[pca] Freezing on a date

[Rajiv Gunja]

To freeze patches to a particular date, I freeze the patchdiag.xref to that
date. Currently in my Org, we use patches available as of March 1st 2008, so
I use "pca -y" option to keep the patchdiag.xref to that date. It is the
only fail-safe way to ensure that you do not get any new patches, as mindays
will fail if the patch has been obsoleted.

-GGR


2008/7/31 Martin Paul <martin at par.univie.ac.at>

> Hi,
>
>  My next  question is how do I " freeze" patches on a certain date.
>>
>> pca --minage=DAYS (always making sure to calculate from the day I' m
>> running pca)
>>
>
> Not recommended. Most of the time, it will work, but you will get problems
> if required patches get updated later. Example: On July 24th, patch
> 111111-01 required 222222-01. Two weeks later, 111111-02 is released. While
> 222222-01 still only requires 111111-01, information about that rev. 01 will
> be gone from the xref file, and pca will show 111111-02 in the list of
> missing patches, although it is newer than minage. There's not enough input
> data in the xref file for pca to ensure that minage alwats gives the
> expected result.
>
>  pca --stop=ID (using this one to figure out what the last patch was on the
>> date)
>>
>
> Wouldn't have thought about that one, but I don't recommend it either.
> Patch requirements can change over time, and this will change the patch
> order as well.
>
>  pca --nocheckxref (and making sure I have the xref file from the date I
>> want to freeze on?)
>>
>
> That's the best solution. You have the complete state of all patches and
> patch dependencies as of the date on which you froze the xref file. As pca
> doesn't user other input (from Sun) for its analysis, it's almost perfect. I
> say almost, as in rare cases Sun changes patch attributes for a patch even
> if no new revision is issued. 111111-01 might be the current 111111 patch
> today and two weeks ago, but it might be marked as "R"ecommended in today's
> xref while it wasn't two weeks ago. This is rare, and not a real problem. If
> you install "missing" (and not "missingr") it's no problem at all.
>
> Hope that helps,
>
> Martin.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.univie.ac.at/mailman/private/pca/attachments/20080731/abec971f/attachment-0001.html