0
0
Fork 0
mirror of https://github.com/NixOS/nixpkgs.git synced 2025-07-14 06:00:33 +03:00

nixos/hadoop: add hadoop module (hdfs, yarn)

This commit is contained in:
Allan Espinosa 2018-05-20 21:09:31 -04:00
parent 978397d358
commit 0c10b2baa6
9 changed files with 352 additions and 0 deletions

View file

@ -0,0 +1,31 @@
{ hadoop, pkgs }:
let
propertyXml = name: value: ''
<property>
<name>${name}</name>
<value>${builtins.toString value}</value>
</property>
'';
siteXml = fileName: properties: pkgs.writeTextDir fileName ''
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!-- generated by NixOS -->
<configuration>
${builtins.concatStringsSep "\n" (pkgs.lib.mapAttrsToList propertyXml properties)}
</configuration>
'';
userFunctions = ''
hadoop_verify_logdir() {
echo Skipping verification of log directory
}
'';
in
pkgs.buildEnv {
name = "hadoop-conf";
paths = [
(siteXml "core-site.xml" hadoop.coreSite)
(siteXml "hdfs-site.xml" hadoop.hdfsSite)
(siteXml "mapred-site.xml" hadoop.mapredSite)
(siteXml "yarn-site.xml" hadoop.yarnSite)
(pkgs.writeTextDir "hadoop-user-functions.sh" userFunctions)
];
}

View file

@ -0,0 +1,63 @@
{ config, lib, pkgs, ...}:
let
cfg = config.services.hadoop;
hadoopConf = import ./conf.nix { hadoop = cfg; pkgs = pkgs; };
in
with lib;
{
imports = [ ./yarn.nix ./hdfs.nix ];
options.services.hadoop = {
coreSite = mkOption {
default = {};
example = {
"fs.defaultFS" = "hdfs://localhost";
};
description = "Hadoop core-site.xml definition";
};
hdfsSite = mkOption {
default = {};
example = {
"dfs.nameservices" = "namenode1";
};
description = "Hadoop hdfs-site.xml definition";
};
mapredSite = mkOption {
default = {};
example = {
"mapreduce.map.cpu.vcores" = "1";
};
description = "Hadoop mapred-site.xml definition";
};
yarnSite = mkOption {
default = {};
example = {
"yarn.resourcemanager.ha.id" = "resourcemanager1";
};
description = "Hadoop yarn-site.xml definition";
};
package = mkOption {
type = types.package;
default = pkgs.hadoop;
defaultText = "pkgs.hadoop";
example = literalExample "pkgs.hadoop";
description = ''
'';
};
};
config = mkMerge [
(mkIf (builtins.hasAttr "yarn" config.users.extraUsers ||
builtins.hasAttr "hdfs" config.users.extraUsers ) {
users.extraGroups.hadoop = {
gid = config.ids.gids.hadoop;
};
})
];
}

View file

@ -0,0 +1,73 @@
{ config, lib, pkgs, ...}:
let
cfg = config.services.hadoop;
hadoopConf = import ./conf.nix { hadoop = cfg; pkgs = pkgs; };
in
with lib;
{
options.services.hadoop.hdfs = {
namenode.enabled = mkOption {
type = types.bool;
default = false;
description = ''
Whether to run the Hadoop YARN NameNode
'';
};
datanode.enabled = mkOption {
type = types.bool;
default = false;
description = ''
Whether to run the Hadoop YARN DataNode
'';
};
};
config = mkMerge [
(mkIf cfg.hdfs.namenode.enabled {
systemd.services."hdfs-namenode" = {
description = "Hadoop HDFS NameNode";
wantedBy = [ "multi-user.target" ];
environment = {
HADOOP_HOME = "${cfg.package}";
};
preStart = ''
${cfg.package}/bin/hdfs --config ${hadoopConf} namenode -format -nonInteractive || true
'';
serviceConfig = {
User = "hdfs";
SyslogIdentifier = "hdfs-namenode";
ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} namenode";
};
};
})
(mkIf cfg.hdfs.datanode.enabled {
systemd.services."hdfs-datanode" = {
description = "Hadoop HDFS DataNode";
wantedBy = [ "multi-user.target" ];
environment = {
HADOOP_HOME = "${cfg.package}";
};
serviceConfig = {
User = "hdfs";
SyslogIdentifier = "hdfs-datanode";
ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} datanode";
};
};
})
(mkIf (
cfg.hdfs.namenode.enabled || cfg.hdfs.datanode.enabled
) {
users.extraUsers.hdfs = {
description = "Hadoop HDFS user";
group = "hadoop";
uid = config.ids.uids.hdfs;
};
})
];
}

View file

@ -0,0 +1,74 @@
{ config, lib, pkgs, ...}:
let
cfg = config.services.hadoop;
hadoopConf = import ./conf.nix { hadoop = cfg; pkgs = pkgs; };
in
with lib;
{
options.services.hadoop.yarn = {
resourcemanager.enabled = mkOption {
type = types.bool;
default = false;
description = ''
Whether to run the Hadoop YARN ResourceManager
'';
};
nodemanager.enabled = mkOption {
type = types.bool;
default = false;
description = ''
Whether to run the Hadoop YARN NodeManager
'';
};
};
config = mkMerge [
(mkIf (
cfg.yarn.resourcemanager.enabled || cfg.yarn.nodemanager.enabled
) {
users.extraUsers.yarn = {
description = "Hadoop YARN user";
group = "hadoop";
uid = config.ids.uids.yarn;
};
})
(mkIf cfg.yarn.resourcemanager.enabled {
systemd.services."yarn-resourcemanager" = {
description = "Hadoop YARN ResourceManager";
wantedBy = [ "multi-user.target" ];
environment = {
HADOOP_HOME = "${cfg.package}";
};
serviceConfig = {
User = "yarn";
SyslogIdentifier = "yarn-resourcemanager";
ExecStart = "${cfg.package}/bin/yarn --config ${hadoopConf} " +
" resourcemanager";
};
};
})
(mkIf cfg.yarn.nodemanager.enabled {
systemd.services."yarn-nodemanager" = {
description = "Hadoop YARN NodeManager";
wantedBy = [ "multi-user.target" ];
environment = {
HADOOP_HOME = "${cfg.package}";
};
serviceConfig = {
User = "yarn";
SyslogIdentifier = "yarn-nodemanager";
ExecStart = "${cfg.package}/bin/yarn --config ${hadoopConf} " +
" nodemanager";
};
};
})
];
}