Exports hadoop metrics via HTTP for Prometheus consumption.
How to run
python hadoop_exporter.py
Help on flags of hadoop_exporter:
$ python hadoop_exporter.py -h
usage: hadoop_exporter.py [-h] [-c cluster_name] [-hdfs namenode_jmx_url]
[-rm resourcemanager_jmx_url] [-dn datanode_jmx_url]
[-jn journalnode_jmx_url] [-mr mapreduce2_jmx_url]
[-hbase hbase_jmx_url] [-hive hive_jmx_url]
[-p metrics_path] [-host ip_or_hostname] [-P port]
hadoop node exporter args, including url, metrics_path, address, port and
cluster.
optional arguments:
-h, --help show this help message and exit
-c cluster_name, --cluster cluster_name
Hadoop cluster labels. (default "cluster_indata")
-hdfs namenode_jmx_url, --namenode-url namenode_jmx_url
Hadoop hdfs metrics URL. (default
"http://indata-10-110-13-165.indata.com:50070/jmx")
-rm resourcemanager_jmx_url, --resourcemanager-url resourcemanager_jmx_url
Hadoop resourcemanager metrics URL. (default
"http://indata-10-110-13-164.indata.com:8088/jmx")
-dn datanode_jmx_url, --datanode-url datanode_jmx_url
Hadoop datanode metrics URL. (default
"http://indata-10-110-13-163.indata.com:1022/jmx")
-jn journalnode_jmx_url, --journalnode-url journalnode_jmx_url
Hadoop journalnode metrics URL. (default
"http://indata-10-110-13-163.indata.com:8480/jmx")
-mr mapreduce2_jmx_url, --mapreduce2-url mapreduce2_jmx_url
Hadoop mapreduce2 metrics URL. (default
"http://indata-10-110-13-165.indata.com:19888/jmx")
-hbase hbase_jmx_url, --hbase-url hbase_jmx_url
Hadoop hbase metrics URL. (default
"http://indata-10-110-13-164.indata.com:16010/jmx")
-hive hive_jmx_url, --hive-url hive_jmx_url
Hadoop hive metrics URL. (default
"http://ip:port/jmx")
-p metrics_path, --path metrics_path
Path under which to expose metrics. (default
"/metrics")
-host ip_or_hostname, -ip ip_or_hostname, --address ip_or_hostname, --addr ip_or_hostname
Polling server on this address. (default "127.0.0.1")
-P port, --port port Listen to this port. (default "9130")
Tested on Apache Hadoop 2.7.3
You can run each Collector under directory cmd/, just like:
cd hadoop_exporter/cmd
python hdfs_namenode.py -h
# input the params the script asked.
Or if you want to run the entire project, you should have an webhook/api with url = http://<rest_api_host_and_port>/alert/getservicesbyhost to provide the jmx urls: the content in the webhook or api should be like:
{
"cluster_name1": [
{
"node1.fqdn.com": {
"DATANODE": {
"jmx": "http://node1.fqdn.com:1022/jmx"
},
"HBASE_REGIONSERVER": {
"jmx": "http://node1.fqdn.com:60030/jmx"
},
"HISTORYSERVER": {
"jmx": "node1.fqdn.com:19888/jmx"
},
"JOURNALNODE": {
"jmx": "node1.fqdn.com:8480/jmx"
},
"NAMENODE": {
"jmx": "node1.fqdn.com:50070/jmx"
},
"NODEMANAGER": {
"jmx": "node1.fqdn.com:8042/jmx"
}
}
},
{
"node2.fqdn.com": {
"DATANODE": {
"jmx": "http://node2.fqdn.com:1022/jmx"
},
"HBASE_REGIONSERVER": {
"jmx": "http://node2.fqdn.com:60030/jmx"
},
"HIVE_LLAP": {
"jmx": "http://node2.fqdn.com:15002/jmx"
},
"HIVE_SERVER_INTERACTIVE": {
"jmx": "http://node2.fqdn.com:10502/jmx"
},
"JOURNALNODE": {
"jmx": "http://node2.fqdn.com:8480/jmx"
},
"NODEMANAGER": {
"jmx": "http://node2.fqdn.com:8042/jmx"
}
}
},
{
"node3.fqdn.com": {
"DATANODE": {
"jmx": "http://node3.fqdn.com:1022/jmx"
},
"HBASE_MASTER": {
"jmx": "http://node3.fqdn.com:16010/jmx"
},
"HBASE_REGIONSERVER": {
"jmx": "http://node3.fqdn.com:60030/jmx"
},
"JOURNALNODE": {
"jmx": "http://node3.fqdn.com:8480/jmx"
},
"NODEMANAGER": {
"jmx": "http://node3.fqdn.com:8042/jmx"
},
"RESOURCEMANAGER": {
"jmx": "http://node3.fqdn.com:8088/jmx"
}
}
}
],
"cluster_name2": [
{
"node4.fqdn.com": {
"DATANODE": {
"jmx": "http://node4.fqdn.com:1022/jmx"
},
"HBASE_REGIONSERVER": {
"jmx": "http://node4.fqdn.com:60030/jmx"
},
"HISTORYSERVER": {
"jmx": "node4.fqdn.com:19888/jmx"
},
"JOURNALNODE": {
"jmx": "node4.fqdn.com:8480/jmx"
},
"NAMENODE": {
"jmx": "node4.fqdn.com:50070/jmx"
},
"NODEMANAGER": {
"jmx": "node4.fqdn.com:8042/jmx"
}
}
},
{
"node5.fqdn.com": {
"DATANODE": {
"jmx": "http://node5.fqdn.com:1022/jmx"
},
"HBASE_REGIONSERVER": {
"jmx": "http://node5.fqdn.com:60030/jmx"
},
"HIVE_LLAP": {
"jmx": "http://node5.fqdn.com:15002/jmx"
},
"HIVE_SERVER_INTERACTIVE": {
"jmx": "http://node5.fqdn.com:10502/jmx"
},
"JOURNALNODE": {
"jmx": "http://node5.fqdn.com:8480/jmx"
},
"NODEMANAGER": {
"jmx": "http://node5.fqdn.com:8042/jmx"
}
}
},
{
"node6.fqdn.com": {
"DATANODE": {
"jmx": "http://node6.fqdn.com:1022/jmx"
},
"HBASE_MASTER": {
"jmx": "http://node6.fqdn.com:16010/jmx"
},
"HBASE_REGIONSERVER": {
"jmx": "http://node6.fqdn.com:60030/jmx"
},
"JOURNALNODE": {
"jmx": "http://node6.fqdn.com:8480/jmx"
},
"NODEMANAGER": {
"jmx": "http://node6.fqdn.com:8042/jmx"
},
"RESOURCEMANAGER": {
"jmx": "http://node6.fqdn.com:8088/jmx"
}
}
}
]
}
Then you can run:
# -s means the rest api or webhook url mentioned above, should be in <host:port> format, no schema and path( I know it's ugly).
# -P (upper) means hadoop_exporter should export metrics in this port. you can get metrics from <http://hostname:9131/metrics>
python hadoop_exporter.py -s "<rest_api_host_and_port>" -P 9131
One more thing: you should run all this steps in all hadoop nodes.
MAYBE I'll improve this project for common use.