We build patch lists every 3 to 6 months which are used to upgrade  
servers to a known state.  We don't patch thousands of servers with  
every list.  But when a server has patches applied it is using one of  
the known internally compiled lists.  As you say, those lists often  
add 100+ patches.

We must use a script very similar to Bernd's that applies each patch  
individually and captures the return code.  At the end we tell the  
sysadmin if there was any error he should go back and check.  And  
yes, we use the internal non documented return codes of the old  
patchadd to check if the error was a simple "package doesn't exist"  
or "patch already applied" ... Obviously we have changed our script  
to use -t on Solaris 10 as well.

I wouldn't mind changing over to 'patchadd -M <patch-dir> <our-list>'  
at all and let you do all the hard work!  But we do like a summary at  
the end that lets us know about the patch errors that we need to go  
and look for (could also be solved by parsing friendly output).  The  
exit code should reflect the difference between a "notice" (e.g.  
didn't add patch as the package it patches isn't installed - nice to  
report but that's about it), "warning" (e.g. couldn't add patch  
because a required patch is missing) and "errors" (tried to add patch  
but something went horribly wrong - we need the sysadmin to go and  
check).  It should report the worst.

I know our sysadmins would hardly ever check through the output of  
patchadd when adding 100+ patches to check if there was any error.   
As you say "And it produce a lot of output - someone even complained  
about it."  Too much output currently leaves you blind to what really  
happened when trying to add all the patches.

- mo


btw, thanks for all your info you post to the list, glad I'm  
subscribed and see it all.  Wouldn't it be better committed to a web  
page as permanent and maintainable documentation rather than hoping  
people might find it again trawling mailing list archives?  Do you  
think "One of the most unnoticed but significant ... improvement"s  
will be noticed by a wider audience this way?  Well, I'm happy, I get  
the info, I did notice it this way :)


On 24 Oct 2006, at 19:23, Vasiliy wrote:

> What exit code should be if some patches installed and some did  
> not? It is not a complete failure on my opinion.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2130 bytes
Desc: not available
URL: 
<http://mail.opensolaris.org/pipermail/install-discuss/attachments/20061031/6ebb7d1d/attachment.bin>

Reply via email to