Re: patch with no testing

cameron.cleland_at_ci.sj.ca.us (horse person) wrote in message news:<ee3c77d3.0305011208.da951e_at_posting.google.com>...
> I have an 8.1.7.0 database that I want to patch to 8.1.7.4 There are
> no resources to test the patch. At all. so my question is, what is a
> bigger risk? Running an unpatched database, or patching without
> testing? If you had no resources for testing, would you go ahead and
> install the patch anyway?

okay, I'll bite.

No one in your company has a workstation or a laptop capable of supporting a subset of your app data? This sounds like an excellent opportunity for you to have your company purchase a new laptop (provided that you're on win32 or linux). For under $1500 - you can pickup a 2.4 GHz unit with 512M of ram and a 60 GB HD. Yes, it won't perform the same as your server - but that is not the point. you're not testing performance so much as errors and loss of functionality, e.g. ORA-00600 {729] [space leak] [1048576] [] [] [] [] [].

I do most of my 9.2 hacking on my laptop during my train time.

the main patchsets (e.g. 8.1.7.4.1) are regression tested. these carry the least risk - but behavior of the CBO could change. Unfortunately, this particular patchset addresses none of the issues raised by security alerts 48-54, which may be one of your major motivating factors for applying patches.

If you have no current issues with the .0.0 release, by all means stay on it.
If you open any issues with Oracle Support - their first line will likely be - upgrade to the current supported release (8.1.7.4.1) - but don't apply a one-off patchset unless required.

you absolutlely do not want to apply patchset 8.1.7.4.10 without thorough testing. make sure that you spool the output of the dictionary rebuild to a file - you'll need it. read the readme.txt file completely.

Paul Received on Thu May 01 2003 - 22:07:01 CDT