The CRUD module allows to perform CRUD operations on the cluster. It also provides the crud-storage role for Tarantool Cartridge.

The CRUD operations should be called from router. All storage replica sets should call crud.init_storage() (or enable the crud-storage role) first to initialize storage-side functions that are used to manipulate data across the cluster. All routers should call crud.init_router() (or enable the crud-router role) to make crud functions callable via net.box.

All operations return a table that contains rows (tuples) and metadata (space format). It can be used to convert received tuples to objects via crud.unflatten_rows function.

For example:

res, err = crud.select('customers')
res
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [1, 12477, 'Elizabeth', 12]
  - [2, 21401, 'David', 33]
...
crud.unflatten_rows(res.rows, res.metadata)
---
- - bucket_id: 12477
    age: 12
    name: Elizabeth
    id: 1
  - bucket_id: 21401
    age: 33
    name: David
    id: 2
...

Notes:

• A space should have a format.

Sharding key and bucket id calculation

Sharding key is a set of tuple field values used for calculation bucket ID. Sharding key definition is a set of tuple field names that describe what tuple field should be a part of sharding key. Bucket ID determines which replicaset stores certain data. Function that used for bucket ID calculation is named sharding function.

By default CRUD calculates bucket ID using primary key and a function vshard.router.bucket_id_strcrc32(key), it happen automatically and doesn't require any actions from user side. However, for operations that accepts tuple/object bucket ID can be specified as tuple/object field as well as opts.bucket_id value.

Starting from 0.10.0 users who don't want to use primary key as a sharding key may set custom sharding key definition as a part of DDL schema or insert manually to the space _ddl_sharding_key (for both cases consider a DDL module documentation). As soon as sharding key for a certain space is available in _ddl_sharding_key space CRUD will use it for bucket ID calculation automatically. Note that CRUD methods delete(), get() and update() requires that sharding key must be a part of primary key.

Starting from 0.11.0 you can specify sharding function to calculate bucket_id with sharding func definition as a part of DDL schema or insert manually to the space _ddl_sharding_func.

Automatic sharding key and function reload is supported since version 0.11.0. Version 0.11.0 contains critical bug that causes some CRUD methods to fail with "Sharding hash mismatch" error if ddl is set and bucket_id is provided explicitly (#278). Please, upgrade to 0.11.1 instead.

CRUD uses strcrc32 as sharding function by default. The reason why using of strcrc32 is undesirable is that this sharding function is not consistent for cdata numbers. In particular, it returns 3 different values for normal Lua numbers like 123, for unsigned long long cdata (like 123ULL, or ffi.cast('unsigned long long', 123)), and for signed long long cdata (like 123LL, or ffi.cast('long long', 123)).

We cannot change default sharding function strcrc32 due to backward compatibility concerns, but please consider using better alternatives for sharding function. mpcrc32 is one of them.

Table below describe what operations supports custom sharding key:

Current limitations for using custom sharding key:

• No support of JSON path for sharding key, see #219.
• primary_index_fieldno_map is not cached, see #243.

-- Insert tuple
local result, err = crud.insert(space_name, tuple, opts)
-- Insert object
local result, err = crud.insert_object(space_name, object, opts)

where:

• space_name (string) - name of the space to insert an object
• tuple / object (table) - tuple/object to insert
• opts:
  • timeout (?number) - vshard.call timeout (in seconds)
  • bucket_id (?number|cdata) - bucket ID
  • fields (?table) - field names for getting only a subset of fields

Returns metadata and array contains one inserted row, error.

Example:

crud.insert('customers', {1, box.NULL, 'Elizabeth', 23})
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [1, 477, 'Elizabeth', 23]
...
crud.insert_object('customers', {
    id = 2, name = 'Elizabeth', age = 24,
})
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [2, 401, 'Elizabeth', 24]
...

local object, err = crud.get(space_name, key, opts)

where:

• space_name (string) - name of the space
• key (any) - primary key value
• opts:
  • fields (?table) - field names for getting only a subset of fields
  • bucket_id (?number|cdata) - bucket ID
  • timeout (?number) - vshard.call timeout (in seconds)
  • mode (?string, read or write) - if write is specified then get is performed on master
  • prefer_replica (?boolean) - if true then the preferred target is one of the replicas
  • balance (?boolean) - use replica according to vshard load balancing policy

Returns metadata and array contains one row, error.

Example:

crud.get('customers', 1)
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [1, 477, 'Elizabeth', 23]
...

local object, err = crud.update(space_name, key, operations, opts)

where:

• space_name (string) - name of the space
• key (any) - primary key value
• operations (table) - update operations
• opts:
  • timeout (?number) - vshard.call timeout (in seconds)
  • bucket_id (?number|cdata) - bucket ID
  • fields (?table) - field names for getting only a subset of fields

Returns metadata and array contains one updated row, error.

Example:

crud.update('customers', 1, {{' ', 'age', 1}})
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [1, 477, 'Elizabeth', 24]
...

local object, err = crud.delete(space_name, key, opts)

where:

• space_name (string) - name of the space
• key (any) - primary key value
• opts:
  • timeout (?number) - vshard.call timeout (in seconds)
  • bucket_id (?number|cdata) - bucket ID
  • fields (?table) - field names for getting only a subset of fields

Returns metadata and array contains one deleted row (empty for vinyl), error.

Example:

crud.delete('customers', 1)
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [1, 477, 'Elizabeth', 24]

-- Replace tuple
local result, err = crud.replace(space_name, tuple, opts)
-- Replace object
local result, err = crud.replace_object(space_name, object, opts)

where:

• space_name (string) - name of the space
• tuple / object (table) - tuple/object to insert or replace exist one
• opts:
  • timeout (?number) - vshard.call timeout (in seconds)
  • bucket_id (?number|cdata) - bucket ID
  • fields (?table) - field names for getting only a subset of fields

Returns inserted or replaced rows and metadata or nil with error.

Example:

crud.replace('customers', {1, box.NULL, 'Alice', 22})
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [1, 477, 'Alice', 22]
...
crud.replace_object('customers', {
    id = 1, name = 'Alice', age = 22,
})
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [1, 477, 'Alice', 22]
...

-- Upsert tuple
local result, err = crud.upsert(space_name, tuple, operations, opts)
-- Upsert object
local result, err = crud.upsert_object(space_name, tuple, operations, opts)

where:

• space_name (string) - name of the space
• tuple / object (table) - tuple/object to insert if there is no existing tuple which matches the key fields
• operations (table) - update operations if there is an existing tuple which matches the key fields of tuple
• opts:
  • timeout (?number) - vshard.call timeout (in seconds)
  • bucket_id (?number|cdata) - bucket ID
  • fields (?table) - field names for getting only a subset of fields

Returns metadata and empty array of rows or nil, error.

Example:

crud.upsert('customers',
    {1, box.NULL, 'Alice', 22},
    {{' ', 'age', 1}})
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows: []
...
crud.upsert_object('customers',
    {id = 1, name = 'Alice', age = 22},
    {{' ', 'age', 1}})
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows: []
...

CRUD supports multi-conditional selects, treating a cluster as a single space. The conditions may include field names or numbers, as well as index names. The recommended first condition is a TREE index; this helps reducing the number of tuples to scan. Otherwise a full scan is performed.

local objects, err = crud.select(space_name, conditions, opts)

where:

• space_name (string) - name of the space
• conditions (?table) - array of select conditions
• opts:
  • first (?number) - the maximum count of the objects to return. If negative value is specified, the last objects are returned (after option is required in this case).
  • after (?table) - tuple after which objects should be selected
  • batch_size (?number) - number of tuples to process per one request to storage
  • bucket_id (?number|cdata) - bucket ID
  • force_map_call (?boolean) - if true then the map call is performed without any optimizations even if full primary key equal condition is specified
  • timeout (?number) - vshard.call timeout (in seconds)
  • fields (?table) - field names for getting only a subset of fields
  • mode (?string, read or write) - if write is specified then select is performed on master
  • prefer_replica (?boolean) - if true then the preferred target is one of the replicas
  • balance (?boolean) - use replica according to vshard load balancing policy

Returns metadata and array of rows, error.

Select conditions are very similar to Tarantool update operations.

Each condition is a table {operator, field-identifier, value}:

• Supported operators are: = (or ==), >, >=, <, <=.
• Field identifier can be field name, field number, or index name.

Example:

crud.select('customers', {{'<=', 'age', 35}})
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [5, 1172, 'Jack', 35]
  - [3, 2804, 'David', 33]
  - [6, 1064, 'William', 25]
  - [7, 693, 'Elizabeth', 18]
  - [1, 477, 'Elizabeth', 12]
...

Note: tuples are sorted by age because space has index age. Otherwise, tuples are sorted by primary key.

See more examples of select queries here.

You can iterate across a distributed space using the crud.pairs function. Its arguments are the same as crud.select arguments, but negative first values aren't allowed. User could pass use_tomap flag (false by default) to iterate over flat tuples or objects.

Example:

local tuples = {}
for _, tuple in crud.pairs('customers', {{'<=', 'age', 35}}, {use_tomap = false}) do
    -- {5, 1172, 'Jack', 35}
    table.insert(tuples, tuple)
end

local objects = {}
for _, object in crud.pairs('customers', {{'<=', 'age', 35}}, {use_tomap = true}) do
    -- {id = 5, name = 'Jack', bucket_id = 1172, age = 35}
    table.insert(objects, object)
end

See more examples of pairs queries here.

-- Find the minimum value in the specified index
local result, err = crud.min(space_name, 'age', opts)
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [1, 477, 'Elizabeth', 12]

-- Find the maximum value in the specified index
local result, err = crud.max(space_name, 'age', opts)
---
- metadata:
  - {'name': 'id', 'type': 'unsigned'}
  - {'name': 'bucket_id', 'type': 'unsigned'}
  - {'name': 'name', 'type': 'string'}
  - {'name': 'age', 'type': 'number'}
  rows:
  - [5, 1172, 'Jack', 35]

You could use crud.cut_rows function to cut off scan key and primary key values that were merged to the select/pairs partial result (select/pairs with fields option).

local res, err = crud.cut_rows(rows, metadata, fields)

where:

• rows (table) - array of tuples for cutting
• matadata (?metadata) - metadata about rows fields
• fields (table) - field names of fields that should be contained in the result

Returns metadata and array of rows, error.

See more examples of crud.cut_rows usage here and here.

If you use pairs with use_tomap flag and you need to cut off scan key and primary key values that were merged to the pairs partial result (pairs with fields option) you should use crud.cut_objects.

local new_objects = crud.cut_objects(objects, fields)

where:

• objects (table) - array of objects for cutting
• fields (table) - field names of fields that should be contained in the result

Returns array of objects.

See more examples of crud.cut_objects usage here.

-- Truncate space
local result, err = crud.truncate(space_name, opts)

where:

• space_name (string) - name of the space
• opts:
  • timeout (?number) - vshard.call timeout (in seconds)

Returns true or nil with error.

Example:

#crud.select('customers', {{'<=', 'age', 35}})
---
- 1
...
crud.truncate('customers', {timeout = 2})
---
- true
...
#crud.select('customers', {{'<=', 'age', 35}})
---
- 0
...

-- Calculates the number of tuples in the space for memtx engine
-- Calculates the maximum approximate number of tuples in the space for vinyl engine
local result, err = crud.len(space_name, opts)

where:

• space_name (string|number) - name of the space as well as numerical id of the space
• opts:
  • timeout (?number) - vshard.call timeout (in seconds)

Returns number or nil with error.

Example:

Using memtx:

#crud.select('customers')
---
- 5
...
crud.len('customers', {timeout = 2})
---
- 5
...

Using vinyl:

crud.len('customers')
---
- 0
...
crud.delete('customers', 1)
---
...
crud.len('customers')
---
- 1
...

CRUD supports multi-conditional count, treating a cluster as a single space. The same as with select() the conditions may include field names or numbers, as well as index names. The recommended first condition is a TREE index; this helps to reduce the number of tuples to scan. Otherwise a full scan is performed. If compared with len(), count() method scans the entire space to count the tuples according user conditions. This method does yield that's why result may be approximate. Number of tuples before next yield() is under control with option yield_every.

local result, err = crud.count(space_name, conditions, opts)

where:

• space_name (string) - name of the space
• conditions (?table) - array of conditions
• opts:
  • yield_every (?number) - number of tuples processed to yield after, yield_every should be > 0, default value is 100
  • timeout (?number) - vshard.call timeout (in seconds), default value is 2
  • bucket_id (?number|cdata) - bucket ID
  • force_map_call (?boolean) - if true then the map call is performed without any optimizations even, default value is false
  • mode (?string, read or write) - if write is specified then count is performed on master, default value is read
  • prefer_replica (?boolean) - if true then the preferred target is one of the replicas, default value is false
  • balance (?boolean) - use replica according to vshard load balancing policy, default value is false

crud.count('customers', {{'<=', 'age', 35}})
---
- 5
...

Combinations of mode, prefer_replica and balance options lead to:

• mode == write - method performed on master with vshard call callrw
• mode == read
  • not prefer_replica, not balance - vshard call callro
  • not prefer_replica, balance - vshard call callbro
  • prefer_replica, not balance - vshard call callre
  • prefer_replica, balance - vshard call callbre

crud routers can provide statistics on called operations.

-- Enable statistics collect.
crud.cfg{ stats = true }

-- Returns table with statistics information.
crud.stats()

-- Returns table with statistics information for specific space.
crud.stats('my_space')

-- Disables statistics collect and destroys all collectors.
crud.cfg{ stats = false }

-- Destroys all statistics collectors and creates them again.
crud.reset_stats()

If metrics 0.10.0 or greater found, metrics collectors will be used by default to store statistics instead of local collectors. Quantiles in metrics summary collections are disabled by default. You can manually choose driver and enable quantiles.

-- Use simple local collectors (default if no required metrics version found).
crud.cfg{ stats = true, stats_driver = 'local' }

-- Use metrics collectors (default if metrics rock found).
crud.cfg{ stats = true, stats_driver = 'metrics' }

-- Use metrics collectors with 0.99 quantiles.
crud.cfg{ stats = true, stats_driver = 'metrics', stats_quantiles = true }

You can use crud.cfg to check current stats state.

crud.cfg
---
- stats_quantiles: true
  stats: true
  stats_driver: metrics
...

Performance overhead is 3-10% in case of local driver and 5-15% in case of metrics driver, up to 20% for metrics with quantiles.

Beware that iterating through crud.cfg with pairs is not supported yet, refer to tarantool/crud#265.

Format is as follows.

crud.stats()
---
- spaces:
    my_space:
      insert:
        ok:
          latency: 0.002
          count: 19800
          time: 39.6
        error:
          latency: 0.000001
          count: 4
          time: 0.000004
...
crud.stats('my_space')
---
- insert:
    ok:
      latency: 0.002
      count: 19800
      time: 39.6
    error:
      latency: 0.000001
      count: 4
      time: 0.000004
...

spaces section contains statistics for each observed space. If operation has never been called for a space, the corresponding field will be empty. If no requests has been called for a space, it will not be represented. Space data is based on client requests rather than storages schema, so requests for non-existing spaces are also collected.

Possible statistics operation labels are insert (for insert and insert_object calls), get, replace (for replace and replace_object calls), update, upsert (for upsert and upsert_object calls), delete, select (for select and pairs calls), truncate, len, count and borders (for min and max calls).

Each operation section consists of different collectors for success calls and error (both error throw and nil, err) returns. count is the total requests count since instance start or stats restart. latency is the 0.99 quantile of request execution time if metrics driver used and quantiles enabled, otherwise latency is the total average. time is the total time of requests execution.

In metrics registry statistics are stored as tnt_crud_stats metrics with operation, status and name labels.

metrics:collect()
---
- - label_pairs:
      status: ok
      operation: insert
      name: customers
    value: 221411
    metric_name: tnt_crud_stats_count
  - label_pairs:
      status: ok
      operation: insert
      name: customers
    value: 10.49834896344692
    metric_name: tnt_crud_stats_sum
  - label_pairs:
      status: ok
      operation: insert
      name: customers
      quantile: 0.99
    value: 0.00023606420935973
    metric_name: tnt_crud_stats
...

If you see -Inf value in quantile metrics, try to decrease the tolerated error:

crud.cfg{stats_quantile_tolerated_error = 1e-4}

See tarantool/metrics#189 for details about the issue.

select section additionally contains details collectors.

crud.stats('my_space').select.details
---
- map_reduces: 4
  tuples_fetched: 10500
  tuples_lookup: 238000
...

map_reduces is the count of planned map reduces (including those not executed successfully). tuples_fetched is the count of tuples fetched from storages during execution, tuples_lookup is the count of tuples looked up on storages while collecting responses for calls (including scrolls for multibatch requests). Details data is updated as part of the request process, so you may get new details before select/pairs call is finished and observed with count, latency and time collectors. In metrics registry they are stored as tnt_crud_map_reduces, tnt_crud_tuples_fetched and tnt_crud_tuples_lookup metrics with { operation = 'select', name = space_name } labels.

Since pairs request behavior differs from any other crud request, its statistics collection also has specific behavior. Statistics (select section) are updated after pairs cycle is finished: you either have iterated through all records or an error was thrown. If your pairs cycle was interrupted with break, statistics will be collected when pairs objects are cleaned up with Lua garbage collector.

Statistics are preserved between package reloads. Statistics are preserved between Tarantool Cartridge role reloads if you use CRUD Cartridge roles. Beware that metrics 0.12.0 and below do not support preserving stats between role reload (see tarantool/metrics#334), thus this feature will be unsupported for metrics driver.

cartridge.roles.crud-storage is a Tarantool Cartridge role that depends on the vshard-storage role, but also initializes functions that are used on the storage side to perform CRUD operations.

cartridge.roles.crud-router is a role that depends on the vshard-router role, but also exposes public crud functions in the global scope, so that you can call them via net.box.

1. Add crud to dependencies in the project rockspec.

Note: it's better to use tagged version than scm-1. Check the latest available release tag and use it.

-- <project-name>-scm-1.rockspec
dependencies = {
    ...
    'crud == <the-latest-tag>-1',
    ...
}

2. Create the role that stores your data and depends on crud-storage.

-- app.roles.customers-storage.lua
local cartridge = require('cartridge')

return {
        role_name = 'customers-storage',
        init = function()
            local customers_space = box.schema.space.create('customers', {
                format = {
                    {name = 'id', type = 'unsigned'},
                    {name = 'bucket_id', type = 'unsigned'},
                    {name = 'name', type = 'string'},
                    {name = 'age', type = 'number'},
                },
                if_not_exists = true,
            })
            customers_space:create_index('id', {
                parts = { {field ='id', is_nullable = false} },
                if_not_exists = true,
            })
            customers_space:create_index('bucket_id', {
                parts = { {field ='bucket_id', is_nullable = false} },
                if_not_exists = true,
            })
            customers_space:create_index('age', {
                parts = { {field ='age'} },
                unique = false,
                if_not_exists = true,
            })
        end,
        dependencies = {'cartridge.roles.crud-storage'},
    }

-- app.roles.customers-router.lua
local cartridge = require('cartridge')
return {
        role_name = 'customers-router',
        dependencies = {'cartridge.roles.crud-router'},
    }

3. Start the application and create customers-storage and customers-router replica sets.

4. Don't forget to bootstrap vshard.

Now your cluster contains storages that are configured to be used for CRUD-operations. You can simply call CRUD functions on the router to insert, select, and update data across the cluster.

BSD-2-Clause. See the LICENSE file.