File and Object Storage

How to establish Ambari & HDP with GPFS transperency

By MIN CAI posted 16 days ago


How to establish Ambari & HDP with GPFS transperency

[Lab Environment]

  • 3 VM nodes with rhel 7.6
  • each with 16G memory, 100G disks for root disk and another 100G disks for GPFS NSD.
  • HDP-
  • ambari2.7.5
  • Spectrum_Scale 5.1.0
  • Spectrum_Scale_Management_Pack_MPACK_for_Ambari-


BDA Info center, overall installation guide:

Ambari Installaton guide from cloudera:

HDP configuration and install all related services guide:

MPack installation guide:

[Overall steps]

  1. Establish yum http repo for ambari/HDP/GPFS
  2. Install ambari
  3. Setup HDP repo and install all services from HDP
  4. Env ready, and install GPFS MPack to change community HDFS to HDFS tranparency

[Step 1. Establish yum http repo for ambari/HDP/GPFS]

a) Install and Create an HTTP server:

yum install -y httpd

b) Download and put ambari/HDP/GPFS packages under default http server path:

[root@hadoop-transparency1 ~]#  ll /var/www/html/
total 4
drwxrwxrwx.  6 root root 4096 Jun 26 09:52 ambari
drwxrwxrwx.  7 root root 4096 Jun 26 21:25 gpfs_rpms
drwxrwxrwx.  3 root root 4096 Jun 26 21:24 hdfs_rpms
drwxrwxrwx. 32 root root 4096 Jun  9 09:23 HDP-
drwxrwxrwx.  5 root root 4096 Jun  9 22:40 HDP-UTILS-
drwxrwxrwx.  3 root root 4096 Jun 22 00:52 libtirpc-devel
drwxrwxrwx.  2 root root 4096 Jun  9 22:38 repos

c) Start httpd service

systemctl start httpd

d) Disable SELinux to avoid permission issue:

setenforce 0

e) Verify if it can be reached via browser properly:

d) Configure yum on all 3 nodes:





list yum repolist:

[root@ambari1 ~]# yum repolist

Loaded plugins: langpacks, product-id, search-disabled-repos, subscription-manager
This system is not registered with an entitlement server. You can use subscription-manager to register.
Repository media is listed more than once in the configuration
repo id                                repo name                     status
!GPFS                                  gpfs-4.2.0                       22
!HDP-3.1-repo-1                        HDP-3.1-repo-1                  201
!HDP-UTILS-             HDP-UTILS-        16
!ambari-                        ambari2.7.5.0                    17
!media                                 ftp                           5,152
repolist: 5,625

[Step 2. Install ambari]

yum install ambari-server

after installed successfully run:

[root@ambari1 ~]# ambari-server start
Using python  /usr/bin/python                                                                                                                                                                                                                 
Starting ambari-server                                                                                                                                                                                                                        
Ambari Server running with administrator privileges.                                                                                                                                                                                          
Organizing resource files at /var/lib/ambari-server/resources...                                                                                                                                                                              
Ambari database consistency check started...                                                                                                                                                                                                  
Server PID at: /var/run/ambari-server/                                                                                                                                                                                       
Server out at: /var/log/ambari-server/ambari-server.out                                                                                                                                                                                       
Server log at: /var/log/ambari-server/ambari-server.log                                                                                                                                                                                       
Waiting for server start......................                                                                                                                                                                                                
Server started listening on 8080                                                                                                                                                                                                              
Ambari Server 'start' completed successfully.   

access ambari gui and launch install wizard to Install HDP:

[Step 3. Setup HDP repo and install all services from HDP]

Follow the installation wizard to set your own cluster:

Specify the HDP and HDP utils repo, which we already prepared, all the service installation rpm will fetch from there:


Provide private key to all target hosts:

Choose service to install, no special, but about Ranger and Ranger kms(Authentication module), you need to install mysql and pre-create the db user and use this user to perform all the db operation.

The steps is follow:

1) Install mariadb
yum install mariadb-server mariadb

2) Configure for ranger and rangerkms dba account:

3) Install mysql-jdbc and ambari setup

yum install mysql-connector-java
ambari-server setup --jdbc-db=mysql --jdbc-driver=/usr/share/java/mysql-connector-java.jar

4) Configure in ambari, 

*This ranger db and db username rangeradmin will be created by ambari

*Following shows use what user to do the db creation things, we just created user rangerdba, set to here and click Test Connection. And Same as ranger kms:

Start to install, waiting:

If install properly, all the services will become green:

If not, services will show RED, try click … and start all service, if failed to start again, check the logs under ambari server: /var/log/{service} if any issue.

[Step 4. Env ready, and install GPFS MPack to change community HDFS to HDFS tranparency]

Now the Ambari+HDP env are ready, but HDFS is still community version (using local disk rather than GPFS fs), we need to install Spectrum Scale MPack to connect to GPFS transparency.

0) Before you install GPFS MPack, you need to install a normal GPFS cluster(5.1) on this ambari cluster, and create a new filesystem named hadoop for example.

[root@ambari1 ~]# mmlscluster

GPFS cluster information


  GPFS cluster name:         hadoop.ambari1

  GPFS cluster id:           16688944068525815190

  GPFS UID domain:           hadoop.ambari1

  Remote shell command:      /usr/bin/ssh

  Remote file copy command:  /usr/bin/scp

  Repository type:           CCR

 Node  Daemon node name  IP address  Admin node name  Designation


   1   ambari1   ambari1          quorum

   2   ambari2   ambari2          

   3   ambari3   ambari3          

[root@ambari1 ~]# mmlsdisk hadoop

disk         driver   sector     failure holds    holds                            storage

name         type       size       group metadata data  status        availability pool

------------ -------- ------ ----------- -------- ----- ------------- ------------ ------------

nsd_a        nsd         512        1001 Yes      Yes   ready         up           system       

nsd_b        nsd         512        1001 Yes      Yes   ready         up           system       

nsd_c        nsd         512        1001 Yes      Yes   ready         up           system       

In our BDA doc, ESS is required, and remote mount is suggested, but this is a test environment so we can just use local nsd based filesystem instead.

a) Download mpack from Fix Center, we can find to download:

b) Extract and install mpack on ambari-server:

c) After you install properly, you can add new service which named: ”Spectrum Scale”

Choose the master node, master node is the node that take care of running operations like collect snap, upgrade transparency check status, etc, normally assign the same node as ambari server.

Set the ambari server credential and gpfs repo url:

*Tips: Here the GPFS repo url should include hdfs_rpm(you can run createrepo in that folder and export to yum by http), ambari mpack would fetch hdfs-protocol rpm from there and install to gpfs nodes.


All done, and now you can verify that the native datanode/namenode services are replaced by transparency services. 

[root@ambari1 ~]# ps -ef| grep -e namenode -e datanode

root      3275     1  0 Jun27 ?        00:06:07 /home/jdk1.8.0_271//bin/java -Dproc_datanode -Dhdp.version= -server -XX:ParallelGCThreads=4 -XX:+UseConcMarkSweepGC -XX:OnOutOfMemoryError="/usr/hdp/current/hadoop-hdfs-datanode/bin/kill-data-node" -XX:ErrorFile=/var/log/hadoop/root/hs_err_pid%p.log -XX:NewSize=200m -XX:MaxNewSize=200m -Xloggc:/var/log/hadoop/root/gc.log-202106272013 -verbose:gc -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+PrintGCDateStamps -Xms1024m -Xmx1024m,DRFAS -Dhdfs.audit.logger=INFO,DRFAAUDIT -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly -Djava.library.path=:/usr/hdp/ -Dhadoop.log.dir=/var/log/hadoop/root -Dhadoop.log.file=hadoop-root-datanode-ambari1.log -Dhadoop.home.dir=/usr/lpp/mmfs/hadoop -Dhadoop.root.logger=INFO,RFA -Dhadoop.policy.file=hadoop-policy.xml org.apache.hadoop.hdfs.server.datanode.DataNode

root      6353     1  1 Jun27 ?        00:20:52 /home/jdk1.8.0_271//bin/java -Dproc_namenode -Dhdp.version= -Dhdfs.audit.logger=INFO,NullAppender -server -XX:ParallelGCThreads=8 -XX:+UseConcMarkSweepGC -XX:ErrorFile=/var/log/hadoop/root/hs_err_pid%p.log -XX:NewSize=128m -XX:MaxNewSize=128m -Xloggc:/var/log/hadoop/root/gc.log-202106272014 -verbose:gc -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+PrintGCDateStamps -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly -Xms1024m -Xmx1024m,DRFAS -Dhdfs.audit.logger=INFO,DRFAAUDIT -XX:OnOutOfMemoryError="/usr/hdp/current/hadoop-hdfs-namenode/bin/kill-name-node" -Dorg.mortbay.jetty.Request.maxFormContentSize=-1 -Djava.library.path=:/usr/hdp/ -Dhadoop.log.dir=/var/log/hadoop/root -Dhadoop.log.file=hadoop-root-namenode-ambari1.log -Dhadoop.home.dir=/usr/lpp/mmfs/hadoop -Dhadoop.root.logger=INFO,RFA -Dhadoop.policy.file=hadoop-policy.xml org.apache.hadoop.hdfs.server.namenode.NameNode

root     32233 25238  0 23:03 pts/1    00:00:00 grep --color=auto -e namenode -e datanode

[root@ambari1 ~]# df -h

Filesystem             Size  Used Avail Use% Mounted on


hadoop                 300G  6.4G  294G   3% /hadoop

[Other tips]

  1. Set java_home to /etc/profile then all the services user could reach java home.

export JAVA_HOME="/root/jdk1.8.0_271/"

export PATH=$PATH:/usr/lpp/mmfs/bin/

export PATH=$JAVA_HOME/bin:$PATH

  1. Hive failed to start due to port 10002 not released

2021-06-25T04:14:50,702 INFO  [main]: server.HiveServer2 ( - Starting Web UI on port 10002

021-06-25T00:03:27,767 INFO  [main]: server.HiveServer2 ( - Created a znode on ZooKeeper for HiveServer2 uri: ambari3:10000

2021-06-25T00:03:27,767 INFO  [main]: metadata.HiveUtils (:()) - Adding metastore authorization provider:

2021-06-25T00:03:27,768 INFO  [main]: imps.CuratorFrameworkImpl (:()) - Starting

2021-06-25T00:03:27,768 INFO  [main]: zookeeper.ZooKeeper (:()) - Initiating client connection, connectString=ambari1:2181,ambari3:2181,ambari2:2181 sessionTimeout=1200000 watcher=org.apache.curator.ConnectionState@17134190

2021-06-25T00:03:27,771 INFO  [main-SendThread(ambari2:2181)]: zookeeper.ClientCnxn (:()) - Opening socket connection to server ambari2/ Will not attempt to authenticate using SASL (unknown error)

2021-06-25T00:03:27,772 INFO  [main-SendThread(ambari2:2181)]: zookeeper.ClientCnxn (:()) - Socket connection established, initiating session, client: /, server: ambari2/

2021-06-25T00:03:27,775 INFO  [main-SendThread(ambari2:2181)]: zookeeper.ClientCnxn (:()) - Session establishment complete on server ambari2/, sessionid = 0x27a414d81d60065, negotiated timeout = 60000

2021-06-25T00:03:27,775 INFO  [main-EventThread]: state.ConnectionStateManager (:()) - State change: CONNECTED

2021-06-25T00:03:27,781 INFO  [main]: server.HiveServer2 ( - Find 1 policy to synchronize, start PrivilegeSynchronizer

2021-06-25T00:03:27,935 INFO  [HiveMaterializedViewsRegistry-0]: metadata.HiveMaterializedViewsRegistry (:()) - Materialized views registry has been initialized

2021-06-25T00:03:27,937 INFO  [main]: server.Server (:()) - jetty-9.3.25.v20180904, build timestamp: 2018-09-04T17:11:46-04:00, git hash: 3ce520221d0240229c862b122d2b06c12a625732

2021-06-25T00:03:27,937 INFO  [PrivilegeSynchronizer]: authorization.PrivilegeSynchronizer (:()) - Start synchronize privilege

2021-06-25T00:03:27,982 INFO  [main]: http.CrossOriginFilter (:()) - Allowed Methods: GET,POST,DELETE,HEAD

2021-06-25T00:03:27,982 INFO  [main]: http.CrossOriginFilter (:()) - Allowed Headers: X-Requested-With,Content-Type,Accept,Origin,X-Requested-By,x-requested-by

2021-06-25T00:03:27,982 INFO  [main]: http.CrossOriginFilter (:()) - Allowed Origins: *

2021-06-25T00:03:27,982 INFO  [main]: http.CrossOriginFilter (:()) - Allow All Origins: true

2021-06-25T00:03:27,982 INFO  [main]: http.CrossOriginFilter (:()) - Max Age: 1800

2021-06-25T00:03:27,983 INFO  [main]: handler.ContextHandler (:()) - Started o.e.j.w.WebAppContext@4a660b34{/,file:///tmp/jetty-,AVAILABLE}{jar:file:/usr/hdp/!/hive-webapps/hiveserver2}

2021-06-25T00:03:27,983 INFO  [main]: handler.ContextHandler (:()) - Started o.e.j.s.ServletContextHandler@362a561e{/static,jar:file:/usr/hdp/!/hive-webapps/static,AVAILABLE}

2021-06-25T00:03:27,983 INFO  [main]: handler.ContextHandler (:()) - Started o.e.j.s.ServletContextHandler@2df3545d{/logs,file:///var/log/hive/,AVAILABLE}

2021-06-25T00:03:27,984 ERROR [main]: server.HiveServer2 ( - Error starting Web UI: Address already in use

        at Method) ~[?:1.8.0_271]

        at ~[?:1.8.0_271]

        at ~[?:1.8.0_271]

        at ~[?:1.8.0_271]

        at ~[?:1.8.0_271]

        at org.eclipse.jetty.server.ServerConnector.openAcceptChannel( ~[jetty-runner-9.3.25.v20180904.jar:9.3.25.v20180904]

        at ~[jetty-runner-9.3.25.v20180904.jar:9.3.25.v20180904]

        at org.eclipse.jetty.server.AbstractNetworkConnector.doStart( ~[jetty-runner-9.3.25.v20180904.jar:9.3.25.v20180904]

        at org.eclipse.jetty.server.ServerConnector.doStart( ~[jetty-runner-9.3.25.v20180904.jar:9.3.25.v20180904]

        at org.eclipse.jetty.util.component.AbstractLifeCycle.start( ~[jetty-runner-9.3.25.v20180904.jar:9.3.25.v20180904]

        at org.eclipse.jetty.server.Server.doStart( ~[jetty-runner-9.3.25.v20180904.jar:9.3.25.v20180904]

        at org.eclipse.jetty.util.component.AbstractLifeCycle.start( ~[jetty-runner-9.3.25.v20180904.jar:9.3.25.v20180904]

        at org.apache.hive.http.HttpServer.start( ~[hive-common-]

        at org.apache.hive.service.server.HiveServer2.start( [hive-service-]

        at org.apache.hive.service.server.HiveServer2.startHiveServer2( [hive-service-]

        at org.apache.hive.service.server.HiveServer2.access$1700( [hive-service-]

        at org.apache.hive.service.server.HiveServer2$StartOptionExecutor.execute( [hive-service-]

        at org.apache.hive.service.server.HiveServer2.main( [hive-service-]

        at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) ~[?:1.8.0_271]

        at sun.reflect.NativeMethodAccessorImpl.invoke( ~[?:1.8.0_271]

        at sun.reflect.DelegatingMethodAccessorImpl.invoke( ~[?:1.8.0_271]