summaryrefslogtreecommitdiffstats
path: root/docs/guides/monitor-hadoop-cluster.md
blob: b536e0fa0389747acd94820d38fbe8450ee3ed2d (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
<!--
title: "Monitor a Hadoop cluster with Netdata"
sidebar_label: "Monitor a Hadoop cluster with Netdata"
custom_edit_url: https://github.com/netdata/netdata/edit/master/docs/guides/monitor-hadoop-cluster.md
learn_status: "Published"
learn_topic_type: "Tasks"
learn_rel_path: "Miscellaneous"
-->

# Monitor a Hadoop cluster with Netdata

Hadoop is an [Apache project](https://hadoop.apache.org/) is a framework for processing large sets of data across a
distributed cluster of systems.

And while Hadoop is designed to be a highly-available and fault-tolerant service, those who operate a Hadoop cluster
will want to monitor the health and performance of their [Hadoop Distributed File System
(HDFS)](https://hadoop.apache.org/docs/r1.2.1/hdfs_design.html) and [Zookeeper](https://zookeeper.apache.org/)
implementations.

Netdata comes with built-in and pre-configured support for monitoring both HDFS and Zookeeper.

This guide assumes you have a Hadoop cluster, with HDFS and Zookeeper, running already. If you don't, please follow
the [official Hadoop
instructions](http://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-common/SingleCluster.html) or an
alternative, like the guide available from
[DigitalOcean](https://www.digitalocean.com/community/tutorials/how-to-install-hadoop-in-stand-alone-mode-on-ubuntu-18-04).

For more specifics on the collection modules used in this guide, read the respective pages in our documentation:

-   [HDFS](https://github.com/netdata/netdata/blob/master/src/go/collectors/go.d.plugin/modules/hdfs/README.md)
-   [Zookeeper](https://github.com/netdata/netdata/blob/master/src/go/collectors/go.d.plugin/modules/zookeeper/README.md)

## Set up your HDFS and Zookeeper installations

As with all data sources, Netdata can auto-detect HDFS and Zookeeper nodes if you installed them using the standard
installation procedure.

For Netdata to collect HDFS metrics, it needs to be able to access the node's `/jmx` endpoint. You can test whether an
JMX endpoint is accessible by using `curl HDFS-IP:PORT/jmx`. For a NameNode, you should see output similar to the
following:

```json
{
  "beans" : [ {
    "name" : "Hadoop:service=NameNode,name=JvmMetrics",
    "modelerType" : "JvmMetrics",
    "MemNonHeapUsedM" : 65.67851,
    "MemNonHeapCommittedM" : 67.3125,
    "MemNonHeapMaxM" : -1.0,
    "MemHeapUsedM" : 154.46341,
    "MemHeapCommittedM" : 215.0,
    "MemHeapMaxM" : 843.0,
    "MemMaxM" : 843.0,
    "GcCount" : 15,
    "GcTimeMillis" : 305,
    "GcNumWarnThresholdExceeded" : 0,
    "GcNumInfoThresholdExceeded" : 0,
    "GcTotalExtraSleepTime" : 92,
    "ThreadsNew" : 0,
    "ThreadsRunnable" : 6,
    "ThreadsBlocked" : 0,
    "ThreadsWaiting" : 7,
    "ThreadsTimedWaiting" : 34,
    "ThreadsTerminated" : 0,
    "LogFatal" : 0,
    "LogError" : 0,
    "LogWarn" : 2,
    "LogInfo" : 348
  }, 
  { ... }
  ]
}
```

The JSON result for a DataNode's `/jmx` endpoint is slightly different:

```json
{
  "beans" : [ {
    "name" : "Hadoop:service=DataNode,name=DataNodeActivity-dev-slave-01.dev.local-9866",
    "modelerType" : "DataNodeActivity-dev-slave-01.dev.local-9866",
    "tag.SessionId" : null,
    "tag.Context" : "dfs",
    "tag.Hostname" : "dev-slave-01.dev.local",
    "BytesWritten" : 500960407,
    "TotalWriteTime" : 463,
    "BytesRead" : 80689178,
    "TotalReadTime" : 41203,
    "BlocksWritten" : 16,
    "BlocksRead" : 16,
    "BlocksReplicated" : 4,
    ...
  },
  { ... }
  ]
}
```

If Netdata can't access the `/jmx` endpoint for either a NameNode or DataNode, it will not be able to auto-detect and
collect metrics from your HDFS implementation.

Zookeeper auto-detection relies on an accessible client port and a allow-listed `mntr` command. For more details on
`mntr`, see Zookeeper's documentation on [cluster
options](https://zookeeper.apache.org/doc/current/zookeeperAdmin.html#sc_clusterOptions) and [Zookeeper
commands](https://zookeeper.apache.org/doc/current/zookeeperAdmin.html#sc_zkCommands).

## Configure the HDFS and Zookeeper modules

To configure Netdata's HDFS module, navigate to your Netdata directory (typically at `/etc/netdata/`) and use
`edit-config` to initialize and edit your HDFS configuration file.

```bash
cd /etc/netdata/
sudo ./edit-config go.d/hdfs.conf
```

At the bottom of the file, you will see two example jobs, both of which are commented out:

```yaml
# [ JOBS ]
#jobs:
#  - name: namenode
#    url: http://127.0.0.1:9870/jmx
#
#  - name: datanode
#    url: http://127.0.0.1:9864/jmx
```

Uncomment these lines and edit the `url` value(s) according to your setup. Now's the time to add any other configuration
details, which you can find inside of the `hdfs.conf` file itself. Most production implementations will require TLS
certificates.

The result for a simple HDFS setup, running entirely on `localhost` and without certificate authentication, might look
like this:

```yaml
# [ JOBS ]
jobs:
  - name: namenode
    url: http://127.0.0.1:9870/jmx

  - name: datanode
    url: http://127.0.0.1:9864/jmx
```

At this point, Netdata should be configured to collect metrics from your HDFS servers. Let's move on to Zookeeper.

Next, use `edit-config` again to initialize/edit your `zookeeper.conf` file.

```bash
cd /etc/netdata/
sudo ./edit-config go.d/zookeeper.conf
```

As with the `hdfs.conf` file, head to the bottom, uncomment the example jobs, and tweak the `address` values according
to your setup. Again, you may need to add additional configuration options, like TLS certificates.

```yaml
jobs:
  - name    : local
    address : 127.0.0.1:2181

  - name    : remote
    address : 203.0.113.10:2182
```

Finally, [restart Netdata](https://github.com/netdata/netdata/blob/master/packaging/installer/README.md#maintaining-a-netdata-agent-installation).

```sh
sudo systemctl restart netdata
```

Upon restart, Netdata should recognize your HDFS/Zookeeper servers, enable the HDFS and Zookeeper modules, and begin
showing real-time metrics for both in your Netdata dashboard. 🎉

## Configuring HDFS and Zookeeper alerts

The Netdata community helped us create sane defaults for alerts related to both HDFS and Zookeeper. You may want to
investigate these to ensure they work well with your Hadoop implementation.

-   [HDFS alerts](https://raw.githubusercontent.com/netdata/netdata/master/src/health/health.d/hdfs.conf)

You can also access/edit these files directly with `edit-config`:

```bash
sudo /etc/netdata/edit-config health.d/hdfs.conf
sudo /etc/netdata/edit-config health.d/zookeeper.conf
```

For more information about editing the defaults or writing new alert entities, see our 
[health monitoring documentation](https://github.com/netdata/netdata/blob/master/src/health/README.md).