Use OEM 13c R2 to Discover Oracle BDA

OEM 13c Cloud Control is a powerful monitoring tool, not only for Exadata and Oracle database, but also for Oracle Big Data Appliance (BDA). There are many articles or blogs about Exadata Discovery using OEM 12c or 13c. But not many places discuss the OEM BDA Discovery, especially using the new version of OEM, 13c Cloud Control. In this blog, I am going to discuss the steps to discover BDA using OEM 13c R2.

First, do not use OEM 13c R1 for BDA Discovery. It is very time consuming and very likely not going to work. OEM 13c R2 is much better, at least I can successfully do the BDA Discovery on all of the BDAs I have worked on.

Secondly, unlike OEM Exadata Discovery, BDA Discovery usually requires one extra step before the Manual OEM BDA Discovery by using bdacli enable em command first. Theoretically if works, I don’t need to do anything in manual BDA discovery process. Unfortunately I have never run into this perfect situation in different BDA environment and always get certain kind of errors at the end.

Preparation
There are a few useful notes about OEM BDA Discovery.
1) Instructions to Install 12.1.0.4 BDA Plug-in on Oracle Big Data Appliance (BDA) V2.*/V3.0.*/V3.1/V4.* (Doc ID 1682558.1)
2) BDA Credentials for Enterprise Manager 13.x Plugin (Doc ID 2206111.1)
3) Instructions to Enable / Disable the 13.x BDA Enterprise Manager Plug-in on Oracle Big Data Appliance (BDA) V4.5-V4.7 (Doc ID 2206207.1)

Execute bdacli command
Run bdacli enable em. For BDA version below 4.5, run command bdacli enable em –force. I am almost 100% guarantee you won’t see the successful completion message from this command. For example, get the following error at the end.

INFO: Running: /opt/oracle/emcli_home/emcli discover_bda_cluster -hostname=enkx4bda1node01.enkitec.local -cloudera_credential=BDA_ENKX4BDA_CM_CRED -host_credential=BDA_ENKX4BDA_HOSTS_CRED -cisco_credential=BDA_ENKX4BDA_CISCO_CRED -ilom_credential=BDA_ENKX4BDA_ILOM_CRED -infiniband_credential=BDA_ENKX4BDA_IB_CRED -pdu_credential=BDA_ENKX4BDA_PDU_CRED -cisco_snmp_string="snmp_v3;;SNMPV3Creds;authUser:none;authPwd:none;authProtocol:none;privPwd:none" -pdu_snmp_string="snmp_v1v2_v3;;SNMPV1Creds;COMMUNITY:none" -switch_snmp_string="snmp_v1v2_v3;;SNMPV3Creds;authUser:none;authPwd:none;authProtocol:none;privPwd:none"
ERROR: Syntax Error: Unrecognized argument -cisco_snmp_string #Step Syntax Error: Unrecognized argument -pdu_snmp_string#
Are you sure you want to completely cleanup em and lose all related state ?

When see the above message, always type in N and not rollback the changes. Basically you have OEM agent deployed, just need to figure out which node you want to use as the start point for Manual OEM BDA Discovery.

On each node, run the following command:

[root@enkx4bda1node06 ~]# java -classpath /opt/oracle/EMAgent/agent_13.2.0.0.0/jlib/*:/opt/oracle/EMAgent/agent_13.2.0.0.0/plugins/oracle.sysman.bda.discovery.plugin_13.2.2.0.0/archives/* oracle.sysman.bda.discovery.pojo.GetHadoopClusters http://enkx4bda1node03.enkitec.local:7180/api/v1/clusters admin admin_password

You should see the error below for the execution on many nodes.

Apr 10, 2017 10:14:44 AM com.sun.jersey.api.client.ClientResponse getEntity
SEVERE: A message body reader for Java class [Loracle.sysman.bda.discovery.pojo.Items;, and Java type class [Loracle.sysman.bda.discovery.pojo.Items;, and MIME media type text/html was not found
Apr 10, 2017 10:14:44 AM com.sun.jersey.api.client.ClientResponse getEntity
SEVERE: The registered message body readers compatible with the MIME media type are:
*/* ->
  com.sun.jersey.core.impl.provider.entity.FormProvider
  com.sun.jersey.core.impl.provider.entity.MimeMultipartProvider
  com.sun.jersey.core.impl.provider.entity.StringProvider
  com.sun.jersey.core.impl.provider.entity.ByteArrayProvider
  com.sun.jersey.core.impl.provider.entity.FileProvider
  com.sun.jersey.core.impl.provider.entity.InputStreamProvider
  com.sun.jersey.core.impl.provider.entity.DataSourceProvider
  com.sun.jersey.core.impl.provider.entity.XMLJAXBElementProvider$General
  com.sun.jersey.core.impl.provider.entity.ReaderProvider
  com.sun.jersey.core.impl.provider.entity.DocumentProvider
  com.sun.jersey.core.impl.provider.entity.SourceProvider$StreamSourceReader
  com.sun.jersey.core.impl.provider.entity.SourceProvider$SAXSourceReader
  com.sun.jersey.core.impl.provider.entity.SourceProvider$DOMSourceReader
  com.sun.jersey.core.impl.provider.entity.XMLRootElementProvider$General
  com.sun.jersey.core.impl.provider.entity.XMLListElementProvider$General
  com.sun.jersey.core.impl.provider.entity.XMLRootObjectProvider$General
  com.sun.jersey.core.impl.provider.entity.EntityHolderReader

For certain node, you could see successful message and showing below.

enkx4bda;;

In my case, it is node 2. So I will use Node 2 for my manual BDA Discovery in the following steps.

Manual OEM BDA Discovery
Logon to OEM as sysman user. Select Add Target -> Add Target Manually.

Select Add Targets Using Guided Process

Select Oracle Big Data Appliance

The Add Targets Manually pages shows up. Select node2 from the list. Click Next.

After it completes, it will show the following hardware information. Click Next.

The Hardware Credentials screen shows up. If all Host credentials show green sign, you don’t need to do anything related to Host. Go to the next one, for example, IB Switch. Select Set Credentials -> All Infiniband Switches . Then set SNMP Credentials type and community string. Majority of the time, input public for community string. Then click OK.

If successful, it shows the green check.

Following the similar procedure for all other hardware components, like ILOM, PDU and Cisco Switch. At the end, you should see the following screen.
One interesting note about PDU. PDU component always behave in a weird way during the discovery. For this case, it shows successful with green check, but later on OEM shows PDUs as DOWN status after the discovery. In my other discovery works for different BDA environments, the green check has never shown up in this page, but PDUs shows UP status after the discovery. So the result is inconsistent.

Click Next. The screen for Cloudera Manager shows up. Click Edit, verify the credential for admin user for Cloudera Manager. Then click Next.

The Software page shows up, click Next.

The review page shows up, click Submit

If successful, will see the screen message below, click OK.

The BDA Discovery is completed.
.
You might notice the new BDA cluster is called BDA Network1. This is not a good way to name a cluster, especially you have multiple BDAs under the management from the same OEM. I don’t understand why not to use BDA’s cluster name or Cloudera Manager’s cluster name. Either one will be much better than this naming. Even worse, you can change a lot of target name in OEM, but not for this one. I have another blog (Change BDA Cluster Name in OEM Cloud Control 13c) discussing a partial workaround for this issue.

To view the detail of a host target, you can have the following:

The presentation looks better than OEM 12c. In general, OEM 13c for BDA is good one. But pay attention to the followings. Otherwise you will spend a lot of additional time.
1) Before performing OEM BDA Discovery, make sure you have changed all of your default passwords on BDA. It’s easier to use default password during the discovery, but a huge pain after you change passwords for certain user accounts used in BDA discovery. Basically, update the Named Credentials is not enough and you have to delete the whole BDA target in OEM and redo the discovery.

2) Similarly, if configure TLS with Cloudera Manager after BDA Discovery, you will have to remove the BDA target and redo the discovery. It is a clearly a bug in OEM, at least not fixed at the time I am writing this blog.

3) Sometimes you might see tons of alerts from almost every ports in the Cisco switch. If from a few ports, I might believe it. But for almost every port, there is no way this is the right alert. As matter of fact, Oracle Support confirmed it seem false alert. At the time I had to do the BDA Rediscovery after configuring TLS with Cloudera Manager, I happened to notice all Cisco port alerts were gone after BDA rediscovery.

4) Both Oracle document and Oracle support says OEM 13c R2 supports BDA v4.5+ and any version below it is not supported. It’s true the lower BDA version would run into additional issues, but I managed to find workaround and make it working for BDA v4.3.

Advertisements

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s