Making storage plugin on Apache Drill to HDFS

问题:

I'm trying to make storage plugin for Hadoop (hdfs) and Apache Drill. Actually I'm confused and I don't know what to set as port for hdfs:// connection, and what to set for location. This is my plugin:

 {
 "type": "file",
 "enabled": true,
 "connection": "hdfs://localhost:54310",
 "workspaces": {
 "root": {
  "location": "/",
  "writable": false,
  "defaultInputFormat": null
},
"tmp": {
  "location": "/tmp",
  "writable": true,
  "defaultInputFormat": null
}
 },
"formats": {
  "psv": {
  "type": "text",
  "extensions": [
    "tbl"
  ],
  "delimiter": "|"
},
"csv": {
  "type": "text",
  "extensions": [
    "csv"
  ],
  "delimiter": ","
},
"tsv": {
  "type": "text",
  "extensions": [
    "tsv"
  ],
  "delimiter": "\t"
},
"parquet": {
  "type": "parquet"
},
"json": {
  "type": "json"
},
"avro": {
  "type": "avro"
   }
 }
}

So, is ti correct to set localhost:54310 because I got that with command:

 hdfs -getconf -nnRpcAddresses 

or it is :8020 ?

Second question, what do I need to set for location? My hadoop folder is in:

/usr/local/hadoop

, and there you can find /etc /bin /lib /log ... So, do I need to set location on my datanode, or?

Third question. When I'm connecting to Drill, I'm going through sqlline and than connecting on my zookeeper like:

  !connect jdbc:drill:zk=localhost:2181 

My question here is, after I make storage plugin and when I connect to Drill with zk, can I query hdfs file?

I'm very sorry if this is a noob question but I haven't find anything useful on internet or at least it haven't helped me. If you are able to explain me some stuff, I'll be very grateful.

回答1:

As per Drill docs,

  {
    "type" : "file",
    "enabled" : true,
    "connection" : "hdfs://10.10.30.156:8020/",
    "workspaces" : {
      "root" : {
        "location" : "/user/root/drill",
        "writable" : true,
        "defaultInputFormat" : null
      }
    },
    "formats" : {
      "json" : {
        "type" : "json"
      }
    }
  }

In "connection",

put namenode server address.

If you are not sure about this address. Check fs.default.name or fs.defaultFS properties in core-site.xml.

Coming to "workspaces",

you can save workspaces in this. In the above example, there is a workspace with name root and location /user/root/drill. This is your HDFS location.

If you have files under /user/root/drill hdfs directory, you can query them using this workspace name.

Example: abc is under this directory.

 select * from dfs.root.`abc.csv`

---

After successfully creating the plugin, you can start drill and start querying .

You can query any directory irrespective to workspaces.

Say you want to query employee.json in /tmp/data hdfs directory.

Query is :

select * from dfs.`/tmp/data/employee.json`

回答2:

I have similar problem, Drill cannot read dfs server. Finally, the problem is cause by namenode port. The default address of namenode web UI is http://localhost:50070/.

The default address of namenode server is hdfs://localhost:8020/.