From 9ca4363191f05147d0bd23ffb61838d9c1ee890e Mon Sep 17 00:00:00 2001
From: illustris <rharikrishnan95@gmail.com>
Date: Tue, 2 Nov 2021 18:34:04 +0530
Subject: nixos/hadoop: add HTTPFS

---
 nixos/modules/services/cluster/hadoop/conf.nix    |  1 +
 nixos/modules/services/cluster/hadoop/default.nix | 14 ++++++-
 nixos/modules/services/cluster/hadoop/hdfs.nix    | 50 ++++++++++++++++++++++-
 nixos/tests/hadoop/hdfs.nix                       | 23 +++++++----
 4 files changed, 78 insertions(+), 10 deletions(-)

(limited to 'nixos')

diff --git a/nixos/modules/services/cluster/hadoop/conf.nix b/nixos/modules/services/cluster/hadoop/conf.nix
index 69472408cab..0caec5cfc20 100644
--- a/nixos/modules/services/cluster/hadoop/conf.nix
+++ b/nixos/modules/services/cluster/hadoop/conf.nix
@@ -35,6 +35,7 @@ pkgs.runCommand "hadoop-conf" {} ''
   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 ${pkgs.writeTextDir "hadoop-user-functions.sh" userFunctions}/* $out/
   cp ${pkgs.writeTextDir "hadoop-env.sh" hadoopEnv}/* $out/
diff --git a/nixos/modules/services/cluster/hadoop/default.nix b/nixos/modules/services/cluster/hadoop/default.nix
index da3e47b95d4..e4dcfde80fb 100644
--- a/nixos/modules/services/cluster/hadoop/default.nix
+++ b/nixos/modules/services/cluster/hadoop/default.nix
@@ -70,6 +70,17 @@ with lib;
       description = "Hadoop yarn-site.xml definition";
     };
 
+    httpfsSite = mkOption {
+      default = { };
+      type = types.attrsOf types.anything;
+      example = literalExpression ''
+        {
+          "hadoop.http.max.threads" = 500;
+        }
+      '';
+      description = "Hadoop httpfs-site.xml definition";
+    };
+
     log4jProperties = mkOption {
       default = "${cfg.package}/lib/${cfg.package.untarDir}/etc/hadoop/log4j.properties";
       type = types.path;
@@ -118,7 +129,8 @@ with lib;
 
   config = mkMerge [
     (mkIf (builtins.hasAttr "yarn" config.users.users ||
-           builtins.hasAttr "hdfs" config.users.users) {
+           builtins.hasAttr "hdfs" config.users.users ||
+           builtins.hasAttr "httpfs" config.users.users) {
       users.groups.hadoop = {
         gid = config.ids.gids.hadoop;
       };
diff --git a/nixos/modules/services/cluster/hadoop/hdfs.nix b/nixos/modules/services/cluster/hadoop/hdfs.nix
index 961aa35a4b1..11b855b0c71 100644
--- a/nixos/modules/services/cluster/hadoop/hdfs.nix
+++ b/nixos/modules/services/cluster/hadoop/hdfs.nix
@@ -86,6 +86,23 @@ in
       };
       inherit restartIfChanged;
     };
+    httpfs = {
+      enabled = mkOption {
+        type = types.bool;
+        default = false;
+        description = ''
+          Whether to run the HDFS httpfs failover controller
+        '';
+      };
+      tempPath = mkOption {
+        type = types.path;
+        default = "/tmp/hadoop/httpfs";
+        description = ''
+          HTTPFS_TEMP path used by HTTPFS
+        '';
+      };
+      inherit restartIfChanged;
+    };
   };
 
   config = mkMerge [
@@ -166,6 +183,31 @@ in
         };
       };
     })
+    (mkIf cfg.hdfs.httpfs.enabled {
+      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 $HTTPFS_TEMP
+        '';
+
+        serviceConfig = {
+          User = "httpfs";
+          SyslogIdentifier = "hdfs-httpfs";
+          ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} httpfs";
+          Restart = "always";
+        };
+      };
+      networking.firewall.allowedTCPPorts = (mkIf cfg.hdfs.datanode.openFirewall [
+        14000 # httpfs.http.port
+      ]);
+    })
     (mkIf (
         cfg.hdfs.namenode.enabled || cfg.hdfs.datanode.enabled || cfg.hdfs.journalnode.enabled || cfg.hdfs.zkfc.enabled
     ) {
@@ -175,6 +217,12 @@ in
         uid = config.ids.uids.hdfs;
       };
     })
-
+    (mkIf cfg.hdfs.httpfs.enabled {
+      users.users.httpfs = {
+        description = "Hadoop HTTPFS user";
+        group = "hadoop";
+        isSystemUser = true;
+      };
+    })
   ];
 }
diff --git a/nixos/tests/hadoop/hdfs.nix b/nixos/tests/hadoop/hdfs.nix
index 9bc32cc7f8a..e2cef32d05d 100644
--- a/nixos/tests/hadoop/hdfs.nix
+++ b/nixos/tests/hadoop/hdfs.nix
@@ -2,19 +2,20 @@
 import ../make-test-python.nix ({...}: {
   nodes = {
     namenode = {pkgs, ...}: {
+      virtualisation.memorySize = 1024;
       services.hadoop = {
         package = pkgs.hadoop;
-        hdfs.namenode = {
-          enabled = true;
-          formatOnInit = true;
+        hdfs = {
+          namenode = {
+            enabled = true;
+            formatOnInit = true;
+          };
+          httpfs.enabled = true;
         };
         coreSite = {
           "fs.defaultFS" = "hdfs://namenode:8020";
-        };
-        hdfsSite = {
-          "dfs.replication" = 1;
-          "dfs.namenode.rpc-bind-host" = "0.0.0.0";
-          "dfs.namenode.http-bind-host" = "0.0.0.0";
+          "hadoop.proxyuser.httpfs.groups" = "*";
+          "hadoop.proxyuser.httpfs.hosts" = "*";
         };
       };
     };
@@ -24,6 +25,8 @@ import ../make-test-python.nix ({...}: {
         hdfs.datanode.enabled = true;
         coreSite = {
           "fs.defaultFS" = "hdfs://namenode:8020";
+          "hadoop.proxyuser.httpfs.groups" = "*";
+          "hadoop.proxyuser.httpfs.hosts" = "*";
         };
       };
     };
@@ -49,5 +52,9 @@ import ../make-test-python.nix ({...}: {
     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")
+
+    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