R12 - Distributed AD Patching

Hello everyone!!

Today I was working on patching of Big RUP patch on Shared appltier - R12 environment. I was running out of time to complete the task. Thought for a while to reduce the patching downtime, took Distributed AD method to reduce the time.

My environment configurations:

3 node mid-tier shared appltier each of 8CPU's and 8Gigs of RAM. (quite healthy).

I decided to have 16 (2* CPU) workers on each node, by summing I will be using 48 workers (whoop!!).


It took one hour less than then normal single node patching. here is the detail how I have applied.

I have exported TWO_TASK (normally two_task is set to Name_BALANCE in RAC with shared appltier) to instance1 of RAC node as AD utilities doesn't use RAC features. (adpatch works perfectly but when patch contains jar files to be uploaded using loadjava utilities it fails making connections to DB using loadbalancing JDBC parameters). This means that I am going to use only node 1 of the DB for all patching. same way I have exported TWO_TASK to instance1 at session level for node2 and node3.

first node I started the patching

adpatch options=hotpatch localworkers=16 workers=48

(response to the prompts, it will start the patching). from the session of node2 run
adctrl distributed=y, respond to the prompt - worker range: 17-32
start the adctrl distributed=y, respond to the prompt - worker range in node3: 33-48

till FND_INSTALL_PROCESS tables get created, distributed adctrl sessions will wait for node1 adpatch. when fnd_install_processes table gets created, distributed workers will be forked and will start the tasks.

its really cool way of reducing downtime - Part No. B31567-02 of R12 document set contains bit more information on this!!

11i note may help you too -note:236469.1

How to check the status of the workers??

You can use the same adctrl command without distributed=Y parameters!!. its going to behave as normal adctrl.

Leave your comments, ideas.!!

Happy learning!!

4 comments:

Anonymous said...

Hi Suresh,
Few doubts please:

why do we need downtime when using hotpatch option.
AND
why are we using hotpatch option for such a big RUP patch.

Suresh Lakshmanan said...

Hi Anonymous,

Good questions. I didn't put the enviornment in maintenance mode. In order to by pass the maintenance mode, I ran the patch in 'hotpatch' option. but my midtiers were down!!
:) hope this clarifies. please mention your name while giving comments to address your with name.
thank you.

Thanks
Suresh

Elan said...

Hi Suresh,

I heard the patch driver file/odf file can be updated suitably to reduce the patch time. It was not very clear to me but the DBA said he reduced the patch time drastically from 120 hours to around 20 hours. Any details on this?

Thanks
Elan

Suresh Lakshmanan said...

Hi Elan,

Thanks for the mail. Look at the page 26 of the powerpoint presentation given by my x director Eugne Weinstein at

http://www.norcaloaug.com/seminar_archive/2008_training_day_pres/4_01_weinstein.ppt

Hope this helps you.

Thanks
Suresh