Upgrade from 10.2.0.2 to 10.2.0.4

Printable View

02-23-2010, 11:37 AM
dimicont

Upgrade from 10.2.0.2 to 10.2.0.4

Hello everybody!
We are using Oracle standard edition 10.2.0.2 on Solaris x86-64 platform, and I'm trying to estimate reasonability of movement to 10.2.0.4.
A disadvantage I see is according to 10.2.0.2 "Availability and Known Issues" amount of "issues introduced" is about 10, but in 10.2.0.4 there are 63 "issues introduced", according to similar document. I suspect that some of bugs can be eliminated by applying 10.2.0.4.3 patchset, but I'm not sure that all of them will be eliminated.
A benefit of 10.2.0.4 (and especially 10.2.0.4.3) is presence of all necessary critical patches (up to Jan2010).
Can you provide me any suggestions in this situation? And do you know does the game is worth the candle?

Thanks in advance, guys!:)
02-23-2010, 02:06 PM
davey23uk

thats why we have test systems, to test these things ouyt to see if they work well for us.

umber of bugs introduced is irrelevant if none of them affect you
02-23-2010, 04:33 PM
dimicont

Quote:

Originally Posted by davey23uk

umber of bugs introduced is irrelevant if none of them affect you

It's an absolute truth! Except a case then they appear AFTER testing was finished and patchset has been applied on production system :)
02-23-2010, 06:50 PM
vnktummala

I have recently upgraded all the databases in one of my clients' shop from 10.2.0.3 to 10.2.0.4 because I need to apply DST v13 patch and DST v13 released for 10.2.0.4.

If you have no reason or bugs you don't need to upgrade just to move to the higher version.

Thanks,
02-24-2010, 04:27 AM
dimicont

Quote:

Originally Posted by vnktummala

I have recently upgraded all the databases in one of my clients' shop from 10.2.0.3 to 10.2.0.4 because I need to apply DST v13 patch and DST v13 released for 10.2.0.4.

If you have no reason or bugs you don't need to upgrade just to move to the higher version.

Thanks,

Thank you for your response, vnktummala! In my everyday life I also use a rule - "There is no necessity to repair thing that wasn't broken" :)
02-24-2010, 05:20 PM
PAVB

Quote:

Originally Posted by dimicont

It's an absolute truth! Except a case then they appear AFTER testing was finished and patchset has been applied on production system :)

I'm really sorry in being the one breaking this news but... that's the definition of bad testing.
02-25-2010, 02:11 AM
hany

I would like to add that even if you did your testing optimally, there is always the case of hitting "Bugs", that's why it's really recommended to have the latest patchset applied so that you can get a fix from Oracle.
Imagine if you are running on 10.2.0.2 and all of a sudden you hit a bug that was not even discovered uptill 10.2.0.4, what would be easier then?
1. To start testing applying 10.2.0.4 which could take some time if you are going to do some proper testing (with all the downtime involved) and then create a SR with Oracle to investigate the bug and create a fix.
OR
2. Create a SR with Oracle to investigate the bug and create a fix since you are already running the latest patchset.

Regards,
Hany.
02-25-2010, 04:26 AM
dimicont

Quote:

Originally Posted by PAVB

I'm really sorry in being the one breaking this news but... that's the definition of bad testing.

yes, but anyone can miss or forget about something during testing - even Oracle Corporation (otherwise why should they produce such huge amount of patches and pathchsets after new release becomes available? ;) )
02-25-2010, 04:34 AM
dimicont

Quote:

Originally Posted by hany

I would like to add that even if you did your testing optimally, there is always the case of hitting "Bugs", that's why it's really recommended to have the latest patchset applied so that you can get a fix from Oracle.
Imagine if you are running on 10.2.0.2 and all of a sudden you hit a bug that was not even discovered uptill 10.2.0.4, what would be easier then?
1. To start testing applying 10.2.0.4 which could take some time if you are going to do some proper testing (with all the downtime involved) and then create a SR with Oracle to investigate the bug and create a fix.
OR
2. Create a SR with Oracle to investigate the bug and create a fix since you are already running the latest patchset.

Regards,
Hany.

Hi Hany!
Thank you very much for an excellent example of why patchset applying is preferable. A really good example indeed.

Regards,
dimicont
02-25-2010, 08:37 AM
PAVB

Quote:

Originally Posted by dimicont

yes, but anyone can miss or forget about something during testing - even Oracle Corporation (otherwise why should they produce such huge amount of patches and pathchsets after new release becomes available? ;) )

Oh, I totally agree.

Let me share two theories about bugs - these two were coined by a couple of old friends of mine.

Theory #1 --Oracle sells bugs, rdbms software is just the packaging.

Theory #2 --Oracle developers work hard to produce as many bugs as possible so to ensure job security.
02-25-2010, 09:06 AM
dimicont

Quote:

Originally Posted by PAVB

Oh, I totally agree.

Let me share two theories about bugs - these two were coined by a couple of old friends of mine.

Theory #1 --Oracle sells bugs, rdbms software is just the packaging.

Theory #2 --Oracle developers work hard to produce as many bugs as possible so to ensure job security.

I suppose both these theories can exist together and complement each other :)