documents
matching a query, and also potentially re-shapes the collection by explicitly specifying 3 shards. Since the target collection
hasn’t been specified in the parameters, a collection with a unique name, e.g., .rx_newCollection_2
, will be created and on success
an alias pointing from newCollection
to .rx_newCollection_2
will be created, effectively replacing the source collection
for the purpose of indexing and searching. The source collection is assumed to be small so a synchronous request was made.
Output
{
"responseHeader":{
"status":0,
"QTime":10757},
"reindexStatus":{
"phase":"done",
"inputDocs":13416,
"processedDocs":376,
"actualSourceCollection":".rx_newCollection_1",
"state":"finished",
"actualTargetCollection":".rx_newCollection_2",
"checkpointCollection":".rx_ck_newCollection"
}
}
As a result a new collection .rx_newCollection_2
has been created, with selected documents reindexed to 3 shards, and
with an alias pointing from newCollection
to this one. The status also shows that the source collection
was already an alias to .rx_newCollection_1
, which was likely a result of a previous reindexing.
COLSTATUS: Detailed Status of a Collection’s Indexes
The COLSTATUS command provides a detailed description of the collection status, including low-level index information about segments and field data.
This command also checks the compliance of Lucene index field types with the current Solr collection schema and indicates the names of non-compliant fields, i.e., Lucene fields with field types incompatible (or different) from the corresponding Solr field types declared in the current schema. Such incompatibilities may result from incompatible schema changes or after migration of data to a different major Solr release.
/admin/collections?action=COLSTATUS&collection=coll&coreInfo=true&segments=true&fieldInfo=true&sizeInfo=true
COLSTATUS Parameters
collection
- Collection name (optional). If missing then it means all collections.
coreInfo
- Optional boolean. If true then additional information will be provided about SolrCore of shard leaders.
segments
- Optional boolean. If true then segment information will be provided.
fieldInfo
- Optional boolean. If true then detailed Lucene field information will be provided and their corresponding Solr schema types.
sizeInfo
- Optional boolean. If true then additional information about the index files size and their RAM usage will be provided.
Index Size Analysis Tool
The COLSTATUS
command also provides a tool for analyzing and estimating the composition of raw index data. Please note that
this tool should be used with care because it generates a significant IO load on all shard leaders of the
analyzed collections. A sampling threshold and a sampling percent parameters can be adjusted to reduce this
load to some degree.
Size estimates produced by this tool are only approximate and represent the aggregated size of uncompressed index data. In reality these values would never occur, because Lucene (and Solr) always stores data in a compressed format - still, these values help to understand what occupies most of the space and the relative size of each type of data and each field in the index.
In the following sections whenever "size" is mentioned it means an estimated aggregated size of uncompressed (raw) data.
The following parameters are specific to this tool:
rawSize
- Optional boolean. If true then run the raw index data analysis tool (other boolean options below imply this option if any of them are true). Command response will include sections that show estimated breakdown of data size per field and per data type.
rawSizeSummary
- Optional boolean. If true then include also a more detailed breakdown of data size per field and per type.
rawSizeDetails
- Optional boolean. If true then provide exhaustive details that include statistical distribution of items per field and per type as well as top 20 largest items per field.
rawSizeSamplingPercent
- Optional float. When the index is larger than a certain threshold (100k documents per shard) only a part of data is actually retrieved and analyzed in order to reduce the IO load, and then the final results are extrapolated. Values must be greater than 0 and less or equal to 100.0. Default value is 5.0. Very small values (between 0.0 and 1.0) may introduce significant estimation errors. Also, values that would result in less than 10 documents being sampled are rejected with an exception.
Response for this command always contains two sections:
fieldsBySize
is a map where field names are keys and values are estimated sizes of raw (uncompressed) data that belongs to the field. The map is sorted by size so that it’s easy to see what field occupies most space.typesBySize
is a map where data types are the keys and values are estimates sizes of raw (uncompressed) data of particular type. This map is also sorted by size.
Optional sections include:
summary
section containing a breakdown of data sizes for each field by data type.details
section containing detailed statistical summary of size distribution within each field, per data type. This section also showstopN
values by size from each field.
Data types shown in the response can be roughly divided into the following groups:
storedFields
- represents the raw uncompressed data in stored fields. For example, for UTF-8 strings this represents the aggregated sum of the number of bytes in the strings' UTF-8 representation, for long numbers this is 8 bytes per value, etc.terms_terms
- represents the aggregated size of the term dictionary. The size of this data is affected by the the number and length of unique terms, which in turn depends on the field size and the analysis chain.terms_postings
- represents the aggregated size of all term position and offset information, if present. This information may be absent if position-based searching, such as phrase queries, is not needed.terms_payloads
- represents the aggregated size of all per-term payload data, if present.norms
- represents the aggregated size of field norm information. This information may be omitted if a field has anomitNorms
flag in the schema, which is common for fields that don’t need weighting or scoring by field length.termVectors
- represents the aggregated size of term vectors.docValues_*
- represents aggregated size of doc values, by type (e.g.,docValues_numeric
,docValues_binary
, etc).points
- represents aggregated size of point values.
COLSTATUS Response
The response will include an overview of the collection status, the number of active or inactive shards and replicas, and additional index information of shard leaders.
Examples using COLSTATUS
Input
http://localhost:8983/solr/admin/collections?action=COLSTATUS&collection=gettingstarted&fieldInfo=true&sizeInfo=true
Output
{
"responseHeader": {
"status": 0,
"QTime": 50
},
"gettingstarted": {
"stateFormat": 2,
"znodeVersion": 16,
"properties": {
"autoAddReplicas": "false",
"maxShardsPerNode": "-1",
"nrtReplicas": "2",
"pullReplicas": "0",
"replicationFactor": "2",
"router": {
"name": "compositeId"
},
"tlogReplicas": "0"
},
"activeShards": 2,
"inactiveShards": 0,
"schemaNonCompliant": [
"(NONE)"
],
"shards": {
"shard1": {
"state": "active",
"range": "80000000-ffffffff",
"replicas": {
"total": 2,
"active": 2,
"down": 0,
"recovering": 0,
"recovery_failed": 0
},
"leader": {
"coreNode": "core_node4",
"core": "gettingstarted_shard1_replica_n1",
"base_url": "http://192.168.0.80:8983/solr",
"node_name": "192.168.0.80:8983_solr",
"state": "active",
"type": "NRT",
"force_set_state": "false",
"leader": "true",
"segInfos": {
"info": {
"minSegmentLuceneVersion": "9.0.0",
"commitLuceneVersion": "9.0.0",
"numSegments": 40,
"segmentsFileName": "segments_w",
"totalMaxDoc": 686953,
"userData": {
"commitCommandVer": "1627350608019193856",
"commitTimeMSec": "1551962478819"
}
},
"fieldInfoLegend": [
"I - Indexed",
"D - DocValues",
"xxx - DocValues type",
"V - TermVector Stored",
"O - Omit Norms",
"F - Omit Term Frequencies & Positions",
"P - Omit Positions",
"H - Store Offsets with Positions",
"p - field has payloads",
"s - field uses soft deletes",
":x:x:x - point data dim : index dim : num bytes"
],
"segments": {
"_i": {
"name": "_i",
"delCount": 738,
"softDelCount": 0,
"hasFieldUpdates": false,
"sizeInBytes": 109398213,
"size": 70958,
"age": "2019-03-07T12:34:24.761Z",
"source": "merge",
"version": "9.0.0",
"createdVersionMajor": 9,
"minVersion": "9.0.0",
"diagnostics": {
"os": "Mac OS X",
"java.vendor": "Oracle Corporation",
"java.version": "1.8.0_191",
"java.vm.version": "25.191-b12",
"lucene.version": "9.0.0",
"mergeMaxNumSegments": "-1",
"os.arch": "x86_64",
"java.runtime.version": "1.8.0_191-b12",
"source": "merge",
"mergeFactor": "10",
"os.version": "10.14.3",
"timestamp": "1551962064761"
},
"attributes": {
"Lucene50StoredFieldsFormat.mode": "BEST_SPEED"
},
"largestFiles": {
"_i.fdt": "42.5 MB",
"_i_Lucene80_0.dvd": "35.3 MB",
"_i_Lucene50_0.pos": "11.1 MB",
"_i_Lucene50_0.doc": "10 MB",
"_i_Lucene50_0.tim": "4.3 MB"
},
"ramBytesUsed": {
"total": 49153,
"postings [PerFieldPostings(segment=_i formats=1)]": {
"total": 31023,
"fields": {
"dc": {
"flags": "I-----------",
"schemaType": "text_general"
},
"dc_str": {
"flags": "-Dsrs-------",
"schemaType": "strings"
},
"dc.title": {
"flags": "I-----------",
"docCount": 70958,
"sumDocFreq": 646756,
"sumTotalTermFreq": 671817,
"schemaType": "text_general"
},
"dc.date": {
"flags": "-Dsrn-------:1:1:8",
"schemaType": "pdates"
}
}}}}}}}}}}}
Example of using the raw index data analysis tool:
Input
http://localhost:8983/solr/admin/collections?action=COLSTATUS&collection=gettingstarted&rawSize=true&rawSizeSamplingPercent=0.1
Output
{
"responseHeader": {
"status": 0,
"QTime": 26812
},
"gettingstarted": {
"stateFormat": 2,
"znodeVersion": 33,
"properties": {
"autoAddReplicas": "false",
"maxShardsPerNode": "-1",
"nrtReplicas": "2",
"pullReplicas": "0",
"replicationFactor": "2",
"router": {
"name": "compositeId"
},
"tlogReplicas": "0"
},
"activeShards": 2,
"inactiveShards": 0,
"schemaNonCompliant": [
"(NONE)"
],
"shards": {
"shard1": {
"state": "active",
"range": "80000000-ffffffff",
"replicas": {
"total": 2,
"active": 2,
"down": 0,
"recovering": 0,
"recovery_failed": 0
},
"leader": {
"coreNode": "core_node5",
"core": "gettingstarted_shard1_replica_n2",
"base_url": "http://192.168.0.80:8983/solr",
"node_name": "192.168.0.80:8983_solr",
"state": "active",
"type": "NRT",
"force_set_state": "false",
"leader": "true",
"segInfos": {
"info": {
"minSegmentLuceneVersion": "9.0.0",
"commitLuceneVersion": "9.0.0",
"numSegments": 46,
"segmentsFileName": "segments_4h",
"totalMaxDoc": 3283741,
"userData": {
"commitCommandVer": "1635676266902323200",
"commitTimeMSec": "1559902446318"
}
},
"rawSize": {
"fieldsBySize": {
"revision.text": "7.9 GB",
"revision.text_str": "734.7 MB",
"revision.comment_str": "259.1 MB",
"revision": "239.2 MB",
"revision.sha1": "211.9 MB",
"revision.comment": "201.3 MB",
"title": "114.9 MB",
"revision.contributor": "103.5 MB",
"revision.sha1_str": "96.4 MB",
"revision.id": "75.2 MB",
"ns": "75.2 MB",
"revision.timestamp": "75.2 MB",
"revision.contributor.id": "74.7 MB",
"revision.format": "69 MB",
"id": "65 MB",
"title_str": "26.8 MB",
"revision.model_str": "25.4 MB",
"_version_": "24.9 MB",
"_root_": "24.7 MB",
"revision.contributor.ip_str": "22 MB",
"revision.contributor_str": "21.8 MB",
"revision_str": "15.5 MB",
"revision.contributor.ip": "13.5 MB",
"restrictions_str": "428.7 KB",
"restrictions": "164.2 KB",
"name_str": "84 KB",
"includes_str": "8.8 KB"
},
"typesBySize": {
"storedFields": "7.8 GB",
"docValues_sortedSet": "1.2 GB",
"terms_postings": "788.8 MB",
"terms_terms": "342.2 MB",
"norms": "237 MB",
"docValues_sortedNumeric": "124.3 MB",
"points": "115.7 MB",
"docValues_numeric": "24.9 MB",
"docValues_sorted": "18.5 MB"
}
}
}
}
},
"shard2": {
"state": "active",
"range": "0-7fffffff",
"replicas": {
"total": 2,
"active": 2,
"down": 0,
"recovering": 0,
"recovery_failed": 0
},
"leader": {
"coreNode": "core_node8",
"core": "gettingstarted_shard2_replica_n6",
"base_url": "http://192.168.0.80:8983/solr",
"node_name": "192.168.0.80:8983_solr",
"state": "active",
"type": "NRT",
"force_set_state": "false",
"leader": "true",
"segInfos": {
"info": {
"minSegmentLuceneVersion": "9.0.0",
"commitLuceneVersion": "9.0.0",
"numSegments": 55,
"segmentsFileName": "segments_4d",
"totalMaxDoc": 3284863,
"userData": {
"commitCommandVer": "1635676259742646272",
"commitTimeMSec": "1559902445005"
}
},
"rawSize": {
"fieldsBySize": {
"revision.text": "8.3 GB",
"revision.text_str": "687.5 MB",
"revision": "238.9 MB",
"revision.sha1": "212 MB",
"revision.comment_str": "211.5 MB",
"revision.comment": "201.7 MB",
"title": "115.9 MB",
"revision.contributor": "103.4 MB",
"revision.sha1_str": "96.3 MB",
"ns": "75.2 MB",
"revision.id": "75.2 MB",
"revision.timestamp": "75.2 MB",
"revision.contributor.id": "74.6 MB",
"revision.format": "69 MB",
"id": "67 MB",
"title_str": "29.5 MB",
"_version_": "24.8 MB",
"revision.model_str": "24 MB",
"revision.contributor_str": "21.7 MB",
"revision.contributor.ip_str": "20.9 MB",
"revision_str": "15.5 MB",
"revision.contributor.ip": "13.8 MB",
"restrictions_str": "411.1 KB",
"restrictions": "132.9 KB",
"name_str": "42 KB",
"includes_str": "41 KB"
},
"typesBySize": {
"storedFields": "8.2 GB",
"docValues_sortedSet": "1.1 GB",
"terms_postings": "787.4 MB",
"terms_terms": "337.5 MB",
"norms": "236.6 MB",
"docValues_sortedNumeric": "124.1 MB",
"points": "115.7 MB",
"docValues_numeric": "24.9 MB",
"docValues_sorted": "20.5 MB"
}
}
}
}
}
}
}
}
BACKUP: Backup Collection
Backs up Solr collections and associated configurations to a shared filesystem - for example a Network File System.
/admin/collections?action=BACKUP&name=myBackupName&collection=myCollectionName&location=/path/to/my/shared/drive
The BACKUP command will backup Solr indexes and configurations for a specified collection. The BACKUP command takes one copy from each shard for the indexes. For configurations, it backs up the configset that was associated with the collection and metadata.
BACKUP Parameters
collection
- The name of the collection to be backed up. This parameter is required.
name
- What to name the backup that is created. This is checked to make sure it doesn’t already exist, and otherwise an error message is raised. This parameter is required.
location
- The location on a shared drive for the backup command to write to. Alternately it can be set as a cluster property.
async
- Request ID to track this action which will be processed asynchronously.
repository
- The name of a repository to be used for the backup. If no repository is specified then the local filesystem repository will be used automatically.
RESTORE: Restore Collection
Restores Solr indexes and associated configurations.
/admin/collections?action=RESTORE&name=myBackupName&location=/path/to/my/shared/drive&collection=myRestoredCollectionName
The RESTORE operation will create a collection with the specified name in the collection parameter. You cannot restore into the same collection the backup was taken from. Also the target collection should not be present at the time the API is called as Solr will create it for you.
The collection created will be have the same number of shards and replicas as the original collection, preserving routing information, etc. Optionally, you can override some parameters documented below.
While restoring, if a configset with the same name exists in ZooKeeper then Solr will reuse that, or else it will upload the backed up configset in ZooKeeper and use that.
You can use the collection CREATEALIAS command to make sure clients don’t need to change the endpoint to query or index against the newly restored collection.
RESTORE Parameters
collection
- The collection where the indexes will be restored into. This parameter is required.
name
- The name of the existing backup that you want to restore. This parameter is required.
location
- The location on a shared drive for the RESTORE command to read from. Alternately it can be set as a cluster property.
async
- Request ID to track this action which will be processed asynchronously.
repository
- The name of a repository to be used for the backup. If no repository is specified then the local filesystem repository will be used automatically.
There’s a number of optional parameters that determine the target collection layout.
The following parameters are currently supported (described in detail in the CREATE collection section):
createNodeSet
, createNodeSet.shuffle
.
Note: for createNodeSet
the special value of EMPTY
is not allowed with this command.
Overridable Parameters
Additionally, there are several parameters that may have been set on the original collection that can be overridden when restoring the backup (described in detail in the CREATE collection section):
collection.configName
, replicationFactor
, nrtReplicas
, tlogReplicas
, pullReplicas
, property.name=value
.
REBALANCELEADERS: Rebalance Leaders
Reassigns leaders in a collection according to the preferredLeader property across active nodes.
/admin/collections?action=REBALANCELEADERS&collection=collectionName
Leaders are assigned in a collection according to the preferredLeader
property on active nodes. This command should be run after the preferredLeader property has been assigned via the BALANCESHARDUNIQUE or ADDREPLICAPROP commands.
It is not required that all shards in a collection have a preferredLeader property. Rebalancing will only attempt to reassign leadership to those replicas that have the preferredLeader property set to true and are not currently the shard leader and are currently active.
|
REBALANCELEADERS Parameters
collection
- The name of the collection to rebalance
preferredLeaders
on. This parameter is required. maxAtOnce
The maximum number of reassignments to have queue up at once. Values <=0 are use the default value Integer.MAX_VALUE.
When this number is reached, the process waits for one or more leaders to be successfully assigned before adding more to the queue.
maxWaitSeconds
Defaults to
60
. This is the timeout value when waiting for leaders to be reassigned. IfmaxAtOnce
is less than the number of reassignments that will take place, this is the maximum interval that any single wait for at least one reassignment.For example, if 10 reassignments are to take place and
maxAtOnce
is1
andmaxWaitSeconds
is60
, the upper bound on the time that the command may wait is 10 minutes.
REBALANCELEADERS Response
The response will include the status of the request. A status of "0" indicates the request was processed, not that all assignments were successful. Examine the "Summary" section for that information.
Examples using REBALANCELEADERS
Input
Either of these commands would cause all the active replicas that had the preferredLeader
property set and were not already the preferred leader to become leaders.
http://localhost:8983/solr/admin/collections?action=REBALANCELEADERS&collection=collection1&wt=json
http://localhost:8983/solr/admin/collections?action=REBALANCELEADERS&collection=collection1&maxAtOnce=5&maxWaitSeconds=30&wt=json
Output
In this example:
In the "alreadyLeaders" section, core_node5 was already the leader, so there were no changes in leadership for shard1.
In the "inactivePreferreds" section, core_node57 had the preferredLeader property set, but the node was not active, the leader for shard7 was not changed. This is considered successful.
In the "successes" section, core_node23 was not the leader for shard3, so leadership was assigned to that replica.
The "Summary" section with the "Success" tag indicates that the command rebalanced all active replicas with the preferredLeader property set as requried. If a replica cannot be made leader due to not being healthy (for example, it is on a Solr instance that is not running), it’s also considered success.
{
"responseHeader":{
"status":0,
"QTime":3054},
"Summary":{
"Success":"All active replicas with the preferredLeader property set are leaders"},
"alreadyLeaders":{
"core_node5":{
"status":"skipped",
"msg":"Replica core_node5 is already the leader for shard shard1. No change necessary"}},
"inactivePreferreds":{
"core_node57":{
"status":"skipped",
"msg":"Replica core_node57 is a referredLeader for shard shard7, but is inactive. No change necessary"}},
"successes":{
"shard3":{
"status":"success",
"msg":"Successfully changed leader of slice shard3 to core_node23"}}}
Examining the clusterstate after issuing this call should show that every active replica that has the preferredLeader
property should also have the "leader" property set to true.
The added work done by an NRT leader is quite small and only present when indexing. The primary use-case is to redistribute the leader role if there are a large number of leaders concentrated on a small number of nodes. Rebalancing will likely not improve performance unless the imbalance of leadership roles is measured in multiples of 10. |
The BALANCESHARDUNIQUE command that distributes the preferredLeader property does not guarantee perfect distribution and in some collection topologies it is impossible to make that guarantee. |