Hi Michael
Actually, the patch utilites patch should be rev'ed to the latest rev available
prior to any other patches being applied.
In the case of zones this is extremely important.
Applying patches to an FCS system using the FCS patch utilties will almost
certainly result in issues with respect to zones
either:
1 Issues creating zones subsequent to patching
2 patchadd failing if non-global zones already exist.
So although most if not all Solaris 10 patches benefit from having the patch
utilties up to date, there is no actual hard dependency listed as it covers
every patch, instead we issues a SunAlert.
Please see SunAlert
http://sunsolve.sun.com/search/document.do?assetkey=1-26-101688-1
Now another question is the use of patchinfos to generate information. Patchinfos
are only used internally within Sun to construct metadata for Sun Update
Client ( SunUC ) to consume. This data is actually maintained seperately
by the team that generate the metadata download for SunUC, so basically the
patchinfo in the patch may differ substantially from that used by the Sun
tools themselves. In saying thatthough, fields such as PATH_REQUIRES should
be more stable than most.
patchadd itself does not use or rely on 'any' data contained in the patchinfo
file. The only tools that use patchinfo metadata are updatemanger from (
SunUC ) which uses smpatch which eventually calls patch add to add the patch
once it has parsed the metadata.
I still have not determined the problem with the cluster in question but
am continuing to look. Bacically all the clusters should install the patch
utilites first ( and their requirements ). Looks like the one in question
is doing that, but I'd like to see the actual cluster itslef.
Enda
Michael Barto wrote:
Our company's configuration management products contains a Perl script which
calculates patch order based using a recursive program. I cannot share the
program because it is part of our products, but I will give you the information
on how it works. In each Sun patch directory after it is decompressed, there
is a file called "patchinfo". This is somewhat similar to the "pkginfo" file
for programs (e.g. pkgadd). In it you will find a keyword "PATCH_REQUIRES"
followed by a list of patches and sometimes some "if-then-else" constructions.
The "PATCH_REQUIRES" information must be installed prior to installing the
or "patchadd" will fail. If you are having issues with a failed patch. Check
this file, it may resolve your problem.
In your issue with 119254-19 and 118712-09 there is only a reference for
in the "patchinfo" file that patch "121333-02 must be installed prior to
installing 119254-19. Hence, I believe the order is unimportant with those
two patch numbers. The conflict is that "121333-02" must be installed prior
to installing "119254-19". Over a period of time, I have found several times
that the patch README file is in conflict to this "patchadd" parameter file
("patchinfo") when I have done a large server migrations for Sun.
Below is the patch info file for 119254-19
patchinfo file for "119254-19 "
-
PATCHINFOVERSION="1.0"
PATCHID=119254-19
PATCH_CORRECTS='BaseOS.Package.Cmds-5.10 AdminInstall.Solaris.SUNWswmt-5.10'
PATCH_ARCH='sparc'
PATCH_OS='SunOS'
PATCH_OSRELEASE='5.10'
PATCH_PROPERTIES='clientusr'
PATCH_REQUIRES="121333-02"
PATCH_OBSOLETES="119015-03"
-
Michael Barto
Software Architect
LogiQwest
Inc.
16458 Bolsa Chica Street, # 15
Huntington Beach, CA 92649
http://www.logiqwest.com/
[EMAIL PROTECTED]
Tel: 714 377 3705
Fax: 714 840 3937
Cell: 714 883 1949
'tis a gift to be simple
This e-mail may contain LogiQwest
proprietary information and should be treated as confidential.
___
zones-discuss mailing list
zones-discuss@opensolaris.org
___
zones-discuss mailing list
zones-discuss@opensolaris.org