This is not an issue with -optimistic/-pessimistic patching, that I was trying 
to introduce for patchadd.ksh failure on one of the zones during patching 
transaction. Which do not suppose to fail because pdo did all the checks, but 
sometimes may happen because of prepatch scripts etc.

As you may see pdo in this case effectively found dependency problems on 
several zones and reported this in the log file and what the problem is:

psc-201n: For patch 120272-13, required patch 118833-36 does not exist.
eule-201n: For patch 120272-13, required patch 118833-36 does not exist.
kops-201n: For patch 120272-13, required patch 118833-36 does not exist.
kops-227n: For patch 120272-13, required patch 118833-36 does not exist.
lsorui-218n: For patch 120272-13, required patch 118833-36 does not exist.

And it did not install it on that zones - I see that log is cut off, because 
after patching global zone it suppose to go and patch local zones and it should 
skip all that zones.

This is what happen if patch does not affect anything in shared area and 
installing it in Global zone outside shared area does not break local zones. If 
patch touch any file in shared area it when you install it in global zone 
changes will be shared with local zones right away. This kind of patches called 
Global Patch.

To handle Global patch pdo doing all those dependency check first for all zones 
before going to real patching to verifi that all zones are compatible with 
Global Patch otherwise pdo will not patch anything at all.

This is what suppose to happen in this case because KU is for sure Global Patch 
because it patching kernel which is shared by all zones. For some reason pdo 
seeing all this unsuticfied dependency proceeds with patching, I may suspect 
that Patch did not marked as an Global Patch or it may be something wrong with 
pdo.

I can only  speculate - whoever responsible for patch tools now should have a 
look at this.

Thanks, Vassili.
 
 
This message posted from opensolaris.org

Reply via email to