Introduce Nodes APIs

In this commit we are introducing a new section in REST API Reference for Nodes APIs.

We are adding documentation for "Nodes hot threads" and "Nodes info". Because Nodes APIs share common request parameters we also add more detailed section into parent index page (notice that the nodes-filter parameter was moved from "Cluster stats" page and was improved, and relevant links were updated).

There are more Nodes APIs that still need to be documented, this is just a good start.

We updated nav_order parameter for some pages as well.

Relevant ticket: #424

Signed-off-by: Lukáš Vlček <lukas.vlcek@aiven.io>
This commit is contained in:
Lukáš Vlček 2022-04-22 18:39:12 +02:00
parent cb02f718cf
commit 2df0f8450b
5 changed files with 306 additions and 2 deletions

View File

@ -31,7 +31,7 @@ All cluster stats parameters are optional.
Parameter | Type | Description
:--- | :--- | :---
&lt;node-filters&gt; | List | A comma-separated list of node-filters that OpenSearch uses to filter results. Available options are `all`, `_local`, `_master`, a node name or ID, `master:true`, `master:false`, `data:true`, `data:false`, `ingest:true`, `ingest:false`, `voting_only:true`, `voting_only:false`, `ml:true`, `ml:false`, `coordinating_only:true`, `coordinating_only:false`, and &lt;custom node attributes&gt; : &lt;attribute values&gt; pairs.
&lt;node-filters&gt; | List | A comma-separated list of [node-filters](../nodes-apis/index/#node-filters) that OpenSearch uses to filter results.
## Response

View File

@ -3,7 +3,7 @@ layout: default
title: Ingest APIs
parent: REST API reference
has_children: true
nav_order: 3
nav_order: 4
redirect_from:
- /opensearch/rest-api/ingest-apis/
---

View File

@ -0,0 +1,72 @@
---
layout: default
title: Nodes APIs
parent: REST API reference
has_children: true
nav_order: 5
---
# Nodes APIs
The nodes API makes it possible to retrieve information about individual cluster nodes.
---
Many nodes APIs support common parameters like `{timeout}`, and `{node-filters}`.
## Timeout
The `{timeout}` parameter can be used to change the default time limit for node response.
Parameter | Type | Description
:--- |:----------| :---
`{timeout}` | TimeValue | Default value is `30s`.
## Node filters
The `{node-filters}` parameter can be used to filter target set of nodes that will be included in the response.
Parameter | Type | Description
:--- |:-------| :---
<code><nobr>{node-filters} | String | A comma-separated list of resolution mechanisms that OpenSearch uses to identify cluster nodes.
Node filters support several node resolution mechanisms:
- pre-defined constants: `_local`, `_cluster_manager` (or deprecated `_master`) or `_all`
- exact match for `nodeID`
- a simple case-sensitive wildcard pattern matching for: `node-name`, `host-name` or `host-IP-address`
- node roles (where `<bool>` value is set either to `true` or `false`):
- `cluster_manager:<bool>` (or deprecated `master:<bool>`)
- `data:<bool>`
- `ingest:<bool>`
- `voting_only:<bool>`
- `ml:<bool>`
- `coordinating_only:<bool>`
- a simple case-sensitive wildcard pattern matching for node attributes: <br>`<node attribute*>:<attribute value*>` (the wildcard matching pattern can be used in both the key and value at the same time)
The resolution mechanisms are applied sequentially in the order specified by the client and each mechanism specification may either add or remove nodes.
### Example
Get statistics from elected cluster-manager node only:
```text
GET /_nodes/_cluster_manager/stats
```
Get statistics from nodes that are data-only nodes:
```text
GET /_nodes/data:true/stats
```
#### Order of resolution mechanisms matters
The order of resolution mechanisms is applied sequentially and each can add or remove nodes, this means that the following two examples yield different results.
Get statistics from all the nodes but the cluster-manager node:
```text
GET /_nodes/_all,cluster_namager:false/stats
```
However, if we switch the resolution mechanisms then the result will include all the cluster nodes including the cluster manager node.
```text
GET /_nodes/cluster_namager:false,_all/stats
```

View File

@ -0,0 +1,101 @@
---
layout: default
title: Nodes hot threads
parent: Nodes APIs
grand_parent: REST API reference
nav_order: 10
---
# Nodes hot threads
This REST API provide information about busy JVM threads for selected cluster nodes.
It provides a unique view of what activity each node spends time on.
## Example
```json
GET /_nodes/hot_threads
```
## Path and HTTP methods
```text
GET /_nodes/hot_threads
GET /_nodes/{nodeId}/hot_threads
```
## URL parameters
You can include the following URL parameters in your request. All parameters are optional.
Parameter | Type | Description
:--- |:----------| :---
nodeId | String | A comma-separated list of nodeIds to filter results. Supports [node filters](../index/#node-filters).<br>Defaults to `_all`.
snapshots | Integer | Number of samples of thread stacktraces.<br>Defaults to `10`.
interval | TimeValue | Interval between consecutive samples.<br>Defaults to `500ms`.
threads | Integer | A number of top bussiest threads to return information about. Defaults to `3`.
ignore_idle_threads | Boolean | Dont show threads that are in known-idle states, such as waiting on a socket select or pulling from an empty task queue.<br>Defaults to `true`.
type | String | Supported thread types are `cpu`, `wait`, or `block`.<br>Defaults to `cpu`.
timeout | TimeValue | A request [timeout](../index/#timeout).<br>Defaults to `30s`.
## Response
Unlike majority of OpenSearch responses this response is in text format.
It consists of one section per each cluster node included in the response.
Each section starts with a single line containing the following segments:
Line segment | Description
:--- |:-------
<code>:::&nbsp;</code> | Line start (a distinct visual symbol).
`{global-eu-35}` | Node name.
`{uFPbKLDOTlOmdnwUlKW8sw}` | NodeId.
`{OAM8OT5CQAyasWuIDeVyUA}` | EphemeralId.
`{global-eu-35.local}` | Host name.
`{[gdv2:a284:2acv:5fa6:0:3a2:7260:74cf]:9300}` | Host address.
`{dimr}` | Node roles (d=data, i=ingest, m=cluster&nbsp;manager, r=remote&nbsp;cluster&nbsp;client).
`{zone=west-a2, shard_indexing_pressure_enabled=true}` | Node attributes.
Then follows information about threads of selected type.
```text
::: {global-eu-35}{uFPbKLDOTlOmdnwUlKW8sw}{OAM8OT5CQAyasWuIDeVyUA}{global-eu-35.local}{[gdv2:a284:2acv:5fa6:0:3a2:7260:74cf]:9300}{dimr}{zone=west-a2, shard_indexing_pressure_enabled=true}
Hot threads at 2022-04-01T15:15:27.658Z, interval=500ms, busiestThreads=3, ignoreIdleThreads=true:
0.1% (645micros out of 500ms) cpu usage by thread 'opensearch[global-eu-35][transport_worker][T#7]'
4/10 snapshots sharing following 3 elements
io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:986)
io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)
java.base@11.0.14.1/java.lang.Thread.run(Thread.java:829)
::: {global-eu-62}{4knOxAdERlOB19zLQIT1bQ}{HJuZs2HiQ_-8Elj0Fvi_1g}{global-eu-62.local}{[gdv2:a284:2acv:5fa6:0:3a2:bba6:fe3f]:9300}{dimr}{zone=west-a2, shard_indexing_pressure_enabled=true}
Hot threads at 2022-04-01T15:15:27.659Z, interval=500ms, busiestThreads=3, ignoreIdleThreads=true:
18.7% (93.4ms out of 500ms) cpu usage by thread 'opensearch[global-eu-62][transport_worker][T#3]'
6/10 snapshots sharing following 3 elements
io.netty.util.concurrent.SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:986)
io.netty.util.internal.ThreadExecutorMap$2.run(ThreadExecutorMap.java:74)
java.base@11.0.14.1/java.lang.Thread.run(Thread.java:829)
::: {global-eu-44}{8WW3hrkcTwGvgah_L8D_jw}{Sok7spHISFyol0jFV6i0kw}{global-eu-44.local}{[gdv2:a284:2acv:5fa6:0:3a2:9120:e79e]:9300}{dimr}{zone=west-a2, shard_indexing_pressure_enabled=true}
Hot threads at 2022-04-01T15:15:27.659Z, interval=500ms, busiestThreads=3, ignoreIdleThreads=true:
42.6% (212.7ms out of 500ms) cpu usage by thread 'opensearch[global-eu-44][write][T#5]'
2/10 snapshots sharing following 43 elements
java.base@11.0.14.1/sun.nio.ch.IOUtil.write1(Native Method)
java.base@11.0.14.1/sun.nio.ch.EPollSelectorImpl.wakeup(EPollSelectorImpl.java:254)
io.netty.channel.nio.NioEventLoop.wakeup(NioEventLoop.java:787)
io.netty.util.concurrent.SingleThreadEventExecutor.execute(SingleThreadEventExecutor.java:846)
io.netty.util.concurrent.SingleThreadEventExecutor.execute(SingleThreadEventExecutor.java:815)
io.netty.channel.AbstractChannelHandlerContext.safeExecute(AbstractChannelHandlerContext.java:989)
io.netty.channel.AbstractChannelHandlerContext.write(AbstractChannelHandlerContext.java:796)
io.netty.channel.AbstractChannelHandlerContext.writeAndFlush(AbstractChannelHandlerContext.java:758)
io.netty.channel.DefaultChannelPipeline.writeAndFlush(DefaultChannelPipeline.java:1020)
io.netty.channel.AbstractChannel.writeAndFlush(AbstractChannel.java:311)
org.opensearch.transport.netty4.Netty4TcpChannel.sendMessage(Netty4TcpChannel.java:159)
app//org.opensearch.transport.OutboundHan...
```
## Required permissions
If you use the security plugin, make sure you have the appropriate permissions:
`cluster:monitor/nodes/hot_threads`
{: .note }

View File

@ -0,0 +1,131 @@
---
layout: default
title: Nodes info
parent: Nodes APIs
grand_parent: REST API reference
nav_order: 20
---
# Nodes info
Represents mostly static information about cluster nodes.
Such as host system information, JVM, processor type, or specific
node information like node settings, thread pools settings, installed plugins, and more.
## Example
```json
# Get information from all cluster nodes
GET /_nodes
# Get thread pool information from the cluster manager node only
GET /_nodes/master:true/thread_pool?pretty
```
## Path and HTTP methods
```text
GET /_nodes
GET /_nodes/{nodeId}
GET /_nodes/{metrics}
GET /_nodes/{nodeId}/{metrics}
# or full path equivalent
GET /_nodes/{nodeId}/info/{metrics}
```
## URL parameters
You can include the following URL parameters in your request. All parameters are optional.
Parameter | Type | Description
:--- |:-------| :---
nodeId | String | A comma-separated list of nodeIds to filter results. Supports [node filters](../index/#node-filters).<br>Defaults to `_all`.
metrics | String | A comma-separated list of metric groups that will be included in the response. For example `jvm,thread_pools`.<br>Defaults to all metrics.
timeout | TimeValue | A request [timeout](../index/#timeout).<br>Defaults to `30s`.
The following are listed all available metric groups:
Metric | Description
:--- |:----
`settings` | A node settings. This is combination of the default settings, custom settings from [configuration file](../../../configuration/#configuration-file) and dynamically [updated settings](../../../configuration/#update-cluster-settings-using-the-api).
`os` | Static information about host os, including version, processor architecture and available/allocated processors.
`process` | Contains process id.
`jvm` | Detailed static information about running JVM, including arguments.
`thread_pool` | Configured options for all individual thread pools (type, size ... etc).
`transport` | Mostly static information about transport layer.
`http` | Mostly static information about http layer.
`plugins` | Information about installed plugins and modules.
`ingest` | Information about ingets pipelines and available ingest processors.
`aggregations` | Information about available [aggregations](../../../aggregations).
`indices` | Static index settings configured at the node level.
## Response
The response contains basic node identification and build info for every node
matching the `{nodeId}` request parameter.
Metric | Description
:--- |:----
name | A node name.
transport_address | A node transport address.
host | A node host address.
ip | A node host ip address.
version | A node OpenSearch version.
build_type | A build type, like `rpm`,`docker`, `zip` ... etc.
build_hash | A git commit hash of the build.
roles | A node roles.
attributes | A node attributes.
On top of that it will contain one or more metric groups depending on `{metrics}` request parameter.
```json
GET /_nodes/master:true/process,transport?pretty
{
"_nodes": {
"total": 1,
"successful": 1,
"failed": 0
},
"cluster_name": "opensearch",
"nodes": {
"VC0d4RgbTM6kLDwuud2XZQ": {
"name": "node-m1-23",
"transport_address": "127.0.0.1:9300",
"host": "127.0.0.1",
"ip": "127.0.0.1",
"version": "1.3.1",
"build_type": "tar",
"build_hash": "c4c0672877bf0f787ca857c7c37b775967f93d81",
"roles": [
"data",
"ingest",
"master",
"remote_cluster_client"
],
"attributes": {
"shard_indexing_pressure_enabled": "true"
},
"process" : {
"refresh_interval_in_millis": 1000,
"id": 44584,
"mlockall": false
},
"transport": {
"bound_address": [
"[::1]:9300",
"127.0.0.1:9300"
],
"publish_address": "127.0.0.1:9300",
"profiles": { }
}
}
}
}
```
## Required permissions
If you use the security plugin, make sure you have the appropriate permissions:
`cluster:monitor/nodes/info`
{: .note }