On Thu, Jun 18, 2015 at 5:53 PM, sad man <asadxflow@gmail.com> wrote:

Thanks Tom, That's a great way to go about it but it totally eliminates any programming/development from the task and as it is summer of code I need to implement an open source code base. So do you have any suggestions for me regarding maybe extending your idea to do some development?

:-/  I manually do it right now.

I build the DVD with all packages in the base, install with that, set the cache, build the rest w/ internet access.

Then I tar up the cache, rebuild the DVD w/ only the used packages and install that.  I copy the base and my tar onto where I want to create the repo.  Then I build the rest w/ that repo enabled and the rest enabled.

It's a two step, tedious process to produce the non-internet version.  I'm sure there are better ways to do it :-)  I wish I had time.

 

Otherwise currently based on responses on this thread, I am going for directly interfacing pack-stack with Anaconda (need to do more research on exact nuts & bolts of it).

On 18 June 2015 at 22:44, Tom Buskey <tom@buskey.name> wrote:

On Thu, Jun 18, 2015 at 11:42 AM, Mohammed Arafa <mohammed.arafa@gmail.com> wrote:

I am no programmer but packstack has taken 2 years or more to get where it is today by a team of developers (well definitely more than one :) ). I would not assume a single person can get anything reasonably close to that in one summer.

from a sysadmin point of view, i'd think a kickstart file with a post install section containing the path to packstack and the parameters needed would be the method i'd use to deploy what you want. you can translate that into anaconda syntax/language/format if you like

packstack works quite well for non-HA, single node installs.

If you want to eliminate internet access after the OS install

• kickstart a minimal base
  
• login & rpm -qa | sort > pkgfile
• set the yum cache to save all yum installed packages
• run packstack & anything else you need to install
• save all the packages from the yum cache
• add the rpms listed in the pkgfile to that store
• run createrepo against it to create your private repo

Now kickstart another base

put the private repo somewhere

create a repo file pointing at the private with file://

I'd suggest putting the original repo files on, but disabled.  If you need a yum update beyond what's on the DVD, you can enable them and get them over the net as needed.

 

On Thu, Jun 18, 2015 at 11:26 AM, sad man <asadxflow@gmail.com> wrote:

Thanks a lot, So you are suggesting that I add RDO packages in ISO and integrate packstack with Anaconda instead of writing my own OpenStack installer script?

On 18 June 2015 at 15:52, Haïkel <hguemar@fedoraproject.org> wrote:

Yeah, but configuring your OpenStack deployment from raw packages may
be tricky and
you won't be able to finish your GSoC if you go that path.

Packstack is quite reliable and it will handle most of errors. I
suggest that you include
RDO packages in our ISO, that will remove the dependency on network hence
the biggest failure cause.

Regards,
H.

--

Cheers, 

Asadullah Hussain

_______________________________________________
Rdo-list mailing list
Rdo-list@redhat.com
https://www.redhat.com/mailman/listinfo/rdo-list

To unsubscribe: rdo-list-unsubscribe@redhat.com

--

805010942448935	GR750055912MA	Link to me on LinkedIn

_______________________________________________
Rdo-list mailing list
Rdo-list@redhat.com
https://www.redhat.com/mailman/listinfo/rdo-list

To unsubscribe: rdo-list-unsubscribe@redhat.com

--

Cheers, 

Asadullah Hussain