UKOUG Day 4: Oracle Cloud Control 12c for DBAs

Today’s first session was a round table session with Andrew Bullock, Director of Application and systems management at Oracle. The main result of this session was that the main people using Enterprise Manager Cloud Control 12c were DBA and not so much system or Weblogic Administrators.

It appears also that not so many DBA have realized in production the upgrade from Grid Control 11g to Cloud Control 12c, because in case of a huge number of targets, the agent upgrade phase could last a very long time. It appeared also that many DBA use Enterprise Manager Cloud 12c in a high availability mode in active/active or active/passive environment.

It was really interesting to share our experiences about agent’s unreachable, problematic agent 12c installation on MS windows, or AIX.
I continued with a rman session resuming some hints to improve our rman scripts.

1-    It is recommended before calling RMAN to export NLS_DATE_FORMAT in a suitable way to have correct date formats in the log files. It is also important to have the total execution time at the end of the log file.
2-    Do not use crosscheck the crosscheck command in your RMAN backup scripts because RMAN will silently ignore the missing files and the recovery might be impossible. The crosscheck command should be a manual activity executed by a DBA to solve an issue.
3-    Backup the control file at the last step of your backup script.
4-    Do not delete archive log based on time only. It is better to use a command like:

rman>delete no prompt archive log all backed {N} times to disk completed before ‘sysdate –x’;
Or for a standby database:
rman> configure archivelog deletion policy to applied on standby;

5-    Do not rely on rman stored configuration. For example if you use the control file auto backup feature, what happens if another DBA switches auto backup to off ? A good advice is to capture the rman settings and to restore it at the end of the RMAN backup script.
6-    Send backup log files to a shared email address. If possible run a log files check in your rman backup script and in case of failure send an alert to the on call DBA. A good advice should also to check the long running backups.
7-    Use the report commands such as :

Rman> report need backup days=2;
Rman> report unrecoverable;
The last session I assisted was again about Enterprise Manager 12c, even after been working on this subject since quite a few years I learnt some hints, I share with you:

1-    An EM 12c infrastructure is considered as little when monitored targets are less than 1000, users less than 10 and the number of management agent are less than 100.
2-    When planning an installation you have to question yourself about the future number of targets, how big they are to define if you decide to install OMS in active / active environment for high availability
3-    In a shared management server storage the recommendation is at least 50 Gb (in fact for the software library)
4-    Concerning the Enterprise Manager Repository one recommendation is to increase the default value of the STREAMS_POOL_SIZE parameter.
5-    The default security settings for logging information in the weblogic server are set at a level that will cause excessive logging data. It is recommended to modify the USER_MEM_ARG –Xmx in the startEMserver.sh at a value of 4 G in a medium environment size.
6-    Of course the repository should be managed as a target.
7-    Don’t forget to use the search box in the Cloud 12c console to find target or knowledge.
8-    Use emcli, not so many DBA use this nice feature which allows creating database, clear stateless incidents, deploying patches for example. This feature is not very well documented but so useful!

Those four days in Manchester UKOUG have been intensive but very interesting. The UKOUG event is really a technical event allowing us to share technical information with other DBA. We impatiently wait for the next year :=))