yum in kickstart %post and HP Insight Mgr

[Bryan Gartner]

Ugo,

Haven't been on the list that long, so haven't reviewed the 3 year old
history archives.

On Fri, Nov 07, 2008 at 03:01:30PM +0000, Ugo Bellavance wrote:
> Brian Long a écrit :
> > I'm trying to use yum in the %post section of kickstart.  With well-
> > written RPMs, this works great.  However, I've come across the lovely HP
> > Insight Manager RPMS (hpasm, hprsm, etc) which seem to run "service
> > hpasm start" during the RPM postinstall (yuck).  This wouldn't be too
> > bad if the daemons actually disassociated from the running terminal.
> >
> > I notice yum install the group "yum groupinstall Insight-Manager" and
> > yum exits, but the rest of the kickstart %post doesn't run.  If I look
> > at the hpasmd processes, they're tied to tty1 and owned by init  :(  If
> > I hit Alt-F2 in kickstart and chroot /mnt/sysimage, service hpasm stop,
> > the rest of the kickstart %post finishes.  I also strace'd the /tmp/ks-
> > script process and it's stuck at waitpid(-1, )
> >
> > I cannot wait on a vendor escalation at this time, so I see one method
> > of a workaround.  Move the yum installs to a firstboot scenario instead
> > of %post.  I'd really like to find another way around this.
> >
> > Any ideas would be appreciated.  Thanks.
> 
> Hi,
> 
> more than 3 years later, is there a solution to this?  We have a bunch
> of HP servers and I'm working to make the installs as automated as
> possible and I'd like to end up with a system with all the HP packages
> installed after anaconda finishes...   How are you guys now proceeding?

I am kind of confused though, since current package versions don't seem
to try to start the services mentioned (since they require a
configuration step).  That said, given some version information for
the query, there is likely a way to cause the agents not to auto-start,
upon package installation, through environment variables at the very least.

Also, since a yum groupinstall command was cited and the packages
aren't referenced that way, I suspect some other interactions may
also be in play with the original post.

Feel free to follow up with the versioning info, and I'll have a look
at it,

bryang