My Oracle Support Banner

Upgrading Oracle Big Data Appliance Cluster to V5.2.0 with Mammoth (Software Deployment Bundle) Release V5.2.0 Frequently Asked Questions (FAQ) (Doc ID 2762231.1)

Last updated on APRIL 13, 2021

Applies to:

Big Data Appliance Integrated Software - Version 5.2.0 to 5.2.0 [Release 5.0]
Linux x86-64

Purpose

This document provides answers to frequently asked questions on how to upgrade and install an Oracle Big Data Appliance cluster to Oracle Big Data Appliance 5.2.0 using Mammoth (Software Deployment Bundle). 

 

Questions and Answers

To view full details, sign in with your My Oracle Support account.

Don't have a My Oracle Support account? Click to get started!


In this Document
Purpose
Questions and Answers
 Is BDA 5.2 a major BDA release?
 Questions on 5.2.0 and Oracle Linux 7
 On BDA V5.2.0 is it necessary to migrate Oracle Linux 6 or Oracle Linux 7?
 On BDA V5.2.0 is there a migration utility yet to migrate OL6 to OL7?
 Is it possible to install BDA V5.2.0 on a Oracle Linux 6 base image?
 On new 5.2.0  installs done on the OL7 4.10.0-<latest> base image is it expected that Step 3 to update the base image will require a reboot?
 After the reboot step of an upgrade to BDA V5.2.0 if there are mixed kernel versions due to one or more nodes not having their kernel upgraded is that going to be a problem?
 When performing an upgrade to Oracle Big Data Appliance (BDA) V5.2.0 is it ok to be in Maintenance Mode to avoid Cloudera Manager upgrade-related alerts?
 Do Mammoth installs and upgrades require internet access?
 If a cluster needs to be upgraded and expanded what activity is recommended to do first cluster upgrade or cluster expansion?
 If the BDA has been changed from its original layout will that be a problem during upgrade?
 Does BDA 5.2.0 provide support for rolling upgrades?
 Does the cluster need to be in "Good Health" before upgrading to Oracle Big Data Appliance (BDA) V5.2.0?
 Can you upgrade with a decomissioned node?
 What happens if ssh login of the 'oracle' user to BDA nodes has been disabled for security purposes?
 Will CM Host Inspector "warnings" impact upgrade?
 What Hadoop (CDH) version is on Oracle Big Data Appliance (BDA) V5.2.0?
 What does Cloudera 6.3.4 Enterprise include?
 Does BDA 5.2 support a Mammoth installed Kakfa cluster?
 Does BDA V5.2.0 continue to support security options for CDH Hadoop clusters?
 Are any new parcels included with BDA V5.2.0?
 Are there any new recommendations regarding using Microsoft Active Directory to configure a cluster?
 Does the BDA upgrade to 5.2.0 impact existing Sentry and Kerberos configurations?
 On BDA V5.2.0 can the R release and the Oracle R Advanced Analytics for Hadoop release be independently upgraded?
 On BDA V5.2.0 the Oracle Distribution for R be independently upgraded?
 On BDA V5.2.0 can Solr be upgraded to a higher version?
 On any CDH version is it possible to upgrade to a higher version of Solr?
 Regarding Big Data SQL will the Exadata prerequisites change in BDA V5.2.0  from previous releases?
 What version of NoSQL is on Oracle Big Data Appliance (BDA) V5.2.0?
 Are there two separate bundles for BDA 5.2.0 one for OL 6 and one for OL 7?
 How many zipfiles does the Mammoth deployment have in BDA V5.2.0?
 Is a BDA Base Image being released for V5.2.0?
 Can the second Mammoth bundle deployment zip file be used for reimaging?
 What new features does Oracle Big Data Appliance 5.2.0 include? 
 What about Scripts for Download & Configuration of Apache Zeppelin, Jupyter Notebook, and RStudio?
 What about improved Configuration of Oracle's R Distribution and ORAAH?
 What about support for Kerberos on a Kafka cluster?
 What about a custom Rprofile for BDA?
 Since a custom Rprofile is now set up on the BDA how do users know to copy the file from $R_HOME/etc/RBDAProfiles to $R_HOME/etc?
 What else is new in BDA V5.2.0?
 What are the details on the puppet replacement technology Ansible which is used in 5.2?
 Are there any other Mammoth deployment changes?
 Based on the above new HA features what about MySQL HA?
 Is Big Data Manager supported on clusters secured by Active Directory?
 After an upgrade are the new kernel files copied into the Internal USB drive(/dev/sdm) automatically?
 Is Java upgraded in BDA V5.2.0?
 Regarding HTTPS/Network Encryption, how is that enabled after upgrade?
 What about upgrading with HDFS transparent encryption, KMS Proxy Servers, and Key Trustee Servers?
 Do CM, Hue and Oozie use HTTPS security by default?
 If the Oozie CM parameter "Enable TLS/SSL for Oozie" is off but the other configurations which are part of cluster_https_hue_oozie are enabled and bdacli cluster_https_cm_hue_oozie returns true will that cause any problems during Mammoth upgrade?
 Is there is an interactive bdscli utility on the BDA?
 Is Oracle Big Data Discovery supported on BDA V5.2.0 at this time?
 Are there any changes for installing/uninstalling the Enterprise Manager plugin on BDA 5.2.0?
 Can an upgrade to BDA V5.2.0 be performed from any previous BDA version?
 Does this mean that an upgrade from 4.3 for example, can not go directly to BDA V5.2.0?
 Does this mean that an upgrade from a V4.3, for example, will have a multiple upgrade path for example?
 If multiple upgrades are required to get to BDA 5.2.0 does the critical metadata need to be backed up each time?
 Will there be any problem upgrading BDA V5.2.0 from a supported BDA version if CDH has been upgraded from the default to one of the certified CDH versions?
 For example can the version of Oracle Instant Client be updated?
 During or after upgrade to BDA V5.2.0 can you rollback to a previous BDA version?
 The BDA servers are at CDH 6.3.4 Is it a problem if client/ servers are running CDH 6.3.<x>?
 If a non BDA edge is removed from a cluster prior to BDA upgrade, how is its CDH version upgraded to that on the upgraded BDA cluster?
 If the BDA is running CDH 6.3.4 with Oracle Linux 6.10 for example can a non BDA edge node be added into the cluster if it is running RHEL 6.10?
 If an edge node is added into the cluster, does the edge node need to have the same kernel as the rest of the cluster?
 On a BDA cluster based on Oracle Linux 6 what Linux variants are supported on edge nodes?
 On a BDA cluster based on Oracle Linux 7 what Linux variants are supported on edge nodes?
 If a BDA cluster is running Oracle Linux 6 are edge nodes running Oracle Linux 7 ok?
 Are all nodes of the cluster, however,  required to have the exact same parcel version?
 What version of the Oracle BDA Configuration Generation Utility is required for this release?
 Is Spark-on-YARN configured on the BDA V5.2.0 ?
 What about Spark 1.6 is that supported?
 What about Impala and Search are they configured on the BDA V5.2.0?
 If the Mammoth upgrade fails is it necessary to create a SR and upload the diagnostic bundle generated?
 Where is the BDA V5.2.0 documentation?
 What is the direct link to the documentation "New features" section?
 Is Sqoop2 supported in BDA 5.2.0/CDH 6.3.4?
 Where is the Big Data Connector's documentation?
 If a patch is installed, will that be impacted by upgrade?
 When upgrading to BDA V5.2.0 do any one-off patches need to be applied?
 BDA V5.2.0 includes CDH 6.3.4.  Is it possible to upgrade to a higher CDH version?
 When upgrading to BDA V5.2.0 if the cluster verification step fails can Mammoth be rerun?
 Does this mean that if the last step of the upgrade fails but re-running "mammoth -c" is successful then the upgrade is considered complete?
 Is it possible to run ./mammoth -c in the middle of an upgrade to see how things look?
 Is it possible to run ./mammoth -c after executing the run command i.e. BDAMammoth-CDH-ol6-<version>.run just to give a final check on cluster state?
 How can the last step run by Mammoth be found?
 If Mammoth upgrade or install fails, how is it possible to find the step to resume from?
 If there is a problem with a Mammoth step on upgrade or install is it possible to skip it?
 Can the client hostnames and/or priviate InfiniBand IP addresses be changed during the upgrade?
 After upgrade the user 'bdatestuser' was not available.  Why is that?
 Does Mammoth create an oracle user for general use?
 If a cluster is running AD Kerberos why is an AD Kerberos Admin password required?
 If a cluster is running AD Kerberos and the cluster verifications are run standalone is an oracle user required?
 After upgrade there is a problem using the 'hdfs' principal when AD Kerberos is enabled.   Why is that?
 What is the recommended way to integrate Kerberos with Active Directory (AD) on the BDA?
 Does the httpd service need to run on Node 1 during upgrade?
 Is Tomcat used in CDH 6?
 In the case of multiple clusters should the VLANs be the same in all clusters or could be different in every cluster?
 Now that 3 node clusters are supported is it possible to have a 3 node production cluster?
 Where is the Cloudera documentation on CDH 6.3?
 Where is the Cloudera documentation on CDH 6.3.4 Known Issues?
 Is a BDA upgrade possible if one of the servers is down?
 If one of the ILOMs is down is it possible that a BDA upgrade can proceed?
 Is it possible to upgrade the ILOM to the latest version outside of Mammoth?
 Will Mammoth 5.2.0 upgrade update the ILOM firmware?
 Questions/Answers on handling predictive disk failures before/during upgrade.
 If any disks that exhibit "other" errors fail or get into a failing (predictive error) state, right before upgrade or during upgrade what problems will be faced?
 What about a disk that is in bad health right before upgrade?
 What about a disk that fails during upgrade?
 Should we remove any failing disks from CM from the DataNode Data Directory dfs.data.dir, i.e. remove the mount /u0x/hadoop/dfs from there?
 What is a summary of the failed/failing disk issue and upgrade?
 Can you expand a cluster if one of the new nodes has a bad disk?
 If MySQL replication is not working on the MySQL slave, Node 2 by default, is it ok to proceed with the upgrade?
 If a MySQL slave is in the process of a mysql import can upgrade proceed?
  If a cluster expansion is underway and MySQL replication is observed to be off  can it be enabled during the expansion?
 In BDA V5.2.0 what are the restrictions on extending a cluster?
 Will changing the root password after install is complete affect the system?
 Can Mammoth installation work when clusters span several racks?
 Is it possible to have different root passwords on different nodes prior to upgrade?
 The ASR password is requested on upgrade. What happens if it is unavailable?
 What is the correct firmware for the PDU, Cisco switches, Infiniband switches(spine/leaf), for Mammoth 5.2.0?
References

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.