The new Oracle online patching utility that comes with R12.2 looks like it can save a lot of down time by using 11GR2 and forward – Database Editions feature and three new file systems “run” / fs1, “patch” fs2 (can be switched), and fs_ne (fs_ne is for files not subject to patching usually).
There is also a “hotpatch=yes” switch, that applies you patch to all file systems and does not require all the standard adop phases – only “apply” is supported with the hotpatch=yes switch.
But, as all seasoned EBS R12.2 Admins know… hotpatch cannot be rolled back or aborted “adop phase=abort”.
So you had better know for certain – when applying an EBS patch with the hotpatch=yes switch. Backing out of it may be difficult to impossible.
See Oracle Support Note: 2054074.1
It explicitly states “adop phase=abort” cannot be used on a hotpatch=yes session. But does allude to (assuming you have a replacement).
2. Apply a replacement patch with abandon=yes.
I did back out of one (hotpatch=yes) in an unsupported manner in a sandbox environment as follows:
sqlplus apps/pw
select adop_session_id, status from apps.ad_adop_sessions;
Important: select the correct ADOP_SESSION_ID
update apps.ad_adop_sessions set status=’C’ where adop_session_id=2;
commit;
Then update the status to “C” and commit – it allowed me to move forward – in an unsupported way in my sandbox environment – I wouldn’t try it in production…
BTW – first time I applied a very large EBS patch set, and tried an abort, I had to add 30GB UNDO space to the 5GB that already existed before “phase=abort” would succeed…
Another BTW – here’s the best “adop” summary blog I know of at the moment (nice job):
http://h2hdba.blogspot.com/p/r122-quick-notes-and-commands.html