Comments on: Undocumented parameter _fix_control Or How to break your database http://www.pythian.com/blogs/1002/undocumented-parameter-_fix_control-or-how-to-break-your-database News and views from Pythian DBAs Thu, 16 Oct 2008 07:33:33 +0000 http://wordpress.org/?v=2.3.2 By: Abel - Oracle Support Services http://www.pythian.com/blogs/1002/undocumented-parameter-_fix_control-or-how-to-break-your-database#comment-271176 Abel - Oracle Support Services Wed, 03 Sep 2008 15:52:53 +0000 http://www.pythian.com/blogs/1002/undocumented-parameter-_fix_control-or-how-to-break-your-database#comment-271176 Thank you for sharing. The article indicates 2 problem. If a bug fix number is invalid the database will not startup signaling an ORA-940 This was reported on bug 6756883. Internally there is a registry service for the fixes.The issue is that if the bug was not registered and attempted to disable it then gives the error. It is possible for an Oracle developer to forget to register the fix or not do registered on purpose so disabling it would not be possible and give ora-940. If it is not registered by mistake it should be reported to oracle to fix it and once it is registered it should work for future versions too. Some examples of bugs left out on purpose are wrong results bugs. The second issue was that if no number was set the instance will become unusable. This was reported on bug 6628797 and a fix is available. Next time, I suggest to contact Oracle Support if something does not appear to function correctly so we can take care of it and then postings like this will provide a solution besides "don't use it". Thank you for sharing.
The article indicates 2 problem.

If a bug fix number is invalid the database will not startup signaling an ORA-940
This was reported on bug 6756883.
Internally there is a registry service for the fixes.The issue is that if the bug was not registered and attempted to disable it then gives the error.
It is possible for an Oracle developer to forget to register the fix or not do registered on purpose so disabling it would not be possible and give ora-940.
If it is not registered by mistake it should be reported to oracle to fix it and once it is registered it should work for future versions too.
Some examples of bugs left out on purpose are wrong results bugs.

The second issue was that if no number was set the instance will become unusable.
This was reported on bug 6628797 and a fix is available.

Next time, I suggest to contact Oracle Support if something does not appear to function correctly so we can take care of it and then postings like this will provide a solution besides “don’t use it”.

]]> By: Alex Gorbachev http://www.pythian.com/blogs/1002/undocumented-parameter-_fix_control-or-how-to-break-your-database#comment-209183 Alex Gorbachev Sat, 31 May 2008 07:41:21 +0000 http://www.pythian.com/blogs/1002/undocumented-parameter-_fix_control-or-how-to-break-your-database#comment-209183 Live and learn - I never heard about this parameter and those tables. Good to know. Thanks Luke! Live and learn - I never heard about this parameter and those tables. Good to know. Thanks Luke!

]]> By: Bradd PIontek http://www.pythian.com/blogs/1002/undocumented-parameter-_fix_control-or-how-to-break-your-database#comment-197323 Bradd PIontek Mon, 12 May 2008 15:28:46 +0000 http://www.pythian.com/blogs/1002/undocumented-parameter-_fix_control-or-how-to-break-your-database#comment-197323 I was curious as to weather unsetting the parameter via 'alter system unset "_fix_control" would be possible or get around the having to bounce the instance? You have got to love these parameters that have no type checking or validation. I suppose that is why oracle always puts such large caveats around the undocumented parameters. I was curious as to weather unsetting the parameter via ‘alter system unset “_fix_control” would be possible or get around the having to bounce the instance?

You have got to love these parameters that have no type checking or validation. I suppose that is why oracle always puts such large caveats around the undocumented parameters.

]]>