RE: applying cpu patch to data guard

Yes. There will be some down time during the switch over. I had actually overlooked this important detail. We can handle this by alerting the users to the fact that they may experience a brief outage during the patching.

What are the implications of having the binaries patched without the database being patched?

Thanks.

From: jason arneil [mailto:jason.arneil_at_gmail.com] Sent: Friday, June 10, 2011 3:03 PM
To: Armstrong-Champ, Joseph William
Cc: ORACLE-L
Subject: Re: applying cpu patch to data guard

Hello,

Surely you get *some* downtime to this application (assuming it's read-write) when you perform the switchover?

In terms of the state of the binaries and catbundle, your method is equivalent to a rac rolling install - you end up running one node with the patched software but without catbundle being applied immediately.

jason.

--
http://jarneil.wordpress.com

On 10 June 2011 19:19, Armstrong-Champ, Joseph William <Joseph.Armstrong-Champ_at_tufts.edu<mailto:Joseph.Armstrong-Champ_at_tufts.edu>> wrote:
We just applied the latest cpu patch to a test data guard environment without downtime to the application by (some steps are left out):



-          Cancelling recovery




-          Patching the standby binaries




-          Switch over to the standby (note: the catbundle has not been applied)




-          Patch the primary (original) binaries




-          Switchover (original primary is now primary again)




-          Run the catbundle script against the primary


Everything worked out ok. We were told that this is ok to do but it isn't the way it is outlined in the cpu doc. Does anyone else do it this way? Any comments?

Joe


--
http://www.freelists.org/webpage/oracle-l