Fixing buildsys CC
On 11/15/2014 01:29 PM, Cole Robinson wrote:
Hi all,
The ISO volume ID for the Fedora 21 workstation livecd is 'Fedora-Live-Workstation-x86_64-2', which doesn't actually say 21 in it, and libosinfo can't use it to detect the fedora version for boxes/virt-manager.
Koji is responsible for generating the volume ID, and there's a bug filed:
https://bugzilla.redhat.com/show_bug.cgi?id=1145264
I also sent a (untested) koji patch to buildsys list:
https://lists.fedoraproject.org/pipermail/buildsys/2014-October/004411.html
However neither have received a response, so I'm reaching out here.
If this isn't fixed before F21 GA, we are stuck with the crappy volume ID forever and tools that are depending on it are in a tough spot. Is there any process to escalate this and get the fix deployed to infrastructure? (FWIW I've just proposed the bug as a freeze exception, but presumably it will take more coordination that that)
Thanks, Cole
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Sat, 15 Nov 2014 13:31:42 -0500 Cole Robinson crobinso@redhat.com wrote:
Fixing buildsys CC
On 11/15/2014 01:29 PM, Cole Robinson wrote:
Hi all,
The ISO volume ID for the Fedora 21 workstation livecd is 'Fedora-Live-Workstation-x86_64-2', which doesn't actually say 21 in it, and libosinfo can't use it to detect the fedora version for boxes/virt-manager.
Koji is responsible for generating the volume ID, and there's a bug filed:
https://bugzilla.redhat.com/show_bug.cgi?id=1145264
I also sent a (untested) koji patch to buildsys list:
https://lists.fedoraproject.org/pipermail/buildsys/2014-October/004411.html
NAK on this patch its the completely wrong place to apply such a thing. the fix belongs in livecd-tools
However neither have received a response, so I'm reaching out here.
If this isn't fixed before F21 GA, we are stuck with the crappy volume ID forever and tools that are depending on it are in a tough spot. Is there any process to escalate this and get the fix deployed to infrastructure? (FWIW I've just proposed the bug as a freeze exception, but presumably it will take more coordination that that)
Dennis
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Sun, 16 Nov 2014 19:09:25 -0600 Dennis Gilmore dennis@ausil.us wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Sat, 15 Nov 2014 13:31:42 -0500 Cole Robinson crobinso@redhat.com wrote:
Fixing buildsys CC
On 11/15/2014 01:29 PM, Cole Robinson wrote:
Hi all,
The ISO volume ID for the Fedora 21 workstation livecd is 'Fedora-Live-Workstation-x86_64-2', which doesn't actually say 21 in it, and libosinfo can't use it to detect the fedora version for boxes/virt-manager.
Koji is responsible for generating the volume ID, and there's a bug filed:
https://bugzilla.redhat.com/show_bug.cgi?id=1145264
I also sent a (untested) koji patch to buildsys list:
https://lists.fedoraproject.org/pipermail/buildsys/2014-October/004411.html
NAK on this patch its the completely wrong place to apply such a thing. the fix belongs in livecd-tools
of course though things need to be done differently in koji. because we would need to pass through the full untrunckated value or do something else that involves ignoring the CLI. But I do stand by koji being the wrong place for that knowledge. especially true if we change the naming scheme going forward. which is something we intend to do. keep them under 32 characters but make them a bit more readable. really libosinfo should be looking in the image and not relying on the filesystem volumeid to determine the release of an image. something outside of the contents could be manipulated, while it is much less likely to happen inside the image.
Denni
However neither have received a response, so I'm reaching out here.
If this isn't fixed before F21 GA, we are stuck with the crappy volume ID forever and tools that are depending on it are in a tough spot. Is there any process to escalate this and get the fix deployed to infrastructure? (FWIW I've just proposed the bug as a freeze exception, but presumably it will take more coordination that that)
Dennis
On 11/16/2014 08:26 PM, Dennis Gilmore wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Sun, 16 Nov 2014 19:09:25 -0600 Dennis Gilmore dennis@ausil.us wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Sat, 15 Nov 2014 13:31:42 -0500 Cole Robinson crobinso@redhat.com wrote:
Fixing buildsys CC
On 11/15/2014 01:29 PM, Cole Robinson wrote:
Hi all,
The ISO volume ID for the Fedora 21 workstation livecd is 'Fedora-Live-Workstation-x86_64-2', which doesn't actually say 21 in it, and libosinfo can't use it to detect the fedora version for boxes/virt-manager.
Koji is responsible for generating the volume ID, and there's a bug filed:
https://bugzilla.redhat.com/show_bug.cgi?id=1145264
I also sent a (untested) koji patch to buildsys list:
https://lists.fedoraproject.org/pipermail/buildsys/2014-October/004411.html
NAK on this patch its the completely wrong place to apply such a thing. the fix belongs in livecd-tools
of course though things need to be done differently in koji. because we would need to pass through the full untrunckated value or do something else that involves ignoring the CLI. But I do stand by koji being the wrong place for that knowledge. especially true if we change the naming scheme going forward. which is something we intend to do. keep them under 32 characters but make them a bit more readable.
ccing bcl who originally moved the bug to koji. I'm all for long term fixup but I'd also really like to see some solution before F21 GA. Ideas anyone?
really libosinfo
should be looking in the image and not relying on the filesystem volumeid to determine the release of an image. something outside of the contents could be manipulated, while it is much less likely to happen inside the image.
Yeah that would be a useful capability to have, but it's a whole different world of hurt and a lot more work that isn't going to materialize anytime in the short to midterm. And honestly just depending on volume ID has worked out pretty well already.
- Cole
However neither have received a response, so I'm reaching out here.
If this isn't fixed before F21 GA, we are stuck with the crappy volume ID forever and tools that are depending on it are in a tough spot. Is there any process to escalate this and get the fix deployed to infrastructure? (FWIW I've just proposed the bug as a freeze exception, but presumably it will take more coordination that that)
Dennis
On Mon, Nov 17, 2014 at 04:14:11PM -0500, Cole Robinson wrote:
On 11/16/2014 08:26 PM, Dennis Gilmore wrote:
NAK on this patch its the completely wrong place to apply such a thing. the fix belongs in livecd-tools
[...]
ccing bcl who originally moved the bug to koji. I'm all for long term fixup but I'd also really like to see some solution before F21 GA. Ideas anyone?
Unless there is another easy change that can be made (suggestions, Dennis?), I think we should Do Whatever Works in the short-term fix, and then file a ticket for long-term fix, including reverting the short-term one if necessary. I appreciate that this adds yet more cruft to an already messy system, but in the absence of alternatives....
buildsys@lists.fedoraproject.org
-
Cole Robinson -
Dennis Gilmore -
Matthew Miller