[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Building db2 v9.5 rpm from payload files

rpm-list-bounces redhat com wrote on 05/14/2008 03:03:50 PM:

> Hallo.  I've been basing my head on this for over a week now (up until
> last 2 days I kept finding possible fixes) and IBM hasnt been all that
> helpful for typical support reasons (why do they care if I can make an
> they provide 2 installers and a manual process! isnt that enough? /end
> sarcasm).  Anyways..  let me provide a bit more information.
> I was going to try building the rpm in the same manner that we do some
> other 3rd party software distributions, which is basically to run the
> installer scripts in the %install portion of the script, gather up the
> files and move on.  This didn't get me where I needed to be, and about
> time IBM released a new draft of their "DB2 Up and Running on Linux"
> provided very basic information about a new installation method as of
> version 9 of DB2, payload files (tarballs).  Basically the procedure is
> pick the payload files you need, extract them all into a specific
> directory, and then run this 1 command and viola, you have a working
> install to begin setting up.  Sounded simple enough, and doing it
> works great.  So I implemented this in the spec file, extracting all the
> files and such in %install and then running the command in the %post
> section (this way any locally specific voodoo can happen).  The
> installation all looks great, but when I try to execute one of the
> that is part of the standard configuration setup (specifically I'm trying
> to run`$IBMDIR/instance/dascrt -u dasusr1`) it fails.
> So far I've verified all the files permissions and ownerships, everything
> matches.  I also have SELinux disabled.  As far as I can tell they are
> identical, except the RPM install won't let me run the commands, but the
> manual one will.  I have also tried setting the RPM to not run the 1
> command so that I do it manually after the files have just been placed on
> the system (which should remove and enviromental idfferences cause now it
> also gets run directly as root) and it still doesnt work.
> My biggest problem is that I just don't know where to go from here, any
> insight would be great.
> thanks
> -greg swift

So after 3+ weeks of bashing my head on this RPM issue I narrowed it down
to the fact that RPM strips the symbols out of objects during the build
process.  the `%define debug_package %{nil}` did not prevent this as most
of what I read stated it would, and so I found an old post[1] from 2002 on
this list telling how Xfree86 prevented the issue.  They used `%define
__spec_install_port /usr/lib/rpm/brp-compress` so now all is better as far
as I can tell.  Anyone in the same boat can reference the details of my
journey on IBM's developerWorks forum[2].

Have fun.



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]