Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -18,17 +18,20 @@ If you have a small data set that can fit in memory, then you should consider us
[.edition]#{enterprise}#

Magma is designed for high performance with very large datasets that do not fit in memory.
It is ideal for use cases that rely primarily on disk access.
It's ideal for use cases that rely primarily on disk access.
The performance of disk access will be as good as the underlying disk sub-systems --
for example, using NVMe SSDs will give higher performance.

In order to get maximum performance from Magma for disk-oriented workloads, it is recommended to set the Writer Threads to `Disk i/o optimized`. This setting will ensure there are enough threads to sustain high write rates.

To learn more about Writer Thread settings, see xref:manage:manage-settings/general-settings.adoc#data-settings[Data Settings]

Magma can work with very low amounts of memory for large datasets: a minimum memory-to-data ratio of 1% is required.
For example, if a node is holding 5{nbsp}TB of data, Magma can be used with only 64{nbsp}GB RAM.

Magma has several thread settings to tune performance for different workloads.
To get maximum performance from Magma for disk-oriented workloads, set the Writer Threads to `Disk i/o optimized`. This setting allocates more threads to sustain high write rates.
See xref:manage:manage-settings/general-settings.adoc#data-settings[Data Settings] for more information about this setting.

You can also tune the number and allocation of threads that Magma uses to compact and flush data to disk.
You may choose to change the ratio of threads that compact data to the threads that flush data if you notice that Magma's data compaction spikes CPU use.


.Magma Supported Services
|===
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,11 @@ For illustrations of how Couchbase Server saves new and updates existing Couchba
[#threading]
== Threading

Couchbase Server uses multiple threads when reading an writing data.
Copy link

Copilot AI Jan 27, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Corrected 'an' to 'and' before 'writing'.

Suggested change
Couchbase Server uses multiple threads when reading an writing data.
Couchbase Server uses multiple threads when reading and writing data.

Copilot uses AI. Check for mistakes.
It offers several configuration settings you can change to optimize performance for your workload and hardware.

=== Reader and Writer Threads

Synchronized, multithreaded _readers_ and _writers_ provide simultaneous, high-performance operations for data on disk.
Conflicts are avoided by assigning each thread (reader or writer) a specific subset of the 1024 vBuckets for each Couchbase bucket.

Expand Down Expand Up @@ -59,7 +64,35 @@ Again, the maximum thread-allocation that can be specified for each is _64_, the
Thread-status can be viewed, by means of the [.cmd]`cbstats` command, specified with the [.param]`raw workload` option.
See xref:cli:cbstats-intro.adoc[cbstats] for information.

For information on using the REST API to manage thread counts, see xref:rest-api:rest-reader-writer-thread-config.adoc[Setting Thread Allocations].
For information about using the REST API to manage thread counts, see xref:rest-api:rest-reader-writer-thread-config.adoc[Setting Thread Allocations].

[#magma-flushing-and-compaction-threads]
=== Magma Flushing and Compaction Threads

Couchbase Server compacts the data it writes to disk for xref:learn:buckets-memory-and-storage/storage-engines.adoc#storage-engine-magma[Magma] buckets.
It allocates a thread pool (containing 20 threads by default) for background compaction and flushing operations for these buckets.
You can change the number of threads in this pool using the `num_storage_threads` setting of the xref:rest-api:rest-reader-writer-thread-config.adoc[thread allocation REST API].

Two types of threads share the Magma thread pool: compactor threads that compact data and flusher threads that write data to disk.
By default, Couchbase Server allocates 20% of the threads to flushing data and 80% to compacting data.
With the default thread pool size and the default flusher allocation, Couchbase Server uses 4 threads to flush data and 16 threads to compact data for Magma buckets.
You can also change the ratio of flusher to compactor threads using the `magma_flusher_thread_percentage` setting of the xref:rest-api:rest-reader-writer-thread-config.adoc[thread allocation REST API]

For most workloads, the default thread pool size and flusher allocation percentage work well.
If you notice CPU use spikes during heavy data mutation workloads, you may want to investigate whether the compactor threads are the cause.
You can monitor the compaction and flushing activity using the `kv_magma_compactions` metric.
This metric counts the number of compactions Couchbase Server has performed.
You can view this metric via the xref:rest-api:rest-statistics.adoc[] REST API or through Prometheus if you have configured it to collect Couchbase Server metrics.
See xref:manage:monitor/set-up-prometheus-for-monitoring.adoc[] for more information about using Prometheus with Couchbase Server.

If you see the CPU use of the `memcached` Linux process on your nodes spike while the `kv_magma_compactions` count is increasing, the compactor threads may be the cause of the spike.
In this case, you may want to reduce the number of compactor threads by increasing the percentage of flusher threads.
This reduction limits the compactor's ability to spike CPU use.
However, reducing the number of compactor threads may lead to higher latency before Couchbase Server writes data to disk.

NOTE: The number of threads in the compactor and flusher pool and the percentages of flusher threads are advanced settings.
Contact Couchbase Support before making changes to them.
Support can help you determine the best settings for your workload and hardware.

[#deletion]
== Deletion
Expand Down
221 changes: 188 additions & 33 deletions modules/rest-api/pages/rest-reader-writer-thread-config.adoc
Original file line number Diff line number Diff line change
@@ -1,82 +1,237 @@
= Setting Thread Allocations
:description: The number of threads allocated per node is configurable.
:description: Couchbase Server has several settings that let you change how it allocates and uses threads for the entire cluster.
:page-topic-type: reference

[abstract]
{description}

== HTTP method and URI

.Get Global Thread Settings
[source,url]
----
GET /pools/default/settings/memcached/global
----

.Set Global Thread Allocations
[source]
----
POST /pools/default/settings/memcached/global
----

[#description]
== Description

Couchbase Server allows the number of threads allocated per node for _reading_, _writing_, and for _NonIO_, and _AuxIO_ thread-pools to be configured by the administrator.
Couchbase Server lets you set the number of threads per node for reading, writing, NonIO, and AuxIO thread pools.
You can also change the number of threads that Magma buckets use for writing data to disk.
Increasing thread allocation can improve performance on systems with a large number of CPU cores.
For example, more writer threads can optimize durable write performance.
See xref:learn:data/durability.adoc[Durability] for details.
However, allocating too many threads can reduce performance on nodes with limited resources.
Test any changes to thread allocation before applying them to production systems.


== Getting Current Thread Settings

To get the current global thread settings for Couchbase Server, use the `GET /pools/default/settings/memcached/global` endpoint.

[#get-curl-syntax]
=== curl Syntax

[source,bash]
----
curl -X GET http[s]://{HOST}:{PORT}/pools/default/settings/memcached/global
-u $USER:$PASSWORD
----


==== Path Parameters
:required-privileges: required-privileges-get
include::partial$user_pwd_host_port_params.adoc[]

[#required-privileges-get]
=== Required Privileges

Your account must have at least 1 of the following roles to make a GET request from this endpoint:

* xref:learn:security/roles.adoc#full-admin[Full Admin]
* xref:learn:security/roles.adoc#cluster-admin[Cluster Admin]
* xref:learn:security/roles.adoc#external-user-security-admin[External User Security Admin]
* xref:learn:security/roles.adoc#local-user-security-admin[Local User Security Admin]

=== Responses

200 OK::
The request was successful.
Returns a JSON object containing the current global thread settings that have been set explicitly.

400 Bad Request::
The request could not be understood.
Also returns a JSON object containing an error message.

=== Example

The following example shows how to get the current global thread settings for Couchbase Server:

[source,bash]
----
curl -X GET -u Administrator:password \
http://localhost:8091/pools/default/settings/memcached/global | jq
----

If successful, the call returns a JSON object containing the any overrides to the default global thread settings:

[source,json]
----
{
"magma_flusher_thread_percentage": 30,
"num_storage_threads": 25
}
----

A high thread-allocation may improve performance on systems whose hardware-resources are commensurately supportive (for example, where the number of CPU cores is high).
In particular, a high number of _writer_ threads on such systems may significantly optimize the performance of _durable writes_: see xref:learn:data/durability.adoc[Durability], for information.
== Setting Global Thread Allocations

Note, however, that a high thread-allocation might _impair_ some aspects of performance on less appropriately resourced nodes.
Consequently, changes to the default thread-allocation should not be made to production systems without prior testing.
To set the global thread allocations for Couchbase Server, perform a `POST` ro the `/pools/default/settings/memcached/global` endpoint.

[#curl-syntax]
== Curl Syntax
=== curl Syntax

[source,bash]
----
curl -X POST -d hostname=<ip-address-or-hostname>:8091
-d num_reader_threads=<int>
-d num_writer_threads=<int>
-d num_nonio_threads=<int>
-d num_auxio_threads=<int>
-d password=<password>
-u <administrator>:<password>
http://<host>:<port>/pools/default/settings/memcached/global
curl -X POST http[s]://{HOST}:{PORT}/pools/default/settings/memcached/global
[-d num_reader_threads=<integer>]
[-d num_writer_threads=<integer>]
[-d num_nonio_threads=<integer>]
[-d num_auxio_threads=<integer>]
[-d num_storage_threads=<integer>]
[-d magma_flusher_thread_percentage=<integer>]
-u $USER:$PASSWORD
----

Each key is specified optionally.


==== Path Parameters
:required-privileges: required-privileges-put
include::partial$user_pwd_host_port_params.adoc[]

==== Parameters

`num_reader_threads`::
(optional) Sets the number threads Couchbase Server uses to read data.

`num_writer_threads`::
(optional) Sets the number threads Couchbase Server uses to write data.

`num_nonio_threads`::
(optional) Sets the number of NonIO threads to be used by Couchbase Server.
The NonIO thread pool runs in-memory tasks.
For example, the durability timeout task uses these threads.
Valid values are:

The keys `num_reader_threads` and `num_writer_threads` specify the number of threads that are to be used for _reading_ and _writing_, respectively.
+
* An integer from `1` to `64`: explicitly sets the number of threads.
* `default`: Couchbase Server calculates and applies an appropriate number of threads.

The keys `num_nonio_threads` and `num_auxio_threads` specify the number of threads that are to be used for _NonIO_ and _AuxIO_ thread pools, respectively.
Note that:
`num_auxio_threads`::
(optional) Sets the number of AuxIO threads to be used by Couchbase Server.
The AuxIO thread pool runs auxiliary I/O tasks -- for example, the access log task.
Valid values are:

* The _NonIO_ thread pool is used to run _in memory_ tasks -- for example, the _durability timeout_ task.
+
* An integer from `1` to `64`: explicitly sets the number of threads.
* `default`: Couchbase Server determines the number of threads.

* The _AuxIO_ thread pool is used to run _auxiliary I/O_ tasks -- for example, the _access log_ task.
`num_storage_threads`::
(optional) Sets the number of threads the Magma storage engine can use when compacting and writing data to disk.
This thread pool is divided into 2 types of threads: flusher threads and compactor threads.
See xref:learn:buckets-memory-and-storage/storage-settings.adoc#magma-flushing-and-compaction-threads[Magma Flushing and Compaction Threads] for more information.
The default value is `20`.

`magma_flusher_thread_percentage`::
(optional) Sets the percentage of Magma storage engine threads to allocate to flushing compacted data to disk.
The remaining threads in the thread pool are allocated to compacting data.
For example, if you set `num_storage_threads` to `20` and set `magma_flusher_thread_percentage` to `50`, Couchbase Server uses `10` threads to flush data and `10` threads to compact data.
Valid values for this setting are integers from `0` to `100`.
The default setting is `20` which allocates 20% of threads to flushing data.

NOTE: `num_storage_threads` and `magma_flusher_thread_percentage` are advanced settings.
Contact Couchbase Support before making changes to them.
Support can help you determine the best settings for your workload and hardware.

[#required-privileges-put]
=== Required Privileges

Your account must have at least 1 of the following roles to make a POST request to this endpoint:

* xref:learn:security/roles.adoc#full-admin[Full Admin]
* xref:learn:security/roles.adoc#cluster-admin[Cluster Admin]

In all cases, the value of `int` should be an integer between `1` and `64`, inclusive.
If in any case the value `default` is specified (for example, `-d num_reader_threads=default`), Couchbase Server itself calculates and applies an appropriate number of threads.

[#responses]
== Responses
=== Responses

202 Accepted::
The request was successful.
This result also returns a JSON object containing all non-default thread settings.

400 Bad Request::
The request had errors or invalid values.
This result also returns a JSON object containing an error message.

Success returns an object whose values confirm the settings that have been made.

[#examples]
== Examples
=== Examples

To set the numbers of reader, writer, NonIO, and AuxIO threads for Couchbase Server, use the `POST /pools/default/settings/memcached/global` http method and endpoint as follows:

[source,bash]
----
curl -v -X POST -u Administrator:password \
http://10.144.220.101:8091/pools/default/settings/memcached/global \
curl -X POST -u Administrator:password \
http://localhost:8091/pools/default/settings/memcached/global \
-d num_reader_threads=12 \
-d num_writer_threads=8 \
-d num_nonio_threads=6 \
-d num_auxio_threads=6
-d num_auxio_threads=6 | jq
----

This sets the number of `reader` threads to `12`, the number of `writer` threads to `8`, and the numbers of `NonIO` and `AuxIO` threads each to `6`.
If successful, the call returns an object that confirms the settings you specified:

[source,json]
----
{
"num_reader_threads": 12,
"num_writer_threads": 8,
"num_auxio_threads": 6,
"num_nonio_threads": 6
}
----

This sets the number of _reader_ threads to `12`, the number of _writer_ threads to `8`, and the numbers of _NonIO_ and _AuxIO_ threads each to `6`.
If successful, the call returns an object whose values confirm the settings that have been made:
The following example increases the Magma storage engine's thread pool to 30 threads and allocates 25% of the threads to flushing data to disk:

[source,bash]
----
{"num_reader_threads":12,"num_writer_threads":8,"num_auxio_threads":6,"num_nonio_threads":6}
curl -X POST -u Administrator:password \
http://localhost:8091/pools/default/settings/memcached/global \
-d num_storage_threads=30 \
-d magma_flusher_thread_percentage=25 | jq
----

If successful, the call returns an object confirming your new settings:

[source,json]
----
{
"magma_flusher_thread_percentage": 25,
"num_storage_threads": 30
}
----



[#see-also]
== See Also

See xref:learn:data/durability.adoc[Durability], for information on durable writes.
See xref:learn:buckets-memory-and-storage/storage-settings.adoc#threading[Threading] for an overview of threads.
* See xref:learn:data/durability.adoc[Durability], for information about durable writes.
* See xref:learn:buckets-memory-and-storage/storage-settings.adoc#threading[Threading] for an overview of threads.
15 changes: 15 additions & 0 deletions modules/rest-api/partials/user_pwd_host_port_params.adoc
Original file line number Diff line number Diff line change
@@ -0,0 +1,15 @@
// Be sure to set the 'required-priviledges' attribute before including this partial.
Copy link

Copilot AI Jan 27, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Corrected spelling of 'priviledges' to 'privileges'.

Copilot uses AI. Check for mistakes.
// It must be the anchor to the Required Privileges section for the specific endpoint.
`HOST`::
Host name or IP address of a Couchbase Server.

`PORT`::
Port number for the REST API.
Defaults are 8091 for unencrypted and 18901 for encrypted connections.

`USER`::

The name of a user who has at least 1 of the roles listed in xref:{required-priviledges}[Required Privileges].
Copy link

Copilot AI Jan 27, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Corrected spelling of 'priviledges' to 'privileges'.

Copilot uses AI. Check for mistakes.

`PASSWORD`::
The password for the user.