Hi testers, on our Monday IRC testing meeting, I proposed to test current Factory between RC1 and RC2. Now I found https://bugzilla.novell.com/show_bug.cgi?id=616940 which might affect all 32-bit systems. As workaround, giving mem=800M on bootloader params seemed to work. Keeping the old 2.6.34-9 kernel around would also be good. bug looks like this: http://www3.zq1.de/opensuse/video/openSUSE-KDE-LiveCD-i686-Build0684.ogv Last good Build was 0682. This once more shows that testing is much needed as it is easy for developers to break things without noticing. Ciao Bernhard M. -- To unsubscribe, e-mail: opensuse-testing+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-testing+help@opensuse.org
On 06/24/2010 03:46 AM, Bernhard Wiedemann wrote:
Hi testers,
on our Monday IRC testing meeting, I proposed to test current Factory between RC1 and RC2. Now I found https://bugzilla.novell.com/show_bug.cgi?id=616940 which might affect all 32-bit systems. As workaround, giving mem=800M on bootloader params seemed to work. Keeping the old 2.6.34-9 kernel around would also be good.
bug looks like this: http://www3.zq1.de/opensuse/video/openSUSE-KDE-LiveCD-i686-Build0684.ogv
Last good Build was 0682.
This once more shows that testing is much needed as it is easy for developers to break things without noticing.
Anyone else have the feeling that we are trying to walk through quicksand? Larry -- To unsubscribe, e-mail: opensuse-testing+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-testing+help@opensuse.org
Re; quicksand, I ran 11.3 Milestone 2 up to Milestone 5 and now am running 11.3 RC1 Gnome version on an Eee PC 701 Celeron and so far it has been very smooth and stable. Installation was without issues as was first boot and final set up. If new issues of a serious nature are popping up going into RC2, is it
Quoting Larry Finger
Hi testers,
on our Monday IRC testing meeting, I proposed to test current Factory between RC1 and RC2. Now I found https://bugzilla.novell.com/show_bug.cgi?id=616940 which might affect all 32-bit systems. As workaround, giving mem=800M on bootloader params seemed to work. Keeping the old 2.6.34-9 kernel around would also be good.
bug looks like this: http://www3.zq1.de/opensuse/video/openSUSE-KDE-LiveCD-i686-Build0684.ogv
Last good Build was 0682.
This once more shows that testing is much needed as it is easy for developers to break things without noticing.
Anyone else have the feeling that we are trying to walk through quicksand?
Larry -- To unsubscribe, e-mail: opensuse-testing+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-testing+help@opensuse.org
-- To unsubscribe, e-mail: opensuse-testing+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-testing+help@opensuse.org
On 06/24/2010 03:46 AM, Bernhard Wiedemann wrote:
Hi testers,
on our Monday IRC testing meeting, I proposed to test current Factory between RC1 and RC2. Now I found https://bugzilla.novell.com/show_bug.cgi?id=616940 which might affect all 32-bit systems. As workaround, giving mem=800M on bootloader params seemed to work. Keeping the old 2.6.34-9 kernel around would also be good.
bug looks like this: http://www3.zq1.de/opensuse/video/openSUSE-KDE-LiveCD-i686-Build0684.ogv
Last good Build was 0682.
I downloaded Build0685 of the i686 KDE Live CD and found that it booted fine on my VirtualBox VM. I also downloaded the i586 NET install iso today and used it to install onto the same VM. Other than the usual problem with HAL not being installed, it seems to work fine. BTW, HAL was needed to get wireless networking running. Build0675, which is the official release of the NET Install iso failed to boot on this machine and hung at the loading basic drivers step. There is some progress; however, despite assurances that the new yast2-nfs-client version has been pushed to Factory, it is not yet available and Bug #608166 is still present. Larry -- To unsubscribe, e-mail: opensuse-testing+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-testing+help@opensuse.org
Larry Finger wrote:
On 06/24/2010 03:46 AM, Bernhard Wiedemann wrote:
Hi testers,
on our Monday IRC testing meeting, I proposed to test current Factory between RC1 and RC2. Now I found https://bugzilla.novell.com/show_bug.cgi?id=616940 which might affect all 32-bit systems. As workaround, giving mem=800M on bootloader params seemed to work. Keeping the old 2.6.34-9 kernel around would also be good.
bug looks like this: http://www3.zq1.de/opensuse/video/openSUSE-KDE-LiveCD-i686-Build0684.ogv
Last good Build was 0682.
I downloaded Build0685 of the i686 KDE Live CD and found that it booted fine on my VirtualBox VM. I also downloaded the i586 NET install iso today and used it to install onto the same VM. Other than the usual problem with HAL not being installed, it seems to work fine. BTW, HAL was needed to get wireless networking running.
Build0675, which is the official release of the NET Install iso failed to boot on this machine and hung at the loading basic drivers step. There is some progress; however, despite assurances that the new yast2-nfs-client version has been pushed to Factory, it is not yet available and Bug #608166 is still present.
Larry
The bugfix was already in the pipeline so that the buggy kernel was only distributed for 20-40 hours and Build0675 is indeed good again. http://www3.zq1.de/opensuse/video/openSUSE-NET-i586-Build0685b.ogv http://www3.zq1.de/opensuse/video/openSUSE-NET-x86_64-Build0685.ogv However, this again illustrates the point I tried to make at http://lizards.opensuse.org/2010/06/12/improve-software-quality/ Until yesterday I only thought of "what could they do". But I found that I could do something about that, too. OpenSUSE is probably a "doocracy". A first step could be that I add to my mirror a "factory-testing" repo directory, that will only get a copy of factory after it passed automated tests. While this will need a few extra hours in the pipeline, it could prevent such breakages, thus improving on the experience for testers without having to wait on milestone releases (which even broke in the past, too, so factory-testing could be better than that in both quality and turn-around-time). As for the yast2-nfs-client issue I found this page useful. https://build.opensuse.org/package/show?package=yast2-nfs-client&project=openSUSE%3AFactory when you click on the green "succeeded" it shows yast2-nfs-client-2.19.3-1.1 to mention the fix for bnc#608166 this version is already on http://download.opensuse.org/factory/repo/oss/suse/noarch/. but maybe the fix was not good? Then the bug should be reopened. also in the news: http://www.phoronix.com/scan.php?page=article&item=linux_2010_fiveway&num=1 Ciao Bernhard M. -- To unsubscribe, e-mail: opensuse-testing+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-testing+help@opensuse.org
On 06/25/2010 01:05 AM, Bernhard M. Wiedemann wrote:
The bugfix was already in the pipeline so that the buggy kernel was only distributed for 20-40 hours and Build0675 is indeed good again. http://www3.zq1.de/opensuse/video/openSUSE-NET-i586-Build0685b.ogv http://www3.zq1.de/opensuse/video/openSUSE-NET-x86_64-Build0685.ogv
However, this again illustrates the point I tried to make at http://lizards.opensuse.org/2010/06/12/improve-software-quality/
That is a good article. I agree that we are in somewhat of a downward spiral. As long as factory is so buggy that it cannot be tested, then no one can/will test, thus it gets even worse.
Until yesterday I only thought of "what could they do". But I found that I could do something about that, too. OpenSUSE is probably a "doocracy". A first step could be that I add to my mirror a "factory-testing" repo directory, that will only get a copy of factory after it passed automated tests. While this will need a few extra hours in the pipeline, it could prevent such breakages, thus improving on the experience for testers without having to wait on milestone releases (which even broke in the past, too, so factory-testing could be better than that in both quality and turn-around-time).
Good idea. It is necessary that it build, but certainly not sufficient. With the new repo, a user can be resonably certain that a new build will at least boot. If you are OK with the bandwidth, let me know when this is setup and I will publicize it in the "prerelease-beta" forum and in the weekly news.
As for the yast2-nfs-client issue I found this page useful. https://build.opensuse.org/package/show?package=yast2-nfs-client&project=openSUSE%3AFactory
when you click on the green "succeeded" it shows yast2-nfs-client-2.19.3-1.1 to mention the fix for bnc#608166
this version is already on http://download.opensuse.org/factory/repo/oss/suse/noarch/. but maybe the fix was not good? Then the bug should be reopened.
That version works on both x86_64 and i386. I forgot that it is a noarch package, thus I didn't find it. It has been in Factory since June 18, and must have just missed RC1.
also in the news: http://www.phoronix.com/scan.php?page=article&item=linux_2010_fiveway&num=1
Well, openSUSE didn't shine in those tests, but it didn't fall flat on its face either. It would have been terrible if RC1 had failed to boot/install on their system. Larry -- To unsubscribe, e-mail: opensuse-testing+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-testing+help@opensuse.org
participants (4)
-
Bernhard M. Wiedemann
-
Bernhard Wiedemann
-
kelsh@sentex.net
-
Larry Finger