From 6efa931c518e8fdc725527e465bceb0b8fb60a4a Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Sat, 8 Jan 2022 12:48:23 +0530
Subject: nixos/hadoop: fix mkenableoption text

---
 nixos/modules/services/cluster/hadoop/hdfs.nix | 10 +++++-----
 nixos/modules/services/cluster/hadoop/yarn.nix |  4 ++--
 2 files changed, 7 insertions(+), 7 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/hdfs.nix b/nixos/modules/services/cluster/hadoop/hdfs.nix
index be667aa82d8..8c373968364 100644
--- a/nixos/modules/services/cluster/hadoop/hdfs.nix
+++ b/nixos/modules/services/cluster/hadoop/hdfs.nix
@@ -17,7 +17,7 @@ in
 {
   options.services.hadoop.hdfs = {
     namenode = {
-      enable = mkEnableOption "Whether to run the HDFS NameNode";
+      enable = mkEnableOption "HDFS NameNode";
       formatOnInit = mkOption {
         type = types.bool;
         default = false;
@@ -37,7 +37,7 @@ in
       };
     };
     datanode = {
-      enable = mkEnableOption "Whether to run the HDFS DataNode";
+      enable = mkEnableOption "HDFS DataNode";
       inherit restartIfChanged;
       openFirewall = mkOption {
         type = types.bool;
@@ -48,7 +48,7 @@ in
       };
     };
     journalnode = {
-      enable = mkEnableOption "Whether to run the HDFS JournalNode";
+      enable = mkEnableOption "HDFS JournalNode";
       inherit restartIfChanged;
       openFirewall = mkOption {
         type = types.bool;
@@ -59,11 +59,11 @@ in
       };
     };
     zkfc = {
-      enable = mkEnableOption "Whether to run the HDFS ZooKeeper failover controller";
+      enable = mkEnableOption "HDFS ZooKeeper failover controller";
       inherit restartIfChanged;
     };
     httpfs = {
-      enable = mkEnableOption "Whether to run the HDFS HTTPfs server";
+      enable = mkEnableOption "HDFS HTTPfs server";
       tempPath = mkOption {
         type = types.path;
         default = "/tmp/hadoop/httpfs";
diff --git a/nixos/modules/services/cluster/hadoop/yarn.nix b/nixos/modules/services/cluster/hadoop/yarn.nix
index 37c26ea10f7..cc42d8f388a 100644
--- a/nixos/modules/services/cluster/hadoop/yarn.nix
+++ b/nixos/modules/services/cluster/hadoop/yarn.nix
@@ -17,7 +17,7 @@ in
 {
   options.services.hadoop.yarn = {
     resourcemanager = {
-      enable = mkEnableOption "Whether to run the Hadoop YARN ResourceManager";
+      enable = mkEnableOption "Hadoop YARN ResourceManager";
       inherit restartIfChanged;
       openFirewall = mkOption {
         type = types.bool;
@@ -28,7 +28,7 @@ in
       };
     };
     nodemanager = {
-      enable = mkEnableOption "Whether to run the Hadoop YARN NodeManager";
+      enable = mkEnableOption "Hadoop YARN NodeManager";
       inherit restartIfChanged;
       addBinBash = mkOption {
         type = types.bool;
-- 
cgit 1.4.1


From dd5f004b06a16e7c291bd159f792b718b7fce0b3 Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Sat, 8 Jan 2022 14:02:32 +0530
Subject: nixos/hadoop: refactor HDFS options

---
 nixos/modules/services/cluster/hadoop/hdfs.nix | 69 +++++++-------------------
 1 file changed, 19 insertions(+), 50 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/hdfs.nix b/nixos/modules/services/cluster/hadoop/hdfs.nix
index 8c373968364..9caa3d07525 100644
--- a/nixos/modules/services/cluster/hadoop/hdfs.nix
+++ b/nixos/modules/services/cluster/hadoop/hdfs.nix
@@ -13,71 +13,40 @@ let
     '';
     default = false;
   };
+  openFirewall = serviceName: mkOption {
+    type = types.bool;
+    default = true;
+    description = "Open firewall ports for ${serviceName}.";
+  };
+  hadoopServiceOption = { serviceName, firewallOption ? true }: {
+    enable = mkEnableOption serviceName;
+    inherit restartIfChanged;
+  } // (if firewallOption then {openFirewall = openFirewall serviceName;} else {});
 in
 {
   options.services.hadoop.hdfs = {
-    namenode = {
-      enable = mkEnableOption "HDFS NameNode";
+    namenode = hadoopServiceOption { serviceName = "HDFS NameNode"; } // {
       formatOnInit = mkOption {
         type = types.bool;
         default = false;
         description = ''
           Format HDFS namenode on first start. This is useful for quickly spinning up ephemeral HDFS clusters with a single namenode.
-          For HA clusters, initialization involves multiple steps across multiple nodes. Follow [this guide](https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs/HDFSHighAvailabilityWithQJM.html)
-          to initialize an HA cluster manually.
-        '';
-      };
-      inherit restartIfChanged;
-      openFirewall = mkOption {
-        type = types.bool;
-        default = true;
-        description = ''
-          Open firewall ports for namenode
-        '';
-      };
-    };
-    datanode = {
-      enable = mkEnableOption "HDFS DataNode";
-      inherit restartIfChanged;
-      openFirewall = mkOption {
-        type = types.bool;
-        default = true;
-        description = ''
-          Open firewall ports for datanode
+          For HA clusters, initialization involves multiple steps across multiple nodes. Follow this guide to initialize an HA cluster manually:
+          <link xlink:href="https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs/HDFSHighAvailabilityWithQJM.html"/>
         '';
       };
     };
-    journalnode = {
-      enable = mkEnableOption "HDFS JournalNode";
-      inherit restartIfChanged;
-      openFirewall = mkOption {
-        type = types.bool;
-        default = true;
-        description = ''
-          Open firewall ports for journalnode
-        '';
-      };
-    };
-    zkfc = {
-      enable = mkEnableOption "HDFS ZooKeeper failover controller";
-      inherit restartIfChanged;
+    datanode = hadoopServiceOption { serviceName = "HDFS DataNode"; };
+    journalnode = hadoopServiceOption { serviceName = "HDFS JournalNode"; };
+    zkfc = hadoopServiceOption {
+      serviceName = "HDFS ZooKeeper failover controller";
+      firewallOption = false;
     };
-    httpfs = {
-      enable = mkEnableOption "HDFS HTTPfs server";
+    httpfs = hadoopServiceOption { serviceName = "HDFS JournalNode"; } // {
       tempPath = mkOption {
         type = types.path;
         default = "/tmp/hadoop/httpfs";
-        description = ''
-          HTTPFS_TEMP path used by HTTPFS
-        '';
-      };
-      inherit restartIfChanged;
-      openFirewall = mkOption {
-        type = types.bool;
-        default = true;
-        description = ''
-          Open firewall ports for HTTPFS
-        '';
+        description = "HTTPFS_TEMP path used by HTTPFS";
       };
     };
   };
-- 
cgit 1.4.1


From f6cf1ced335869002697efa8f18cb77493a38e56 Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Sat, 8 Jan 2022 16:02:32 +0530
Subject: nixos/hadoop: refactor HDFS configs

---
 nixos/modules/services/cluster/hadoop/hdfs.nix | 211 ++++++++++++-------------
 1 file changed, 100 insertions(+), 111 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/hdfs.nix b/nixos/modules/services/cluster/hadoop/hdfs.nix
index 9caa3d07525..71bd4478670 100644
--- a/nixos/modules/services/cluster/hadoop/hdfs.nix
+++ b/nixos/modules/services/cluster/hadoop/hdfs.nix
@@ -1,47 +1,89 @@
-{ config, lib, pkgs, ...}:
+{ config, lib, pkgs, ... }:
 with lib;
 let
   cfg = config.services.hadoop;
+
+  # Config files for hadoop services
   hadoopConf = "${import ./conf.nix { inherit cfg pkgs lib; }}/";
-  restartIfChanged  = mkOption {
-    type = types.bool;
-    description = ''
-      Automatically restart the service on config change.
-      This can be set to false to defer restarts on clusters running critical applications.
-      Please consider the security implications of inadvertently running an older version,
-      and the possibility of unexpected behavior caused by inconsistent versions across a cluster when disabling this option.
-    '';
-    default = false;
-  };
-  openFirewall = serviceName: mkOption {
-    type = types.bool;
-    default = true;
-    description = "Open firewall ports for ${serviceName}.";
-  };
+
+  # Generator for HDFS service options
   hadoopServiceOption = { serviceName, firewallOption ? true }: {
     enable = mkEnableOption serviceName;
-    inherit restartIfChanged;
-  } // (if firewallOption then {openFirewall = openFirewall serviceName;} else {});
+    restartIfChanged = mkOption {
+      type = types.bool;
+      description = ''
+        Automatically restart the service on config change.
+        This can be set to false to defer restarts on clusters running critical applications.
+        Please consider the security implications of inadvertently running an older version,
+        and the possibility of unexpected behavior caused by inconsistent versions across a cluster when disabling this option.
+      '';
+      default = false;
+    };
+  } // (optionalAttrs firewallOption {
+    openFirewall = mkOption {
+      type = types.bool;
+      default = true;
+      description = "Open firewall ports for ${serviceName}.";
+    };
+  });
+
+  # Generator for HDFS service configs
+  hadoopServiceConfig =
+    { name
+    , serviceOptions ? cfg.hdfs."${toLower name}"
+    , description ? "Hadoop HDFS ${name}"
+    , User ? "hdfs"
+    , allowedTCPPorts ? [ ]
+    , preStart ? ""
+    , environment ? { }
+    }: (
+
+      mkIf serviceOptions.enable {
+        systemd.services."hdfs-${toLower name}" = {
+          inherit description preStart environment;
+          wantedBy = [ "multi-user.target" ];
+          inherit (serviceOptions) restartIfChanged;
+          serviceConfig = {
+            inherit User;
+            SyslogIdentifier = "hdfs-${toLower name}";
+            ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} ${toLower name}";
+            Restart = "always";
+          };
+        };
+
+        networking.firewall.allowedTCPPorts = mkIf
+          ((builtins.hasAttr "openFirewall" serviceOptions) && serviceOptions.openFirewall)
+          allowedTCPPorts;
+      }
+    );
+
 in
 {
   options.services.hadoop.hdfs = {
+
     namenode = hadoopServiceOption { serviceName = "HDFS NameNode"; } // {
       formatOnInit = mkOption {
         type = types.bool;
         default = false;
         description = ''
-          Format HDFS namenode on first start. This is useful for quickly spinning up ephemeral HDFS clusters with a single namenode.
-          For HA clusters, initialization involves multiple steps across multiple nodes. Follow this guide to initialize an HA cluster manually:
+          Format HDFS namenode on first start. This is useful for quickly spinning up
+          ephemeral HDFS clusters with a single namenode.
+          For HA clusters, initialization involves multiple steps across multiple nodes.
+          Follow this guide to initialize an HA cluster manually:
           <link xlink:href="https://hadoop.apache.org/docs/stable/hadoop-project-dist/hadoop-hdfs/HDFSHighAvailabilityWithQJM.html"/>
         '';
       };
     };
+
     datanode = hadoopServiceOption { serviceName = "HDFS DataNode"; };
+
     journalnode = hadoopServiceOption { serviceName = "HDFS JournalNode"; };
+
     zkfc = hadoopServiceOption {
       serviceName = "HDFS ZooKeeper failover controller";
       firewallOption = false;
     };
+
     httpfs = hadoopServiceOption { serviceName = "HDFS JournalNode"; } // {
       tempPath = mkOption {
         type = types.path;
@@ -49,118 +91,65 @@ in
         description = "HTTPFS_TEMP path used by HTTPFS";
       };
     };
+
   };
 
   config = mkMerge [
-    (mkIf cfg.hdfs.namenode.enable {
-      systemd.services.hdfs-namenode = {
-        description = "Hadoop HDFS NameNode";
-        wantedBy = [ "multi-user.target" ];
-        inherit (cfg.hdfs.namenode) restartIfChanged;
-
-        preStart = (mkIf cfg.hdfs.namenode.formatOnInit ''
-          ${cfg.package}/bin/hdfs --config ${hadoopConf} namenode -format -nonInteractive || true
-        '');
-
-        serviceConfig = {
-          User = "hdfs";
-          SyslogIdentifier = "hdfs-namenode";
-          ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} namenode";
-          Restart = "always";
-        };
-      };
-
-      networking.firewall.allowedTCPPorts = (mkIf cfg.hdfs.namenode.openFirewall [
+    (hadoopServiceConfig {
+      name = "NameNode";
+      allowedTCPPorts = [
         9870 # namenode.http-address
         8020 # namenode.rpc-address
         8022 # namenode. servicerpc-address
-      ]);
+      ];
+      preStart = (mkIf cfg.hdfs.namenode.formatOnInit
+        "${cfg.package}/bin/hdfs --config ${hadoopConf} namenode -format -nonInteractive || true"
+      );
     })
-    (mkIf cfg.hdfs.datanode.enable {
-      systemd.services.hdfs-datanode = {
-        description = "Hadoop HDFS DataNode";
-        wantedBy = [ "multi-user.target" ];
-        inherit (cfg.hdfs.datanode) restartIfChanged;
-
-        serviceConfig = {
-          User = "hdfs";
-          SyslogIdentifier = "hdfs-datanode";
-          ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} datanode";
-          Restart = "always";
-        };
-      };
 
-      networking.firewall.allowedTCPPorts = (mkIf cfg.hdfs.datanode.openFirewall [
+    (hadoopServiceConfig {
+      name = "DataNode";
+      allowedTCPPorts = [
         9864 # datanode.http.address
         9866 # datanode.address
         9867 # datanode.ipc.address
-      ]);
+      ];
     })
-    (mkIf cfg.hdfs.journalnode.enable {
-      systemd.services.hdfs-journalnode = {
-        description = "Hadoop HDFS JournalNode";
-        wantedBy = [ "multi-user.target" ];
-        inherit (cfg.hdfs.journalnode) restartIfChanged;
-
-        serviceConfig = {
-          User = "hdfs";
-          SyslogIdentifier = "hdfs-journalnode";
-          ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} journalnode";
-          Restart = "always";
-        };
-      };
 
-      networking.firewall.allowedTCPPorts = (mkIf cfg.hdfs.journalnode.openFirewall [
+    (hadoopServiceConfig {
+      name = "JournalNode";
+      allowedTCPPorts = [
         8480 # dfs.journalnode.http-address
         8485 # dfs.journalnode.rpc-address
-      ]);
+      ];
     })
-    (mkIf cfg.hdfs.zkfc.enable {
-      systemd.services.hdfs-zkfc = {
-        description = "Hadoop HDFS ZooKeeper failover controller";
-        wantedBy = [ "multi-user.target" ];
-        inherit (cfg.hdfs.zkfc) restartIfChanged;
-
-        serviceConfig = {
-          User = "hdfs";
-          SyslogIdentifier = "hdfs-zkfc";
-          ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} zkfc";
-          Restart = "always";
-        };
-      };
-    })
-    (mkIf cfg.hdfs.httpfs.enable {
-      systemd.services.hdfs-httpfs = {
-        description = "Hadoop httpfs";
-        wantedBy = [ "multi-user.target" ];
-        inherit (cfg.hdfs.httpfs) restartIfChanged;
-
-        environment.HTTPFS_TEMP = cfg.hdfs.httpfs.tempPath;
 
-        preStart = ''
-          mkdir -p $HTTPFS_TEMP
-        '';
+    (hadoopServiceConfig {
+      name = "zkfc";
+      description = "Hadoop HDFS ZooKeeper failover controller";
+    })
 
-        serviceConfig = {
-          User = "httpfs";
-          SyslogIdentifier = "hdfs-httpfs";
-          ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} httpfs";
-          Restart = "always";
-        };
-      };
-      networking.firewall.allowedTCPPorts = (mkIf cfg.hdfs.httpfs.openFirewall [
+    (hadoopServiceConfig {
+      name = "HTTPFS";
+      environment.HTTPFS_TEMP = cfg.hdfs.httpfs.tempPath;
+      preStart = "mkdir -p $HTTPFS_TEMP";
+      User = "httpfs";
+      allowedTCPPorts = [
         14000 # httpfs.http.port
-      ]);
+      ];
     })
-    (mkIf (
+
+    (mkIf
+      (
         cfg.hdfs.namenode.enable || cfg.hdfs.datanode.enable || cfg.hdfs.journalnode.enable || cfg.hdfs.zkfc.enable
-    ) {
-      users.users.hdfs = {
-        description = "Hadoop HDFS user";
-        group = "hadoop";
-        uid = config.ids.uids.hdfs;
-      };
-    })
+      )
+      {
+        users.users.hdfs = {
+          description = "Hadoop HDFS user";
+          group = "hadoop";
+          uid = config.ids.uids.hdfs;
+        };
+      })
     (mkIf cfg.hdfs.httpfs.enable {
       users.users.httpfs = {
         description = "Hadoop HTTPFS user";
-- 
cgit 1.4.1


From 0f97c9ae82bae1d445b14483864d784e1245204e Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Sat, 8 Jan 2022 16:19:15 +0530
Subject: nixos/hadoop: disable openFirewall by default

---
 nixos/modules/services/cluster/hadoop/hdfs.nix |  2 +-
 nixos/modules/services/cluster/hadoop/yarn.nix |  4 +--
 nixos/tests/hadoop/hadoop.nix                  | 45 ++++++++++++++++++++------
 nixos/tests/hadoop/hdfs.nix                    | 11 +++++--
 nixos/tests/hadoop/yarn.nix                    | 10 ++++--
 5 files changed, 56 insertions(+), 16 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/hdfs.nix b/nixos/modules/services/cluster/hadoop/hdfs.nix
index 71bd4478670..451e74df712 100644
--- a/nixos/modules/services/cluster/hadoop/hdfs.nix
+++ b/nixos/modules/services/cluster/hadoop/hdfs.nix
@@ -22,7 +22,7 @@ let
   } // (optionalAttrs firewallOption {
     openFirewall = mkOption {
       type = types.bool;
-      default = true;
+      default = false;
       description = "Open firewall ports for ${serviceName}.";
     };
   });
diff --git a/nixos/modules/services/cluster/hadoop/yarn.nix b/nixos/modules/services/cluster/hadoop/yarn.nix
index cc42d8f388a..90ae75a44b7 100644
--- a/nixos/modules/services/cluster/hadoop/yarn.nix
+++ b/nixos/modules/services/cluster/hadoop/yarn.nix
@@ -21,7 +21,7 @@ in
       inherit restartIfChanged;
       openFirewall = mkOption {
         type = types.bool;
-        default = true;
+        default = false;
         description = ''
           Open firewall ports for resourcemanager
         '';
@@ -39,7 +39,7 @@ in
       };
       openFirewall = mkOption {
         type = types.bool;
-        default = true;
+        default = false;
         description = ''
           Open firewall ports for nodemanager.
           Because containers can listen on any ephemeral port, TCP ports 1024–65535 will be opened.
diff --git a/nixos/tests/hadoop/hadoop.nix b/nixos/tests/hadoop/hadoop.nix
index 48737debab5..adc3c9f393c 100644
--- a/nixos/tests/hadoop/hadoop.nix
+++ b/nixos/tests/hadoop/hadoop.nix
@@ -55,14 +55,20 @@ import ../make-test-python.nix ({pkgs, ...}: {
     nn1 = {pkgs, options, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
-        hdfs.namenode.enable = true;
+        hdfs.namenode = {
+          enable = true;
+          openFirewall = true;
+        };
         hdfs.zkfc.enable = true;
       };
     };
     nn2 = {pkgs, options, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
-        hdfs.namenode.enable = true;
+        hdfs.namenode = {
+          enable = true;
+          openFirewall = true;
+        };
         hdfs.zkfc.enable = true;
       };
     };
@@ -70,26 +76,38 @@ import ../make-test-python.nix ({pkgs, ...}: {
     jn1 = {pkgs, options, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
-        hdfs.journalnode.enable = true;
+        hdfs.journalnode = {
+          enable = true;
+          openFirewall = true;
+        };
       };
     };
     jn2 = {pkgs, options, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
-        hdfs.journalnode.enable = true;
+        hdfs.journalnode = {
+          enable = true;
+          openFirewall = true;
+        };
       };
     };
     jn3 = {pkgs, options, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
-        hdfs.journalnode.enable = true;
+        hdfs.journalnode = {
+          enable = true;
+          openFirewall = true;
+        };
       };
     };
 
     dn1 = {pkgs, options, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
-        hdfs.datanode.enable = true;
+        hdfs.datanode = {
+          enable = true;
+          openFirewall = true;
+        };
       };
     };
 
@@ -98,14 +116,20 @@ import ../make-test-python.nix ({pkgs, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
         yarnSite = options.services.hadoop.yarnSite.default // yarnSiteHA;
-        yarn.resourcemanager.enable = true;
+        yarn.resourcemanager = {
+          enable = true;
+          openFirewall = true;
+        };
       };
     };
     rm2 = {pkgs, options, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
         yarnSite = options.services.hadoop.yarnSite.default // yarnSiteHA;
-        yarn.resourcemanager.enable = true;
+        yarn.resourcemanager = {
+          enable = true;
+          openFirewall = true;
+        };
       };
     };
     nm1 = {pkgs, options, ...}: {
@@ -113,7 +137,10 @@ import ../make-test-python.nix ({pkgs, ...}: {
       services.hadoop = {
         inherit package coreSite hdfsSite;
         yarnSite = options.services.hadoop.yarnSite.default // yarnSiteHA;
-        yarn.nodemanager.enable = true;
+        yarn.nodemanager = {
+          enable = true;
+          openFirewall = true;
+        };
       };
     };
   };
diff --git a/nixos/tests/hadoop/hdfs.nix b/nixos/tests/hadoop/hdfs.nix
index b63cbf48032..c5aee0d5ee7 100644
--- a/nixos/tests/hadoop/hdfs.nix
+++ b/nixos/tests/hadoop/hdfs.nix
@@ -7,9 +7,13 @@ import ../make-test-python.nix ({...}: {
         hdfs = {
           namenode = {
             enable = true;
+            openFirewall = true;
             formatOnInit = true;
           };
-          httpfs.enable = true;
+          httpfs = {
+            enable = true;
+            openFirewall = true;
+          };
         };
         coreSite = {
           "fs.defaultFS" = "hdfs://namenode:8020";
@@ -21,7 +25,10 @@ import ../make-test-python.nix ({...}: {
     datanode = {pkgs, ...}: {
       services.hadoop = {
         package = pkgs.hadoop;
-        hdfs.datanode.enable = true;
+        hdfs.datanode = {
+          enable = true;
+          openFirewall = true;
+        };
         coreSite = {
           "fs.defaultFS" = "hdfs://namenode:8020";
           "hadoop.proxyuser.httpfs.groups" = "*";
diff --git a/nixos/tests/hadoop/yarn.nix b/nixos/tests/hadoop/yarn.nix
index 09bdb35791c..fbf05b19cd2 100644
--- a/nixos/tests/hadoop/yarn.nix
+++ b/nixos/tests/hadoop/yarn.nix
@@ -3,14 +3,20 @@ import ../make-test-python.nix ({...}: {
   nodes = {
     resourcemanager = {pkgs, ...}: {
       services.hadoop.package = pkgs.hadoop;
-      services.hadoop.yarn.resourcemanager.enable = true;
+      services.hadoop.yarn.resourcemanager = {
+        enable = true;
+        openFirewall = true;
+      };
       services.hadoop.yarnSite = {
         "yarn.resourcemanager.scheduler.class" = "org.apache.hadoop.yarn.server.resourcemanager.scheduler.fifo.FifoScheduler";
       };
     };
     nodemanager = {pkgs, ...}: {
       services.hadoop.package = pkgs.hadoop;
-      services.hadoop.yarn.nodemanager.enable = true;
+      services.hadoop.yarn.nodemanager = {
+        enable = true;
+        openFirewall = true;
+      };
       services.hadoop.yarnSite = {
         "yarn.resourcemanager.hostname" = "resourcemanager";
         "yarn.nodemanager.log-dirs" = "/tmp/userlogs";
-- 
cgit 1.4.1


From c82d48913f581a5c878e9f31075c23563f92b9a3 Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Sat, 8 Jan 2022 16:47:26 +0530
Subject: nixos/hadoop: add HADOOP_CONF_DIR to env

---
 nixos/modules/services/cluster/hadoop/default.nix | 1 +
 1 file changed, 1 insertion(+)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/default.nix b/nixos/modules/services/cluster/hadoop/default.nix
index a1a95fe31ca..0adedaf1906 100644
--- a/nixos/modules/services/cluster/hadoop/default.nix
+++ b/nixos/modules/services/cluster/hadoop/default.nix
@@ -169,6 +169,7 @@ with lib;
         etc."hadoop-conf".source = let
           hadoopConf = "${import ./conf.nix { inherit cfg pkgs lib; }}/";
         in "${hadoopConf}";
+        variables.HADOOP_CONF_DIR = "/etc/hadoop-conf/";
       };
     })
 
-- 
cgit 1.4.1


From 8aeb60f034d7581245bdcb3f26e5bc3078dfe0ea Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Sun, 27 Feb 2022 10:37:35 +0530
Subject: nixos/hadoop: use FairScheduler by default

---
 nixos/modules/services/cluster/hadoop/default.nix | 2 +-
 nixos/tests/hadoop/yarn.nix                       | 3 ---
 2 files changed, 1 insertion(+), 4 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/default.nix b/nixos/modules/services/cluster/hadoop/default.nix
index 0adedaf1906..9968706d5a9 100644
--- a/nixos/modules/services/cluster/hadoop/default.nix
+++ b/nixos/modules/services/cluster/hadoop/default.nix
@@ -77,7 +77,7 @@ with lib;
         "yarn.nodemanager.linux-container-executor.path" = "/run/wrappers/yarn-nodemanager/bin/container-executor";
         "yarn.nodemanager.log-dirs" = "/var/log/hadoop/yarn/nodemanager";
         "yarn.resourcemanager.bind-host" = "0.0.0.0";
-        "yarn.resourcemanager.scheduler.class" = "org.apache.hadoop.yarn.server.resourcemanager.scheduler.fifo.FifoScheduler";
+        "yarn.resourcemanager.scheduler.class" = "org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FairScheduler";
       };
       type = types.attrsOf types.anything;
       example = literalExpression ''
diff --git a/nixos/tests/hadoop/yarn.nix b/nixos/tests/hadoop/yarn.nix
index c121f6556d5..ba39e257dc8 100644
--- a/nixos/tests/hadoop/yarn.nix
+++ b/nixos/tests/hadoop/yarn.nix
@@ -10,9 +10,6 @@ import ../make-test-python.nix ({ package, ... }: {
           enable = true;
           openFirewall = true;
         };
-        yarnSite = {
-          "yarn.resourcemanager.scheduler.class" = "org.apache.hadoop.yarn.server.resourcemanager.scheduler.fifo.FifoScheduler";
-        };
       };
     };
     nodemanager = { ... }: {
-- 
cgit 1.4.1


From d39056d165c9dc480284f3bd1e63b6213f0e518d Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Sun, 27 Feb 2022 10:41:05 +0530
Subject: nixos/hadoop: fix tests for hadoop 2 and 3.2

---
 nixos/modules/services/cluster/hadoop/default.nix |  1 +
 nixos/modules/services/cluster/hadoop/hdfs.nix    | 10 ++++--
 nixos/tests/hadoop/hdfs.nix                       | 40 +++++++++++++++--------
 3 files changed, 35 insertions(+), 16 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/default.nix b/nixos/modules/services/cluster/hadoop/default.nix
index 9968706d5a9..636bb406727 100644
--- a/nixos/modules/services/cluster/hadoop/default.nix
+++ b/nixos/modules/services/cluster/hadoop/default.nix
@@ -25,6 +25,7 @@ with lib;
     hdfsSite = mkOption {
       default = {
         "dfs.namenode.rpc-bind-host" = "0.0.0.0";
+        "dfs.namenode.http-address" = "0.0.0.0:9870";
       };
       type = types.attrsOf types.anything;
       example = literalExpression ''
diff --git a/nixos/modules/services/cluster/hadoop/hdfs.nix b/nixos/modules/services/cluster/hadoop/hdfs.nix
index 451e74df712..61d9941298a 100644
--- a/nixos/modules/services/cluster/hadoop/hdfs.nix
+++ b/nixos/modules/services/cluster/hadoop/hdfs.nix
@@ -100,7 +100,8 @@ in
       allowedTCPPorts = [
         9870 # namenode.http-address
         8020 # namenode.rpc-address
-        8022 # namenode. servicerpc-address
+        8022 # namenode.servicerpc-address
+        8019 # dfs.ha.zkfc.port
       ];
       preStart = (mkIf cfg.hdfs.namenode.formatOnInit
         "${cfg.package}/bin/hdfs --config ${hadoopConf} namenode -format -nonInteractive || true"
@@ -109,10 +110,15 @@ in
 
     (hadoopServiceConfig {
       name = "DataNode";
-      allowedTCPPorts = [
+      # port numbers for datanode changed between hadoop 2 and 3
+      allowedTCPPorts = if versionAtLeast cfg.package.version "3" then [
         9864 # datanode.http.address
         9866 # datanode.address
         9867 # datanode.ipc.address
+      ] else [
+        50075 # datanode.http.address
+        50010 # datanode.address
+        50020 # datanode.ipc.address
       ];
     })
 
diff --git a/nixos/tests/hadoop/hdfs.nix b/nixos/tests/hadoop/hdfs.nix
index e60d1410917..cc70fb8ecaf 100644
--- a/nixos/tests/hadoop/hdfs.nix
+++ b/nixos/tests/hadoop/hdfs.nix
@@ -1,8 +1,16 @@
 # Test a minimal HDFS cluster with no HA
-import ../make-test-python.nix ({ package, ... }: {
+import ../make-test-python.nix ({ package, lib, ... }:
+with lib;
+{
   name = "hadoop-hdfs";
 
-  nodes = {
+  nodes = let
+    coreSite = {
+      "fs.defaultFS" = "hdfs://namenode:8020";
+      "hadoop.proxyuser.httpfs.groups" = "*";
+      "hadoop.proxyuser.httpfs.hosts" = "*";
+    };
+    in {
     namenode = { pkgs, ... }: {
       services.hadoop = {
         inherit package;
@@ -13,15 +21,12 @@ import ../make-test-python.nix ({ package, ... }: {
             formatOnInit = true;
           };
           httpfs = {
-            enable = true;
+            # The NixOS hadoop module only support webHDFS on 3.3 and newer
+            enable = mkIf (versionAtLeast package.version "3.3") true;
             openFirewall = true;
           };
         };
-        coreSite = {
-          "fs.defaultFS" = "hdfs://namenode:8020";
-          "hadoop.proxyuser.httpfs.groups" = "*";
-          "hadoop.proxyuser.httpfs.hosts" = "*";
-        };
+        inherit coreSite;
       };
     };
     datanode = { pkgs, ... }: {
@@ -31,11 +36,7 @@ import ../make-test-python.nix ({ package, ... }: {
           enable = true;
           openFirewall = true;
         };
-        coreSite = {
-          "fs.defaultFS" = "hdfs://namenode:8020";
-          "hadoop.proxyuser.httpfs.groups" = "*";
-          "hadoop.proxyuser.httpfs.hosts" = "*";
-        };
+        inherit coreSite;
       };
     };
   };
@@ -46,21 +47,32 @@ import ../make-test-python.nix ({ package, ... }: {
     namenode.wait_for_unit("hdfs-namenode")
     namenode.wait_for_unit("network.target")
     namenode.wait_for_open_port(8020)
+    namenode.succeed("ss -tulpne | systemd-cat")
+    namenode.succeed("cat /etc/hadoop*/hdfs-site.xml | systemd-cat")
     namenode.wait_for_open_port(9870)
 
     datanode.wait_for_unit("hdfs-datanode")
     datanode.wait_for_unit("network.target")
+  '' + ( if versionAtLeast package.version "3" then ''
     datanode.wait_for_open_port(9864)
     datanode.wait_for_open_port(9866)
     datanode.wait_for_open_port(9867)
 
-    namenode.succeed("curl -f http://namenode:9870")
     datanode.succeed("curl -f http://datanode:9864")
+  '' else ''
+    datanode.wait_for_open_port(50075)
+    datanode.wait_for_open_port(50010)
+    datanode.wait_for_open_port(50020)
+
+    datanode.succeed("curl -f http://datanode:50075")
+  '' ) + ''
+    namenode.succeed("curl -f http://namenode:9870")
 
     datanode.succeed("sudo -u hdfs hdfs dfsadmin -safemode wait")
     datanode.succeed("echo testfilecontents | sudo -u hdfs hdfs dfs -put - /testfile")
     assert "testfilecontents" in datanode.succeed("sudo -u hdfs hdfs dfs -cat /testfile")
 
+  '' + optionalString ( versionAtLeast package.version "3.3" ) ''
     namenode.wait_for_unit("hdfs-httpfs")
     namenode.wait_for_open_port(14000)
     assert "testfilecontents" in datanode.succeed("curl -f \"http://namenode:14000/webhdfs/v1/testfile?user.name=hdfs&op=OPEN\" 2>&1")
-- 
cgit 1.4.1


From 716b0dfaaf12afe83ff54b793dc52c022ab62155 Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Sun, 27 Feb 2022 11:52:18 +0530
Subject: nixos/hadoop: add gateway role

---
 nixos/modules/services/cluster/hadoop/default.nix | 31 +++++++--------
 nixos/modules/services/cluster/hadoop/hdfs.nix    | 21 +++++-----
 nixos/modules/services/cluster/hadoop/yarn.nix    | 10 +++--
 nixos/tests/hadoop/hadoop.nix                     | 47 +++++++++++++----------
 4 files changed, 57 insertions(+), 52 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/default.nix b/nixos/modules/services/cluster/hadoop/default.nix
index 636bb406727..5c7ea79c301 100644
--- a/nixos/modules/services/cluster/hadoop/default.nix
+++ b/nixos/modules/services/cluster/hadoop/default.nix
@@ -149,6 +149,8 @@ with lib;
       description = "Directories containing additional config files to be added to HADOOP_CONF_DIR";
     };
 
+    gatewayRole.enable = mkEnableOption "gateway role for deploying hadoop configs";
+
     package = mkOption {
       type = types.package;
       default = pkgs.hadoop;
@@ -158,21 +160,16 @@ with lib;
   };
 
 
-  config = mkMerge [
-    (mkIf (builtins.hasAttr "yarn" config.users.users ||
-           builtins.hasAttr "hdfs" config.users.users ||
-           builtins.hasAttr "httpfs" config.users.users) {
-      users.groups.hadoop = {
-        gid = config.ids.gids.hadoop;
-      };
-      environment = {
-        systemPackages = [ cfg.package ];
-        etc."hadoop-conf".source = let
-          hadoopConf = "${import ./conf.nix { inherit cfg pkgs lib; }}/";
-        in "${hadoopConf}";
-        variables.HADOOP_CONF_DIR = "/etc/hadoop-conf/";
-      };
-    })
-
-  ];
+  config = mkIf cfg.gatewayRole.enable {
+    users.groups.hadoop = {
+      gid = config.ids.gids.hadoop;
+    };
+    environment = {
+      systemPackages = [ cfg.package ];
+      etc."hadoop-conf".source = let
+        hadoopConf = "${import ./conf.nix { inherit cfg pkgs lib; }}/";
+      in "${hadoopConf}";
+      variables.HADOOP_CONF_DIR = "/etc/hadoop-conf/";
+    };
+  };
 }
diff --git a/nixos/modules/services/cluster/hadoop/hdfs.nix b/nixos/modules/services/cluster/hadoop/hdfs.nix
index 61d9941298a..1725dc62d0c 100644
--- a/nixos/modules/services/cluster/hadoop/hdfs.nix
+++ b/nixos/modules/services/cluster/hadoop/hdfs.nix
@@ -51,6 +51,8 @@ let
           };
         };
 
+        services.hadoop.gatewayRole.enable = true;
+
         networking.firewall.allowedTCPPorts = mkIf
           ((builtins.hasAttr "openFirewall" serviceOptions) && serviceOptions.openFirewall)
           allowedTCPPorts;
@@ -145,17 +147,13 @@ in
       ];
     })
 
-    (mkIf
-      (
-        cfg.hdfs.namenode.enable || cfg.hdfs.datanode.enable || cfg.hdfs.journalnode.enable || cfg.hdfs.zkfc.enable
-      )
-      {
-        users.users.hdfs = {
-          description = "Hadoop HDFS user";
-          group = "hadoop";
-          uid = config.ids.uids.hdfs;
-        };
-      })
+    (mkIf cfg.gatewayRole.enable {
+      users.users.hdfs = {
+        description = "Hadoop HDFS user";
+        group = "hadoop";
+        uid = config.ids.uids.hdfs;
+      };
+    })
     (mkIf cfg.hdfs.httpfs.enable {
       users.users.httpfs = {
         description = "Hadoop HTTPFS user";
@@ -163,5 +161,6 @@ in
         isSystemUser = true;
       };
     })
+
   ];
 }
diff --git a/nixos/modules/services/cluster/hadoop/yarn.nix b/nixos/modules/services/cluster/hadoop/yarn.nix
index 90ae75a44b7..9f0d3f85db7 100644
--- a/nixos/modules/services/cluster/hadoop/yarn.nix
+++ b/nixos/modules/services/cluster/hadoop/yarn.nix
@@ -49,10 +49,7 @@ in
   };
 
   config = mkMerge [
-    (mkIf (
-        cfg.yarn.resourcemanager.enable || cfg.yarn.nodemanager.enable
-    ) {
-
+    (mkIf cfg.gatewayRole.enable {
       users.users.yarn = {
         description = "Hadoop YARN user";
         group = "hadoop";
@@ -74,6 +71,9 @@ in
           Restart = "always";
         };
       };
+
+      services.hadoop.gatewayRole.enable = true;
+
       networking.firewall.allowedTCPPorts = (mkIf cfg.yarn.resourcemanager.openFirewall [
         8088 # resourcemanager.webapp.address
         8030 # resourcemanager.scheduler.address
@@ -119,6 +119,8 @@ in
         };
       };
 
+      services.hadoop.gatewayRole.enable = true;
+
       networking.firewall.allowedTCPPortRanges = [
         (mkIf (cfg.yarn.nodemanager.openFirewall) {from = 1024; to = 65535;})
       ];
diff --git a/nixos/tests/hadoop/hadoop.nix b/nixos/tests/hadoop/hadoop.nix
index e84a56f302f..42c238ef853 100644
--- a/nixos/tests/hadoop/hadoop.nix
+++ b/nixos/tests/hadoop/hadoop.nix
@@ -145,7 +145,14 @@ import ../make-test-python.nix ({ package, ... }: {
           };
         };
       };
-    };
+      client = { options, ... }: {
+        services.hadoop = {
+          gatewayRole.enable = true;
+          inherit package coreSite hdfsSite;
+          yarnSite = options.services.hadoop.yarnSite.default // yarnSiteHA;
+        };
+      };
+  };
 
   testScript = ''
     start_all()
@@ -202,26 +209,26 @@ import ../make-test-python.nix ({ package, ... }: {
     # DN should have started by now, but confirm anyway
     dn1.wait_for_unit("hdfs-datanode")
     # Print states of namenodes
-    dn1.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState | systemd-cat")
+    client.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState | systemd-cat")
     # Wait for cluster to exit safemode
-    dn1.succeed("sudo -u hdfs hdfs dfsadmin -safemode wait")
-    dn1.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState | systemd-cat")
+    client.succeed("sudo -u hdfs hdfs dfsadmin -safemode wait")
+    client.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState | systemd-cat")
     # test R/W
-    dn1.succeed("echo testfilecontents | sudo -u hdfs hdfs dfs -put - /testfile")
-    assert "testfilecontents" in dn1.succeed("sudo -u hdfs hdfs dfs -cat /testfile")
+    client.succeed("echo testfilecontents | sudo -u hdfs hdfs dfs -put - /testfile")
+    assert "testfilecontents" in client.succeed("sudo -u hdfs hdfs dfs -cat /testfile")
 
     # Test NN failover
     nn1.succeed("systemctl stop hdfs-namenode")
-    assert "active" in dn1.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState")
-    dn1.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState | systemd-cat")
-    assert "testfilecontents" in dn1.succeed("sudo -u hdfs hdfs dfs -cat /testfile")
+    assert "active" in client.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState")
+    client.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState | systemd-cat")
+    assert "testfilecontents" in client.succeed("sudo -u hdfs hdfs dfs -cat /testfile")
 
     nn1.succeed("systemctl start hdfs-namenode")
     nn1.wait_for_open_port(9870)
     nn1.wait_for_open_port(8022)
     nn1.wait_for_open_port(8020)
-    assert "standby" in dn1.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState")
-    dn1.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState | systemd-cat")
+    assert "standby" in client.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState")
+    client.succeed("sudo -u hdfs hdfs haadmin -getAllServiceState | systemd-cat")
 
     #### YARN tests ####
 
@@ -237,21 +244,21 @@ import ../make-test-python.nix ({ package, ... }: {
     nm1.wait_for_unit("yarn-nodemanager")
     nm1.wait_for_open_port(8042)
     nm1.wait_for_open_port(8040)
-    nm1.wait_until_succeeds("yarn node -list | grep Nodes:1")
-    nm1.succeed("sudo -u yarn yarn rmadmin -getAllServiceState | systemd-cat")
-    nm1.succeed("sudo -u yarn yarn node -list | systemd-cat")
+    client.wait_until_succeeds("yarn node -list | grep Nodes:1")
+    client.succeed("sudo -u yarn yarn rmadmin -getAllServiceState | systemd-cat")
+    client.succeed("sudo -u yarn yarn node -list | systemd-cat")
 
     # Test RM failover
     rm1.succeed("systemctl stop yarn-resourcemanager")
-    assert "standby" not in nm1.succeed("sudo -u yarn yarn rmadmin -getAllServiceState")
-    nm1.succeed("sudo -u yarn yarn rmadmin -getAllServiceState | systemd-cat")
+    assert "standby" not in client.succeed("sudo -u yarn yarn rmadmin -getAllServiceState")
+    client.succeed("sudo -u yarn yarn rmadmin -getAllServiceState | systemd-cat")
     rm1.succeed("systemctl start yarn-resourcemanager")
     rm1.wait_for_unit("yarn-resourcemanager")
     rm1.wait_for_open_port(8088)
-    assert "standby" in nm1.succeed("sudo -u yarn yarn rmadmin -getAllServiceState")
-    nm1.succeed("sudo -u yarn yarn rmadmin -getAllServiceState | systemd-cat")
+    assert "standby" in client.succeed("sudo -u yarn yarn rmadmin -getAllServiceState")
+    client.succeed("sudo -u yarn yarn rmadmin -getAllServiceState | systemd-cat")
 
-    assert "Estimated value of Pi is" in nm1.succeed("HADOOP_USER_NAME=hdfs yarn jar $(readlink $(which yarn) | sed -r 's~bin/yarn~lib/hadoop-*/share/hadoop/mapreduce/hadoop-mapreduce-examples-*.jar~g') pi 2 10")
-    assert "SUCCEEDED" in nm1.succeed("yarn application -list -appStates FINISHED")
+    assert "Estimated value of Pi is" in client.succeed("HADOOP_USER_NAME=hdfs yarn jar $(readlink $(which yarn) | sed -r 's~bin/yarn~lib/hadoop-*/share/hadoop/mapreduce/hadoop-mapreduce-examples-*.jar~g') pi 2 10")
+    assert "SUCCEEDED" in client.succeed("yarn application -list -appStates FINISHED")
   '';
 })
-- 
cgit 1.4.1


From bef71d7c530aeecdcfb66290e6c0b4948d1fb223 Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Mon, 28 Feb 2022 20:41:43 +0530
Subject: nixos/hadoop: use CGroups to enforce container limits by default

---
 nixos/modules/services/cluster/hadoop/default.nix | 5 +++++
 nixos/modules/services/cluster/hadoop/yarn.nix    | 3 ++-
 nixos/tests/hadoop/yarn.nix                       | 4 ++--
 3 files changed, 9 insertions(+), 3 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/default.nix b/nixos/modules/services/cluster/hadoop/default.nix
index 5c7ea79c301..57b1d7a90d7 100644
--- a/nixos/modules/services/cluster/hadoop/default.nix
+++ b/nixos/modules/services/cluster/hadoop/default.nix
@@ -79,6 +79,10 @@ with lib;
         "yarn.nodemanager.log-dirs" = "/var/log/hadoop/yarn/nodemanager";
         "yarn.resourcemanager.bind-host" = "0.0.0.0";
         "yarn.resourcemanager.scheduler.class" = "org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FairScheduler";
+        "yarn.nodemanager.linux-container-executor.cgroups.hierarchy" = "/hadoop-yarn";
+        "yarn.nodemanager.linux-container-executor.resources-handler.class" = "org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler";
+        "yarn.nodemanager.linux-container-executor.cgroups.mount" = "true";
+        "yarn.nodemanager.linux-container-executor.cgroups.mount-path" = "/run/wrappers/yarn-nodemanager/cgroup";
       };
       type = types.attrsOf types.anything;
       example = literalExpression ''
@@ -124,6 +128,7 @@ with lib;
         "yarn.nodemanager.linux-container-executor.group"="hadoop";
         "min.user.id"=1000;
         "feature.terminal.enabled"=1;
+        "feature.mount-cgroup.enabled" = 1;
       };
       type = types.attrsOf types.anything;
       example = literalExpression ''
diff --git a/nixos/modules/services/cluster/hadoop/yarn.nix b/nixos/modules/services/cluster/hadoop/yarn.nix
index 9f0d3f85db7..373d8a70a12 100644
--- a/nixos/modules/services/cluster/hadoop/yarn.nix
+++ b/nixos/modules/services/cluster/hadoop/yarn.nix
@@ -101,8 +101,9 @@ in
           chown yarn:hadoop /var/log/hadoop/yarn/nodemanager
 
           # set up setuid container executor binary
+          umount /run/wrappers/yarn-nodemanager/cgroup/cpu || true
           rm -rf /run/wrappers/yarn-nodemanager/ || true
-          mkdir -p /run/wrappers/yarn-nodemanager/{bin,etc/hadoop}
+          mkdir -p /run/wrappers/yarn-nodemanager/{bin,etc/hadoop,cgroup/cpu}
           cp ${cfg.package}/lib/${cfg.package.untarDir}/bin/container-executor /run/wrappers/yarn-nodemanager/bin/
           chgrp hadoop /run/wrappers/yarn-nodemanager/bin/container-executor
           chmod 6050 /run/wrappers/yarn-nodemanager/bin/container-executor
diff --git a/nixos/tests/hadoop/yarn.nix b/nixos/tests/hadoop/yarn.nix
index ba39e257dc8..1bf8e3831f6 100644
--- a/nixos/tests/hadoop/yarn.nix
+++ b/nixos/tests/hadoop/yarn.nix
@@ -12,14 +12,14 @@ import ../make-test-python.nix ({ package, ... }: {
         };
       };
     };
-    nodemanager = { ... }: {
+    nodemanager = { options, lib, ... }: {
       services.hadoop = {
         inherit package;
         yarn.nodemanager = {
           enable = true;
           openFirewall = true;
         };
-        yarnSite = {
+        yarnSite = options.services.hadoop.yarnSite.default // {
           "yarn.resourcemanager.hostname" = "resourcemanager";
           "yarn.nodemanager.log-dirs" = "/tmp/userlogs";
         };
-- 
cgit 1.4.1


From e1017adb328da98e7942ec08354dd686653370a4 Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Wed, 2 Mar 2022 12:50:01 +0530
Subject: nixos/hadoop: add module options for commonly used service configs

---
 nixos/modules/services/cluster/hadoop/conf.nix    | 22 +++----
 nixos/modules/services/cluster/hadoop/default.nix | 67 ++++++++++++++++----
 nixos/modules/services/cluster/hadoop/hdfs.nix    | 52 ++++++++++++---
 nixos/modules/services/cluster/hadoop/yarn.nix    | 77 +++++++++++++++++++++--
 nixos/tests/hadoop/hadoop.nix                     | 21 ++-----
 nixos/tests/hadoop/hdfs.nix                       |  4 ++
 6 files changed, 194 insertions(+), 49 deletions(-)

(limited to 'nixos/modules/services/cluster/hadoop')

diff --git a/nixos/modules/services/cluster/hadoop/conf.nix b/nixos/modules/services/cluster/hadoop/conf.nix
index 0caec5cfc20..e3c26a0d550 100644
--- a/nixos/modules/services/cluster/hadoop/conf.nix
+++ b/nixos/modules/services/cluster/hadoop/conf.nix
@@ -1,6 +1,6 @@
 { cfg, pkgs, lib }:
 let
-  propertyXml = name: value: ''
+  propertyXml = name: value: lib.optionalString (value != null) ''
     <property>
       <name>${name}</name>
       <value>${builtins.toString value}</value>
@@ -29,16 +29,16 @@ let
     export HADOOP_LOG_DIR=/tmp/hadoop/$USER
   '';
 in
-pkgs.runCommand "hadoop-conf" {} ''
+pkgs.runCommand "hadoop-conf" {} (with cfg; ''
   mkdir -p $out/
-  cp ${siteXml "core-site.xml" cfg.coreSite}/* $out/
-  cp ${siteXml "hdfs-site.xml" cfg.hdfsSite}/* $out/
-  cp ${siteXml "mapred-site.xml" cfg.mapredSite}/* $out/
-  cp ${siteXml "yarn-site.xml" cfg.yarnSite}/* $out/
-  cp ${siteXml "httpfs-site.xml" cfg.httpfsSite}/* $out/
-  cp ${cfgFile "container-executor.cfg" cfg.containerExecutorCfg}/* $out/
+  cp ${siteXml "core-site.xml" (coreSite // coreSiteInternal)}/* $out/
+  cp ${siteXml "hdfs-site.xml" (hdfsSiteDefault // hdfsSite // hdfsSiteInternal)}/* $out/
+  cp ${siteXml "mapred-site.xml" (mapredSiteDefault // mapredSite)}/* $out/
+  cp ${siteXml "yarn-site.xml" (yarnSiteDefault // yarnSite // yarnSiteInternal)}/* $out/
+  cp ${siteXml "httpfs-site.xml" httpfsSite}/* $out/
+  cp ${cfgFile "container-executor.cfg" containerExecutorCfg}/* $out/
   cp ${pkgs.writeTextDir "hadoop-user-functions.sh" userFunctions}/* $out/
   cp ${pkgs.writeTextDir "hadoop-env.sh" hadoopEnv}/* $out/
-  cp ${cfg.log4jProperties} $out/log4j.properties
-  ${lib.concatMapStringsSep "\n" (dir: "cp -r ${dir}/* $out/") cfg.extraConfDirs}
-''
+  cp ${log4jProperties} $out/log4j.properties
+  ${lib.concatMapStringsSep "\n" (dir: "cp -r ${dir}/* $out/") extraConfDirs}
+'')
diff --git a/nixos/modules/services/cluster/hadoop/default.nix b/nixos/modules/services/cluster/hadoop/default.nix
index 57b1d7a90d7..a4fdea81037 100644
--- a/nixos/modules/services/cluster/hadoop/default.nix
+++ b/nixos/modules/services/cluster/hadoop/default.nix
@@ -21,25 +21,50 @@ with lib;
         <link xlink:href="https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-common/core-default.xml"/>
       '';
     };
+    coreSiteInternal = mkOption {
+      default = {};
+      type = types.attrsOf types.anything;
+      internal = true;
+      description = ''
+        Internal option to add configs to core-site.xml based on module options
+      '';
+    };
 
-    hdfsSite = mkOption {
+    hdfsSiteDefault = mkOption {
       default = {
         "dfs.namenode.rpc-bind-host" = "0.0.0.0";
         "dfs.namenode.http-address" = "0.0.0.0:9870";
+        "dfs.namenode.servicerpc-bind-host" = "0.0.0.0";
+        "dfs.namenode.http-bind-host" = "0.0.0.0";
       };
       type = types.attrsOf types.anything;
+      description = ''
+        Default options for hdfs-site.xml
+      '';
+    };
+    hdfsSite = mkOption {
+      default = {};
+      type = types.attrsOf types.anything;
       example = literalExpression ''
         {
           "dfs.nameservices" = "namenode1";
         }
       '';
       description = ''
-        Hadoop hdfs-site.xml definition
+        Additional options and overrides for hdfs-site.xml
         <link xlink:href="https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/hdfs-default.xml"/>
       '';
     };
+    hdfsSiteInternal = mkOption {
+      default = {};
+      type = types.attrsOf types.anything;
+      internal = true;
+      description = ''
+        Internal option to add configs to hdfs-site.xml based on module options
+      '';
+    };
 
-    mapredSite = mkOption {
+    mapredSiteDefault = mkOption {
       default = {
         "mapreduce.framework.name" = "yarn";
         "yarn.app.mapreduce.am.env" = "HADOOP_MAPRED_HOME=${cfg.package}/lib/${cfg.package.untarDir}";
@@ -55,18 +80,25 @@ with lib;
         }
       '';
       type = types.attrsOf types.anything;
+      description = ''
+        Default options for mapred-site.xml
+      '';
+    };
+    mapredSite = mkOption {
+      default = {};
+      type = types.attrsOf types.anything;
       example = literalExpression ''
-        options.services.hadoop.mapredSite.default // {
+        {
           "mapreduce.map.java.opts" = "-Xmx900m -XX:+UseParallelGC";
         }
       '';
       description = ''
-        Hadoop mapred-site.xml definition
+        Additional options and overrides for mapred-site.xml
         <link xlink:href="https://hadoop.apache.org/docs/current/hadoop-mapreduce-client/hadoop-mapreduce-client-core/mapred-default.xml"/>
       '';
     };
 
-    yarnSite = mkOption {
+    yarnSiteDefault = mkOption {
       default = {
         "yarn.nodemanager.admin-env" = "PATH=$PATH";
         "yarn.nodemanager.aux-services" = "mapreduce_shuffle";
@@ -79,22 +111,33 @@ with lib;
         "yarn.nodemanager.log-dirs" = "/var/log/hadoop/yarn/nodemanager";
         "yarn.resourcemanager.bind-host" = "0.0.0.0";
         "yarn.resourcemanager.scheduler.class" = "org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FairScheduler";
-        "yarn.nodemanager.linux-container-executor.cgroups.hierarchy" = "/hadoop-yarn";
-        "yarn.nodemanager.linux-container-executor.resources-handler.class" = "org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler";
-        "yarn.nodemanager.linux-container-executor.cgroups.mount" = "true";
-        "yarn.nodemanager.linux-container-executor.cgroups.mount-path" = "/run/wrappers/yarn-nodemanager/cgroup";
       };
       type = types.attrsOf types.anything;
+      description = ''
+        Default options for yarn-site.xml
+      '';
+    };
+    yarnSite = mkOption {
+      default = {};
+      type = types.attrsOf types.anything;
       example = literalExpression ''
-        options.services.hadoop.yarnSite.default // {
+        {
           "yarn.resourcemanager.hostname" = "''${config.networking.hostName}";
         }
       '';
       description = ''
-        Hadoop yarn-site.xml definition
+        Additional options and overrides for yarn-site.xml
         <link xlink:href="https://hadoop.apache.org/docs/current/hadoop-yarn/hadoop-yarn-common/yarn-default.xml"/>
       '';
     };
+    yarnSiteInternal = mkOption {
+      default = {};
+      type = types.attrsOf types.anything;
+      internal = true;
+      description = ''
+        Internal option to add configs to yarn-site.xml based on module options
+      '';
+    };
 
     httpfsSite = mkOption {
       default = { };
diff --git a/nixos/modules/services/cluster/hadoop/hdfs.nix b/nixos/modules/services/cluster/hadoop/hdfs.nix
index 1725dc62d0c..325a002ad32 100644
--- a/nixos/modules/services/cluster/hadoop/hdfs.nix
+++ b/nixos/modules/services/cluster/hadoop/hdfs.nix
@@ -7,7 +7,7 @@ let
   hadoopConf = "${import ./conf.nix { inherit cfg pkgs lib; }}/";
 
   # Generator for HDFS service options
-  hadoopServiceOption = { serviceName, firewallOption ? true }: {
+  hadoopServiceOption = { serviceName, firewallOption ? true, extraOpts ? null }: {
     enable = mkEnableOption serviceName;
     restartIfChanged = mkOption {
       type = types.bool;
@@ -19,13 +19,27 @@ let
       '';
       default = false;
     };
+    extraFlags = mkOption{
+      type = with types; listOf str;
+      default = [];
+      description = "Extra command line flags to pass to ${serviceName}";
+      example = [
+        "-Dcom.sun.management.jmxremote"
+        "-Dcom.sun.management.jmxremote.port=8010"
+      ];
+    };
+    extraEnv = mkOption{
+      type = with types; attrsOf str;
+      default = {};
+      description = "Extra environment variables for ${serviceName}";
+    };
   } // (optionalAttrs firewallOption {
     openFirewall = mkOption {
       type = types.bool;
       default = false;
       description = "Open firewall ports for ${serviceName}.";
     };
-  });
+  }) // (optionalAttrs (extraOpts != null) extraOpts);
 
   # Generator for HDFS service configs
   hadoopServiceConfig =
@@ -36,17 +50,19 @@ let
     , allowedTCPPorts ? [ ]
     , preStart ? ""
     , environment ? { }
+    , extraConfig ? { }
     }: (
 
-      mkIf serviceOptions.enable {
+      mkIf serviceOptions.enable ( mkMerge [{
         systemd.services."hdfs-${toLower name}" = {
-          inherit description preStart environment;
+          inherit description preStart;
+          environment = environment // serviceOptions.extraEnv;
           wantedBy = [ "multi-user.target" ];
           inherit (serviceOptions) restartIfChanged;
           serviceConfig = {
             inherit User;
             SyslogIdentifier = "hdfs-${toLower name}";
-            ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} ${toLower name}";
+            ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} ${toLower name} ${escapeShellArgs serviceOptions.extraFlags}";
             Restart = "always";
           };
         };
@@ -56,7 +72,7 @@ let
         networking.firewall.allowedTCPPorts = mkIf
           ((builtins.hasAttr "openFirewall" serviceOptions) && serviceOptions.openFirewall)
           allowedTCPPorts;
-      }
+      } extraConfig])
     );
 
 in
@@ -77,7 +93,27 @@ in
       };
     };
 
-    datanode = hadoopServiceOption { serviceName = "HDFS DataNode"; };
+    datanode = hadoopServiceOption { serviceName = "HDFS DataNode"; } // {
+      dataDirs = mkOption {
+        default = null;
+        description = "Tier and path definitions for datanode storage.";
+        type = with types; nullOr (listOf (submodule {
+          options = {
+            type = mkOption {
+              type = enum [ "SSD" "DISK" "ARCHIVE" "RAM_DISK" ];
+              description = ''
+                Storage types ([SSD]/[DISK]/[ARCHIVE]/[RAM_DISK]) for HDFS storage policies.
+              '';
+            };
+            path = mkOption {
+              type = path;
+              example = [ "/var/lib/hadoop/hdfs/dn" ];
+              description = "Determines where on the local filesystem a data node should store its blocks.";
+            };
+          };
+        }));
+      };
+    };
 
     journalnode = hadoopServiceOption { serviceName = "HDFS JournalNode"; };
 
@@ -122,6 +158,8 @@ in
         50010 # datanode.address
         50020 # datanode.ipc.address
       ];
+      extraConfig.services.hadoop.hdfsSiteInternal."dfs.datanode.data.dir" = let d = cfg.hdfs.datanode.dataDirs; in
+        if (d!= null) then (concatMapStringsSep "," (x: "["+x.type+"]file://"+x.path) cfg.hdfs.datanode.dataDirs) else d;
     })
 
     (hadoopServiceConfig {
diff --git a/nixos/modules/services/cluster/hadoop/yarn.nix b/nixos/modules/services/cluster/hadoop/yarn.nix
index 373d8a70a12..74e16bdec68 100644
--- a/nixos/modules/services/cluster/hadoop/yarn.nix
+++ b/nixos/modules/services/cluster/hadoop/yarn.nix
@@ -13,12 +13,27 @@ let
     '';
     default = false;
   };
+  extraFlags = mkOption{
+    type = with types; listOf str;
+    default = [];
+    description = "Extra command line flags to pass to the service";
+    example = [
+      "-Dcom.sun.management.jmxremote"
+      "-Dcom.sun.management.jmxremote.port=8010"
+    ];
+  };
+  extraEnv = mkOption{
+    type = with types; attrsOf str;
+    default = {};
+    description = "Extra environment variables";
+  };
 in
 {
   options.services.hadoop.yarn = {
     resourcemanager = {
       enable = mkEnableOption "Hadoop YARN ResourceManager";
-      inherit restartIfChanged;
+      inherit restartIfChanged extraFlags extraEnv;
+
       openFirewall = mkOption {
         type = types.bool;
         default = false;
@@ -29,7 +44,46 @@ in
     };
     nodemanager = {
       enable = mkEnableOption "Hadoop YARN NodeManager";
-      inherit restartIfChanged;
+      inherit restartIfChanged extraFlags extraEnv;
+
+      resource = {
+        cpuVCores = mkOption {
+          description = "Number of vcores that can be allocated for containers.";
+          type = with types; nullOr ints.positive;
+          default = null;
+        };
+        maximumAllocationVCores = mkOption {
+          description = "The maximum virtual CPU cores any container can be allocated.";
+          type = with types; nullOr ints.positive;
+          default = null;
+        };
+        memoryMB = mkOption {
+          description = "Amount of physical memory, in MB, that can be allocated for containers.";
+          type = with types; nullOr ints.positive;
+          default = null;
+        };
+        maximumAllocationMB = mkOption {
+          description = "The maximum physical memory any container can be allocated.";
+          type = with types; nullOr ints.positive;
+          default = null;
+        };
+      };
+
+      useCGroups = mkOption {
+        type = types.bool;
+        default = true;
+        description = ''
+          Use cgroups to enforce resource limits on containers
+        '';
+      };
+
+      localDir = mkOption {
+        description = "List of directories to store localized files in.";
+        type = with types; nullOr (listOf path);
+        example = [ "/var/lib/hadoop/yarn/nm" ];
+        default = null;
+      };
+
       addBinBash = mkOption {
         type = types.bool;
         default = true;
@@ -62,12 +116,13 @@ in
         description = "Hadoop YARN ResourceManager";
         wantedBy = [ "multi-user.target" ];
         inherit (cfg.yarn.resourcemanager) restartIfChanged;
+        environment = cfg.yarn.resourcemanager.extraEnv;
 
         serviceConfig = {
           User = "yarn";
           SyslogIdentifier = "yarn-resourcemanager";
           ExecStart = "${cfg.package}/bin/yarn --config ${hadoopConf} " +
-                      " resourcemanager";
+                      " resourcemanager ${escapeShellArgs cfg.yarn.resourcemanager.extraFlags}";
           Restart = "always";
         };
       };
@@ -94,6 +149,7 @@ in
         description = "Hadoop YARN NodeManager";
         wantedBy = [ "multi-user.target" ];
         inherit (cfg.yarn.nodemanager) restartIfChanged;
+        environment = cfg.yarn.nodemanager.extraEnv;
 
         preStart = ''
           # create log dir
@@ -115,13 +171,26 @@ in
           SyslogIdentifier = "yarn-nodemanager";
           PermissionsStartOnly = true;
           ExecStart = "${cfg.package}/bin/yarn --config ${hadoopConf} " +
-                      " nodemanager";
+                      " nodemanager ${escapeShellArgs cfg.yarn.nodemanager.extraFlags}";
           Restart = "always";
         };
       };
 
       services.hadoop.gatewayRole.enable = true;
 
+      services.hadoop.yarnSiteInternal = with cfg.yarn.nodemanager; {
+        "yarn.nodemanager.local-dirs" = localDir;
+        "yarn.scheduler.maximum-allocation-vcores" = resource.maximumAllocationVCores;
+        "yarn.scheduler.maximum-allocation-mb" = resource.maximumAllocationMB;
+        "yarn.nodemanager.resource.cpu-vcores" = resource.cpuVCores;
+        "yarn.nodemanager.resource.memory-mb" = resource.memoryMB;
+      } // mkIf useCGroups {
+        "yarn.nodemanager.linux-container-executor.cgroups.hierarchy" = "/hadoop-yarn";
+        "yarn.nodemanager.linux-container-executor.resources-handler.class" = "org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler";
+        "yarn.nodemanager.linux-container-executor.cgroups.mount" = "true";
+        "yarn.nodemanager.linux-container-executor.cgroups.mount-path" = "/run/wrappers/yarn-nodemanager/cgroup";
+      };
+
       networking.firewall.allowedTCPPortRanges = [
         (mkIf (cfg.yarn.nodemanager.openFirewall) {from = 1024; to = 65535;})
       ];
diff --git a/nixos/tests/hadoop/hadoop.nix b/nixos/tests/hadoop/hadoop.nix
index 42c238ef853..b132f4fa58b 100644
--- a/nixos/tests/hadoop/hadoop.nix
+++ b/nixos/tests/hadoop/hadoop.nix
@@ -10,15 +10,10 @@ import ../make-test-python.nix ({ package, ... }: {
         "fs.defaultFS" = "hdfs://ns1";
       };
       hdfsSite = {
-        "dfs.namenode.rpc-bind-host" = "0.0.0.0";
-        "dfs.namenode.http-bind-host" = "0.0.0.0";
-        "dfs.namenode.servicerpc-bind-host" = "0.0.0.0";
-
         # HA Quorum Journal Manager configuration
         "dfs.nameservices" = "ns1";
         "dfs.ha.namenodes.ns1" = "nn1,nn2";
-        "dfs.namenode.shared.edits.dir.ns1.nn1" = "qjournal://jn1:8485;jn2:8485;jn3:8485/ns1";
-        "dfs.namenode.shared.edits.dir.ns1.nn2" = "qjournal://jn1:8485;jn2:8485;jn3:8485/ns1";
+        "dfs.namenode.shared.edits.dir.ns1" = "qjournal://jn1:8485;jn2:8485;jn3:8485/ns1";
         "dfs.namenode.rpc-address.ns1.nn1" = "nn1:8020";
         "dfs.namenode.rpc-address.ns1.nn2" = "nn2:8020";
         "dfs.namenode.servicerpc-address.ns1.nn1" = "nn1:8022";
@@ -32,7 +27,7 @@ import ../make-test-python.nix ({ package, ... }: {
         "dfs.ha.fencing.methods" = "shell(true)";
         "ha.zookeeper.quorum" = "zk1:2181";
       };
-      yarnSiteHA = {
+      yarnSite = {
         "yarn.resourcemanager.zk-address" = "zk1:2181";
         "yarn.resourcemanager.ha.enabled" = "true";
         "yarn.resourcemanager.ha.rm-ids" = "rm1,rm2";
@@ -116,8 +111,7 @@ import ../make-test-python.nix ({ package, ... }: {
       # YARN cluster
       rm1 = { options, ... }: {
         services.hadoop = {
-          inherit package coreSite hdfsSite;
-          yarnSite = options.services.hadoop.yarnSite.default // yarnSiteHA;
+          inherit package coreSite hdfsSite yarnSite;
           yarn.resourcemanager = {
             enable = true;
             openFirewall = true;
@@ -126,8 +120,7 @@ import ../make-test-python.nix ({ package, ... }: {
       };
       rm2 = { options, ... }: {
         services.hadoop = {
-          inherit package coreSite hdfsSite;
-          yarnSite = options.services.hadoop.yarnSite.default // yarnSiteHA;
+          inherit package coreSite hdfsSite yarnSite;
           yarn.resourcemanager = {
             enable = true;
             openFirewall = true;
@@ -137,8 +130,7 @@ import ../make-test-python.nix ({ package, ... }: {
       nm1 = { options, ... }: {
         virtualisation.memorySize = 2048;
         services.hadoop = {
-          inherit package coreSite hdfsSite;
-          yarnSite = options.services.hadoop.yarnSite.default // yarnSiteHA;
+          inherit package coreSite hdfsSite yarnSite;
           yarn.nodemanager = {
             enable = true;
             openFirewall = true;
@@ -148,8 +140,7 @@ import ../make-test-python.nix ({ package, ... }: {
       client = { options, ... }: {
         services.hadoop = {
           gatewayRole.enable = true;
-          inherit package coreSite hdfsSite;
-          yarnSite = options.services.hadoop.yarnSite.default // yarnSiteHA;
+          inherit package coreSite hdfsSite yarnSite;
         };
       };
   };
diff --git a/nixos/tests/hadoop/hdfs.nix b/nixos/tests/hadoop/hdfs.nix
index cc70fb8ecaf..9415500463d 100644
--- a/nixos/tests/hadoop/hdfs.nix
+++ b/nixos/tests/hadoop/hdfs.nix
@@ -35,6 +35,10 @@ with lib;
         hdfs.datanode = {
           enable = true;
           openFirewall = true;
+          dataDirs = [{
+            type = "DISK";
+            path = "/tmp/dn1";
+          }];
         };
         inherit coreSite;
       };
-- 
cgit 1.4.1