commit 9ca4363191f05147d0bd23ffb61838d9c1ee890e · pyrox.dev/nixpkgs

nixos/modules/services/cluster/hadoop/conf.nix

···

       35
       35
        
         cp ${siteXml "hdfs-site.xml" cfg.hdfsSite}/* $out/

     

       36
       36
        
         cp ${siteXml "mapred-site.xml" cfg.mapredSite}/* $out/

     

       37
       37
        
         cp ${siteXml "yarn-site.xml" cfg.yarnSite}/* $out/

     

       38
       38
       +
         cp ${siteXml "httpfs-site.xml" cfg.httpfsSite}/* $out/

     

       38
       39
        
         cp ${cfgFile "container-executor.cfg" cfg.containerExecutorCfg}/* $out/

     

       39
       40
        
         cp ${pkgs.writeTextDir "hadoop-user-functions.sh" userFunctions}/* $out/

     

       40
       41
        
         cp ${pkgs.writeTextDir "hadoop-env.sh" hadoopEnv}/* $out/

+13 -1

nixos/modules/services/cluster/hadoop/default.nix

···

       70
       70
        
             description = "Hadoop yarn-site.xml definition";

     

       71
       71
        
           };

     

       72
       72
        
       

     

       73
       73
       +
           httpfsSite = mkOption {

     

       74
       74
       +
             default = { };

     

       75
       75
       +
             type = types.attrsOf types.anything;

     

       76
       76
       +
             example = literalExpression ''

     

       77
       77
       +
               {

     

       78
       78
       +
                 "hadoop.http.max.threads" = 500;

     

       79
       79
       +
               }

     

       80
       80
       +
             '';

     

       81
       81
       +
             description = "Hadoop httpfs-site.xml definition";

     

       82
       82
       +
           };

     

       83
       83
       +
       

     

       73
       84
        
           log4jProperties = mkOption {

     

       74
       85
        
             default = "${cfg.package}/lib/${cfg.package.untarDir}/etc/hadoop/log4j.properties";

     

       75
       86
        
             type = types.path;

     
···

       118
       129
        
       

     

       119
       130
        
         config = mkMerge [

     

       120
       131
        
           (mkIf (builtins.hasAttr "yarn" config.users.users ||

     

       121
       121
       -
                  builtins.hasAttr "hdfs" config.users.users) {

     

       132
       132
       +
                  builtins.hasAttr "hdfs" config.users.users ||

     

       133
       133
       +
                  builtins.hasAttr "httpfs" config.users.users) {

     

       122
       134
        
             users.groups.hadoop = {

     

       123
       135
        
               gid = config.ids.gids.hadoop;

     

       124
       136
        
             };

+49 -1

nixos/modules/services/cluster/hadoop/hdfs.nix

···

       86
       86
        
             };

     

       87
       87
        
             inherit restartIfChanged;

     

       88
       88
        
           };

     

       89
       89
       +
           httpfs = {

     

       90
       90
       +
             enabled = mkOption {

     

       91
       91
       +
               type = types.bool;

     

       92
       92
       +
               default = false;

     

       93
       93
       +
               description = ''

     

       94
       94
       +
                 Whether to run the HDFS httpfs failover controller

     

       95
       95
       +
               '';

     

       96
       96
       +
             };

     

       97
       97
       +
             tempPath = mkOption {

     

       98
       98
       +
               type = types.path;

     

       99
       99
       +
               default = "/tmp/hadoop/httpfs";

     

       100
       100
       +
               description = ''

     

       101
       101
       +
                 HTTPFS_TEMP path used by HTTPFS

     

       102
       102
       +
               '';

     

       103
       103
       +
             };

     

       104
       104
       +
             inherit restartIfChanged;

     

       105
       105
       +
           };

     

       89
       106
        
         };

     

       90
       107
        
       

     

       91
       108
        
         config = mkMerge [

     
···

       166
       183
        
               };

     

       167
       184
        
             };

     

       168
       185
        
           })

     

       186
       186
       +
           (mkIf cfg.hdfs.httpfs.enabled {

     

       187
       187
       +
             systemd.services.hdfs-httpfs = {

     

       188
       188
       +
               description = "Hadoop httpfs";

     

       189
       189
       +
               wantedBy = [ "multi-user.target" ];

     

       190
       190
       +
               inherit (cfg.hdfs.httpfs) restartIfChanged;

     

       191
       191
       +
       

     

       192
       192
       +
               environment = {

     

       193
       193
       +
                 HTTPFS_TEMP = cfg.hdfs.httpfs.tempPath;

     

       194
       194
       +
               };

     

       195
       195
       +
       

     

       196
       196
       +
               preStart = ''

     

       197
       197
       +
                 mkdir -p $HTTPFS_TEMP

     

       198
       198
       +
               '';

     

       199
       199
       +
       

     

       200
       200
       +
               serviceConfig = {

     

       201
       201
       +
                 User = "httpfs";

     

       202
       202
       +
                 SyslogIdentifier = "hdfs-httpfs";

     

       203
       203
       +
                 ExecStart = "${cfg.package}/bin/hdfs --config ${hadoopConf} httpfs";

     

       204
       204
       +
                 Restart = "always";

     

       205
       205
       +
               };

     

       206
       206
       +
             };

     

       207
       207
       +
             networking.firewall.allowedTCPPorts = (mkIf cfg.hdfs.datanode.openFirewall [

     

       208
       208
       +
               14000 # httpfs.http.port

     

       209
       209
       +
             ]);

     

       210
       210
       +
           })

     

       169
       211
        
           (mkIf (

     

       170
       212
        
               cfg.hdfs.namenode.enabled || cfg.hdfs.datanode.enabled || cfg.hdfs.journalnode.enabled || cfg.hdfs.zkfc.enabled

     

       171
       213
        
           ) {

     
···

       175
       217
        
               uid = config.ids.uids.hdfs;

     

       176
       218
        
             };

     

       177
       219
        
           })

     

       178
       178
       -
       

     

       220
       220
       +
           (mkIf cfg.hdfs.httpfs.enabled {

     

       221
       221
       +
             users.users.httpfs = {

     

       222
       222
       +
               description = "Hadoop HTTPFS user";

     

       223
       223
       +
               group = "hadoop";

     

       224
       224
       +
               isSystemUser = true;

     

       225
       225
       +
             };

     

       226
       226
       +
           })

     

       179
       227
        
         ];

     

       180
       228
        
       }

+15 -8

nixos/tests/hadoop/hdfs.nix

···

       2
       2
        
       import ../make-test-python.nix ({...}: {

     

       3
       3
        
         nodes = {

     

       4
       4
        
           namenode = {pkgs, ...}: {

     

       5
       5
       +
             virtualisation.memorySize = 1024;

     

       5
       6
        
             services.hadoop = {

     

       6
       7
        
               package = pkgs.hadoop;

     

       7
       7
       -
               hdfs.namenode = {

     

       8
       8
       -
                 enabled = true;

     

       9
       9
       -
                 formatOnInit = true;

     

       8
       8
       +
               hdfs = {

     

       9
       9
       +
                 namenode = {

     

       10
       10
       +
                   enabled = true;

     

       11
       11
       +
                   formatOnInit = true;

     

       12
       12
       +
                 };

     

       13
       13
       +
                 httpfs.enabled = true;

     

       10
       14
        
               };

     

       11
       15
        
               coreSite = {

     

       12
       16
        
                 "fs.defaultFS" = "hdfs://namenode:8020";

     

       13
       13
       -
               };

     

       14
       14
       -
               hdfsSite = {

     

       15
       15
       -
                 "dfs.replication" = 1;

     

       16
       16
       -
                 "dfs.namenode.rpc-bind-host" = "0.0.0.0";

     

       17
       17
       -
                 "dfs.namenode.http-bind-host" = "0.0.0.0";

     

       17
       17
       +
                 "hadoop.proxyuser.httpfs.groups" = "*";

     

       18
       18
       +
                 "hadoop.proxyuser.httpfs.hosts" = "*";

     

       18
       19
        
               };

     

       19
       20
        
             };

     

       20
       21
        
           };

     
···

       24
       25
        
               hdfs.datanode.enabled = true;

     

       25
       26
        
               coreSite = {

     

       26
       27
        
                 "fs.defaultFS" = "hdfs://namenode:8020";

     

       28
       28
       +
                 "hadoop.proxyuser.httpfs.groups" = "*";

     

       29
       29
       +
                 "hadoop.proxyuser.httpfs.hosts" = "*";

     

       27
       30
        
               };

     

       28
       31
        
             };

     

       29
       32
        
           };

     
···

       49
       52
        
           datanode.succeed("sudo -u hdfs hdfs dfsadmin -safemode wait")

     

       50
       53
        
           datanode.succeed("echo testfilecontents | sudo -u hdfs hdfs dfs -put - /testfile")

     

       51
       54
        
           assert "testfilecontents" in datanode.succeed("sudo -u hdfs hdfs dfs -cat /testfile")

     

       55
       55
       +
       

     

       56
       56
       +
           namenode.wait_for_unit("hdfs-httpfs")

     

       57
       57
       +
           namenode.wait_for_open_port(14000)

     

       58
       58
       +
           assert "testfilecontents" in datanode.succeed("curl -f \"http://namenode:14000/webhdfs/v1/testfile?user.name=hdfs&op=OPEN\" 2>&1")

     

       52
       59
        
         '';

     

       53
       60
        
       })