Hi,
there is a request for not installing the PReP partition on PPC system that don't really require it in bug 1303219 [1]. I've spend some time working on it and I think I have the solution, see [2] for details. Because the testing requires a bare-metal Power machine I'm now thinking how to allow testing it with the community in the least intrusive way. Right now my plan is to have the anaconda PR merged and use an updates.img for the corresponding blivet change. This way anaconda itself should work as before and only when used with the updates.img it should change the behaviour. What do you think?
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1303219 [2] https://bugzilla.redhat.com/show_bug.cgi?id=1303219#c5
With regards,
Dan
Hi Dan,
You don't need to merge anything. All changes can be done by one updates image. We are creating updates image with new Blivet changes all the time.
You can do this manually from anaconda source (with your changes) by calling:
./scripts/makeupdates -t <ANACONDA_TAG> -k
Then you will see ./updates/ folder in the anaconda git root. You can copy whole blivet to the path:
./updates/run/install/updates/blivet
That way you can test both Anaconda and Blivet changes together.
Another solution is to use this tool:
https://github.com/rhinstaller/devel-tools/tree/master/anaconda_updates
Anaconda updates should be able to automate some steps but it requires additional configuration.
Jirka
On Mon, 2019-06-10 at 13:54 +0200, Dan Horák wrote:
Hi,
there is a request for not installing the PReP partition on PPC system that don't really require it in bug 1303219 [1]. I've spend some time working on it and I think I have the solution, see [2] for details. Because the testing requires a bare-metal Power machine I'm now thinking how to allow testing it with the community in the least intrusive way. Right now my plan is to have the anaconda PR merged and use an updates.img for the corresponding blivet change. This way anaconda itself should work as before and only when used with the updates.img it should change the behaviour. What do you think?
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1303219 [2] https://bugzilla.redhat.com/show_bug.cgi?id=1303219#c5
With regards,
Dan
Anaconda-devel-list mailing list Anaconda-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/anaconda-devel-list
On Mon, Jun 10, 2019 at 1:56 PM Dan Horák dan@danny.cz wrote:
Hi,
there is a request for not installing the PReP partition on PPC system that don't really require it in bug 1303219 [1]. I've spend some time working on it and I think I have the solution, see [2] for details. Because the testing requires a bare-metal Power machine I'm now thinking how to allow testing it with the community in the least intrusive way. Right now my plan is to have the anaconda PR merged and use an updates.img for the corresponding blivet change. This way anaconda itself should work as before and only when used with the updates.img it should change the behaviour. What do you think?
Hello Dan, the changes for anaconda seem to be safe and reasonable, so I think we could merge them if you want to.
As Jirka suggested, it is also possible to create one updates image with anaconda and blivet changes. I can help you with that.
Vendy
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1303219 [2] https://bugzilla.redhat.com/show_bug.cgi?id=1303219#c5
With regards, Dan
Anaconda-devel-list mailing list Anaconda-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/anaconda-devel-list
On Mon, 10 Jun 2019 15:26:21 +0200 Vendula Poncova vponcova@redhat.com wrote:
On Mon, Jun 10, 2019 at 1:56 PM Dan Horák dan@danny.cz wrote:
Hi,
there is a request for not installing the PReP partition on PPC system that don't really require it in bug 1303219 [1]. I've spend some time working on it and I think I have the solution, see [2] for details. Because the testing requires a bare-metal Power machine I'm now thinking how to allow testing it with the community in the least intrusive way. Right now my plan is to have the anaconda PR merged and use an updates.img for the corresponding blivet change. This way anaconda itself should work as before and only when used with the updates.img it should change the behaviour. What do you think?
Hello Dan, the changes for anaconda seem to be safe and reasonable, so I think we could merge them if you want to.
As Jirka suggested, it is also possible to create one updates image with anaconda and blivet changes. I can help you with that.
finally I've managed to test both physical PowewNV hardware and a KVM guest and the is no problem in the installer. KVM guest installation works as before. The bare metal PowerNV one prints a error during the "bootloader installation" phase, but succeeds anyway. The problem is in grub due "grub2-install: error: the chosen partition is not a PReP partition." which will need fixing there by loosening that requirement.
I'm going to open PRs from my powernv branches in blivet and anaconda.
Dan
On Mon, 24 Jun 2019 15:22:27 +0200 Dan Horák dan@danny.cz wrote:
On Mon, 10 Jun 2019 15:26:21 +0200 Vendula Poncova vponcova@redhat.com wrote:
On Mon, Jun 10, 2019 at 1:56 PM Dan Horák dan@danny.cz wrote:
Hi,
there is a request for not installing the PReP partition on PPC system that don't really require it in bug 1303219 [1]. I've spend some time working on it and I think I have the solution, see [2] for details. Because the testing requires a bare-metal Power machine I'm now thinking how to allow testing it with the community in the least intrusive way. Right now my plan is to have the anaconda PR merged and use an updates.img for the corresponding blivet change. This way anaconda itself should work as before and only when used with the updates.img it should change the behaviour. What do you think?
Hello Dan, the changes for anaconda seem to be safe and reasonable, so I think we could merge them if you want to.
As Jirka suggested, it is also possible to create one updates image with anaconda and blivet changes. I can help you with that.
finally I've managed to test both physical PowewNV hardware and a KVM guest and the is no problem in the installer. KVM guest installation works as before. The bare metal PowerNV one prints a error during the "bootloader installation" phase, but succeeds anyway. The problem is in grub due "grub2-install: error: the chosen partition is not a PReP partition." which will need fixing there by loosening that requirement.
I think I have a solution for this problem as well. There is already a request to loosen the check in grub2 [1], but IMHO a more appropriate solution is to skip installing the grub binaries in anaconda and let anaconda write only the grub config. Which could be achieved by introducing of a PowerNVGRUB2 class with an empty install() method.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1723425
Dan
On Wed, Jun 26, 2019 at 2:10 PM Dan Horák dan@danny.cz wrote:
On Mon, 24 Jun 2019 15:22:27 +0200 Dan Horák dan@danny.cz wrote:
On Mon, 10 Jun 2019 15:26:21 +0200 Vendula Poncova vponcova@redhat.com wrote:
On Mon, Jun 10, 2019 at 1:56 PM Dan Horák dan@danny.cz wrote:
Hi,
there is a request for not installing the PReP partition on PPC system that don't really require it in bug 1303219 [1]. I've spend some time working on it and I think I have the solution, see [2] for details. Because the testing requires a bare-metal Power machine I'm now thinking how to allow testing it with the community in the least intrusive way. Right now my plan is to have the anaconda PR merged and use an updates.img for the corresponding blivet change. This way anaconda itself should work as before and only when used with the updates.img it should change the behaviour. What do you think?
Hello Dan, the changes for anaconda seem to be safe and reasonable, so I think we could merge them if you want to.
As Jirka suggested, it is also possible to create one updates image with anaconda and blivet changes. I can help you with that.
finally I've managed to test both physical PowewNV hardware and a KVM guest and the is no problem in the installer. KVM guest installation works as before. The bare metal PowerNV one prints a error during the "bootloader installation" phase, but succeeds anyway. The problem is in grub due "grub2-install: error: the chosen partition is not a PReP partition." which will need fixing there by loosening that requirement.
I think I have a solution for this problem as well. There is already a request to loosen the check in grub2 [1], but IMHO a more appropriate solution is to skip installing the grub binaries in anaconda and let anaconda write only the grub config. Which could be achieved by introducing of a PowerNVGRUB2 class with an empty install() method.
That sounds good to me.
Vendy
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1723425
Dan
Anaconda-devel-list mailing list Anaconda-devel-list@redhat.com https://www.redhat.com/mailman/listinfo/anaconda-devel-list
anaconda-devel@lists.stg.fedoraproject.org
-
Dan Horák -
jkonecny＠redhat.com -
Vendula Poncova