[opensuse-factory] Maintainability of base system packages
Hi List, I am wondering what the strategy of the openSUSE distribution is with regards to the base system packages. Lately I have seen that quite a number of these packages (e.g. systemd, dracut) are on very old versions, but contain a very high number of patches (most of the time even more than one hundred). Given this situation I am wonder what happened to the strategy to follow upstream releases and to update packages to the latest versions available (something that Tumbleweed is promising). It seems as if the maintainers here are following an old approach where a package is updated to the latest release when a new distro version is started and then only update the package with patches. This approach might be understandable for a product like SLE, but is this also the right approach for openSUSE ? Also I wonder if with this amount of patches the stability of these important packages (as they are part of the base system) is really guaranteed. In my opinion it would be very hard to keep all those patches separated and updating such packages to the latest version could become a nightmare. In many areas we are trying to stay as close to upstream as possible, but it seems that this is not happening in the most important area (the base system) and I am wondering why. Regards Raymond -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Raymond Wooninck composed on 2015-02-09 21:06 (UTC+0100):
I am wondering what the strategy of the openSUSE distribution is with regards to the base system packages. Lately I have seen that quite a number of these packages (e.g. systemd, dracut) are on very old versions, but contain a very high number of patches (most of the time even more than one hundred).
Given this situation I am wonder what happened to the strategy to follow upstream releases and to update packages to the latest versions available (something that Tumbleweed is promising). It seems as if the maintainers here are following an old approach where a package is updated to the latest release when a new distro version is started and then only update the package with patches. This approach might be understandable for a product like SLE, but is this also the right approach for openSUSE ?
Also I wonder if with this amount of patches the stability of these important packages (as they are part of the base system) is really guaranteed. In my opinion it would be very hard to keep all those patches separated and updating such packages to the latest version could become a nightmare.
In many areas we are trying to stay as close to upstream as possible, but it seems that this is not happening in the most important area (the base system) and I am wondering why.
It seems to me that maintaining an old version with a high number of patches is more like a guarantee of instability than stability, at least WRT systemd, which is in relative terms quite youthful, and in constant state of change as new discoveries of incompatibility are made and dealt with, or, or put on a TODO list, or not. One of these instabilities has been the reboot process. I do a lot of testing that involves restarts. Since the advent of systemd, rebooting has often been a far more lengthy process than it was in sysvinit releases. Last night as example I must have spent over an hour waiting for shutdowns to complete, or fscks on reboot after pulling the plug when shutdown refused to proceed past "failed to store sound card state" or some "starting..." process at a time when stoppings should be the only things happening. http://lists.freedesktop.org/archives/systemd-devel/2015-January/027536.html notes a fix directed to reboot failures. I have to wonder whether it will be seen in openSUSE before the current Tumbleweed morphs into the next openSUSE release, which would not help Evergreen-next, aka 13.1, or 13.2, both of which lately have been stubborn in delaying or preventing shutdown in reasonable time, if allowing shutdown/reboot at all short of using the power switch or plug. -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Mon, Feb 09, 2015 at 03:47:17PM -0500, Felix Miata wrote:
Raymond Wooninck composed on 2015-02-09 21:06 (UTC+0100):
I am wondering what the strategy of the openSUSE distribution is with regards to the base system packages. Lately I have seen that quite a number of these packages (e.g. systemd, dracut) are on very old versions, but contain a very high number of patches (most of the time even more than one hundred).
Given this situation I am wonder what happened to the strategy to follow upstream releases and to update packages to the latest versions available (something that Tumbleweed is promising). It seems as if the maintainers here are following an old approach where a package is updated to the latest release when a new distro version is started and then only update the package with patches. This approach might be understandable for a product like SLE, but is this also the right approach for openSUSE ?
Also I wonder if with this amount of patches the stability of these important packages (as they are part of the base system) is really guaranteed. In my opinion it would be very hard to keep all those patches separated and updating such packages to the latest version could become a nightmare.
In many areas we are trying to stay as close to upstream as possible, but it seems that this is not happening in the most important area (the base system) and I am wondering why.
It seems to me that maintaining an old version with a high number of patches is more like a guarantee of instability than stability, at least WRT systemd, which is in relative terms quite youthful, and in constant state of change as new discoveries of incompatibility are made and dealt with, or, or put on a TODO list, or not.
One of these instabilities has been the reboot process. I do a lot of testing that involves restarts. Since the advent of systemd, rebooting has often been a far more lengthy process than it was in sysvinit releases. Last night as example I must have spent over an hour waiting for shutdowns to complete, or fscks on reboot after pulling the plug when shutdown refused to proceed past "failed to store sound card state" or some "starting..." process at a time when stoppings should be the only things happening.
http://lists.freedesktop.org/archives/systemd-devel/2015-January/027536.html notes a fix directed to reboot failures. I have to wonder whether it will be seen in openSUSE before the current Tumbleweed morphs into the next openSUSE release, which would not help Evergreen-next, aka 13.1, or 13.2, both of which lately have been stubborn in delaying or preventing shutdown in reasonable time, if allowing shutdown/reboot at all short of using the power switch or plug.
Jan already sent a version update request for systemd. Well, Werner is our main systemd maintainer. Perhaps he can comment? For dracut we backported a lot of upstream fixes and Thomas is working on it (and I think is thinking about rebasing it.) Ciao, Marcus -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Le lundi 09 février 2015 à 21:06 +0100, Raymond Wooninck a écrit :
Hi List,
I am wondering what the strategy of the openSUSE distribution is with regards to the base system packages. Lately I have seen that quite a number of these packages (e.g. systemd, dracut) are on very old versions, but contain a very high number of patches (most of the time even more than one hundred).
Given this situation I am wonder what happened to the strategy to follow upstream releases and to update packages to the latest versions available (something that Tumbleweed is promising). It seems as if the maintainers here are following an old approach where a package is updated to the latest release when a new distro version is started and then only update the package with patches. This approach might be understandable for a product like SLE, but is this also the right approach for openSUSE ?
This is only the consequence of being short on manpower in maintaining and fixing packages. Let me explain why : - in 2014, when SLE12 development branch was forked from Factory, version update didn't occur in packages in SLE12 anymore (mostly) - during the stabilisation of the SLE12 codebase, a lot of patches were either written or backported for upstream into SLE12 codebase. We also had a requirement (not followed at 100% unfortunately, we will improve on this) to ensure all changes in SLE12 were also in Factory (so Factory could benefit from stability work done on SLE12) - on some base system packages, some maintainers decided to not spread their time and effort (which was already huge) on trying to maintain two codebases at the same time. Moreover, it was ensuring the stabilisation effort from SLE12 hits Factory (and therefore 13.2) without too much regression caused by updating to new version. This has happened for systemd and dracut and it was maintainers decision. - even after SLE12 (and 13.2) was out, there was still quite some time to ensure bug fixes on those components could benefit both codebase (and again, having the same version in both helps a lot). Now that SLE12 and 13.2 are out for some time, it should be possible to upgrade Factory packages to new upstream versions, but we want to be sure we don't cause regressions compared to what landed in SLE12 ? Why, you might ask ? Simply because it would mean new bugs in Factory and potentially new bugs in SLE13 in the future. I'm basing this on our SLE12 work where we discovered several times fixes in SLE11 which regressed at one point in Factory (I'm not trying to blame anybody here) and was only discovered when we started to harden SLE12 codebase ; we don't want this to happen again, because it does hurt everybody
Also I wonder if with this amount of patches the stability of these important packages (as they are part of the base system) is really guaranteed. In my opinion it would be very hard to keep all those patches separated and updating such packages to the latest version could become a nightmare.
Most of the patches written during SLE12 on systemd / dracut were either: - backport from upstream (new features or bug fixes) - new bug fixes which were pushed to upstream (sometime accepted directly, sometime not, sometime rewritten by upstream later) - new features (in that case, we submit them to upstream). Trust me, our goal is to lower the number of patches to maintain in the long run. But it is difficult to reach that goal, even more with fast moving targets and when some fixes or features are either controversial or too distribution specific.
In many areas we are trying to stay as close to upstream as possible, but it seems that this is not happening in the most important area (the base system) and I am wondering why.
I hope this explains a bit more what happened. -- Frederic Crozat Project Manager Enterprise Desktop SUSE -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Hi Frederic, On Tuesday 10 February 2015 09:38:31 Frederic Crozat wrote:
This is only the consequence of being short on manpower in maintaining and fixing packages.
Let me explain why :
Thanks for the explanation. It confirms what I suspected.
Trust me, our goal is to lower the number of patches to maintain in the long run. But it is difficult to reach that goal, even more with fast moving targets and when some fixes or features are either controversial or too distribution specific.
I fully understand that this is an approach that works for a distribution as SUSE SLE, where stability is key and fast moving targets could potentially cause issues. However the question remain if this is also valid for a distribution like openSUSE where people would expect that even fast moving targets are followed closely. I am not trying to criticize nor blaming anybody here, but the question is if the SLE methodology is really good for openSUSE. As you indicated manpower is in many cases the issue, but in the past we had community maintainers for a number of fast moving packages in Base:System, but somehow those community maintainers disappeared and I wonder if this has anything to do with the fact that a SLE methodology was pushed onto openSUSE. The reason why I am saying this is that somehow this started after the decision that the code-base of SLE and openSUSE should be kept equally so that both could benefit. Regards Raymond -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Tue, Feb 10, 2015 at 10:34:02AM +0100, Raymond Wooninck wrote:
Hi Frederic,
On Tuesday 10 February 2015 09:38:31 Frederic Crozat wrote:
This is only the consequence of being short on manpower in maintaining and fixing packages.
Let me explain why :
Thanks for the explanation. It confirms what I suspected.
Trust me, our goal is to lower the number of patches to maintain in the long run. But it is difficult to reach that goal, even more with fast moving targets and when some fixes or features are either controversial or too distribution specific.
I fully understand that this is an approach that works for a distribution as SUSE SLE, where stability is key and fast moving targets could potentially cause issues. However the question remain if this is also valid for a distribution like openSUSE where people would expect that even fast moving targets are followed closely. I am not trying to criticize nor blaming anybody here, but the question is if the SLE methodology is really good for openSUSE.
As you indicated manpower is in many cases the issue, but in the past we had community maintainers for a number of fast moving packages in Base:System, but somehow those community maintainers disappeared and I wonder if this has anything to do with the fact that a SLE methodology was pushed onto openSUSE. The reason why I am saying this is that somehow this started after the decision that the code-base of SLE and openSUSE should be kept equally so that both could benefit.
As you know Tumbleweed as our rolling release is supposed to be stable. It needs quite some effort to follow systemd version jumps and stay stable. There is a conflict, which is not just in SLE. Ciao, Marcus -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Tue, Feb 10, 2015 at 12:37 PM, Marcus Meissner <meissner@suse.de> wrote:
On Tue, Feb 10, 2015 at 10:34:02AM +0100, Raymond Wooninck wrote:
Hi Frederic,
On Tuesday 10 February 2015 09:38:31 Frederic Crozat wrote:
This is only the consequence of being short on manpower in maintaining and fixing packages.
Let me explain why :
Thanks for the explanation. It confirms what I suspected.
Trust me, our goal is to lower the number of patches to maintain in the long run. But it is difficult to reach that goal, even more with fast moving targets and when some fixes or features are either controversial or too distribution specific.
I fully understand that this is an approach that works for a distribution as SUSE SLE, where stability is key and fast moving targets could potentially cause issues. However the question remain if this is also valid for a distribution like openSUSE where people would expect that even fast moving targets are followed closely. I am not trying to criticize nor blaming anybody here, but the question is if the SLE methodology is really good for openSUSE.
As you indicated manpower is in many cases the issue, but in the past we had community maintainers for a number of fast moving packages in Base:System, but somehow those community maintainers disappeared and I wonder if this has anything to do with the fact that a SLE methodology was pushed onto openSUSE. The reason why I am saying this is that somehow this started after the decision that the code-base of SLE and openSUSE should be kept equally so that both could benefit.
As you know Tumbleweed as our rolling release is supposed to be stable.
It needs quite some effort to follow systemd version jumps and stay stable.
There is a conflict, which is not just in SLE.
Fedora has newer version of systemd than our "Rolling" Tumbleweed. I think Arch has less contributors than openSUSE, but staying very close to upstream allows them to be stable, using lots of patches WILL cause more time wasted on fixing something that is fixed upstream. http://distrowatch.com/table.php?distribution=arch http://distrowatch.com/table.php?distribution=suse There is a "Refresh" button to see all tracked packages, see the versions and compare. I was so excited to hear about rolling openSUSE, but using Fedora stable gives me newer packages. Regards Mustafa Muhammad
Ciao, Marcus -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
-- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On 02/10/2015 04:06 PM, Mustafa Muhammad wrote:
... gives me ...
That's exactly the point what Marcus and Frederic wanted to say: the systemd packagers are short of resources - and they have to maintain both the SLE and the openSUSE versions. So the question is not only what openSUSE *gives me*, but also what I/you *give to* openSUSE? So if you're some kind of programming guy and have some spare time, then pick the chance and help us out. And it's not only primarily about systemd: if you take away some workload of them which they need to spend for other packages, this would be great, too. Have a nice day, Berny -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Tue, Feb 10, 2015 at 6:36 PM, Bernhard Voelker <mail@bernhard-voelker.de> wrote:
On 02/10/2015 04:06 PM, Mustafa Muhammad wrote:
... gives me ...
That's exactly the point what Marcus and Frederic wanted to say: the systemd packagers are short of resources - and they have to maintain both the SLE and the openSUSE versions.
So the question is not only what openSUSE *gives me*, but also what I/you *give to* openSUSE? So if you're some kind of programming guy and have some spare time, then pick the chance and help us out. And it's not only primarily about systemd: if you take away some workload of them which they need to spend for other packages, this would be great, too.
I wish I could maintain some packages, but I can't (due to working and studying), I use pre-releases and report bugs, sometimes make suggestion to improve the project. So this was a suggestion to stay as close to upstream as possible, especially in the rolling Tumbleweed. Regards Mustafa Muhammad
Have a nice day, Berny
-- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Marcus Meissner schrieb:
[...] As you know Tumbleweed as our rolling release is supposed to be stable.
It needs quite some effort to follow systemd version jumps and stay stable.
I'd not call Tumbleweed 'stable' but rather e.g. 'usable for the target audience'. 'stable' is what we have, well, stable releases for. Tumbleweed must be able to follow upstream developments reasonably quick, _especially_ core packages like systemd. Otherwise we'd miss the purpose. The whole idea of the Factory/Tumbleweed merger was to also get the core rolling. By doing that we want to motivate contributors to participate in the core components. We will not achieve that by keeping the SLE12 versions of packages with a heap of patches on top. Preventing bad breakages in core components to hit the published repo is in fact what new development process with openQA and staging projects was made for. So please go ahead and submit the new systemd. The staging process will have to prove itself on packages like that. If a badly broken systemd manages to pass through we have to analyze why and improve the process. cu Ludwig -- (o_ Ludwig Nussel //\ V_/_ http://www.suse.de/ SUSE LINUX GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 21284 (AG Nürnberg) Maxfeldstraße 5; 90409 Nürnberg; Germany -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Hi, after offline discussion with Ludwig and Coolo I have now accepted Jan Engelhardts systemd 218 submission to Base:System and forwarded it to Factory. I will not fix all potential bugs myself though, please help. Ciao, Marcus On Fri, Feb 13, 2015 at 09:51:11AM +0100, Ludwig Nussel wrote:
Marcus Meissner schrieb:
[...] As you know Tumbleweed as our rolling release is supposed to be stable.
It needs quite some effort to follow systemd version jumps and stay stable.
I'd not call Tumbleweed 'stable' but rather e.g. 'usable for the target audience'. 'stable' is what we have, well, stable releases for. Tumbleweed must be able to follow upstream developments reasonably quick, _especially_ core packages like systemd. Otherwise we'd miss the purpose. The whole idea of the Factory/Tumbleweed merger was to also get the core rolling. By doing that we want to motivate contributors to participate in the core components. We will not achieve that by keeping the SLE12 versions of packages with a heap of patches on top. Preventing bad breakages in core components to hit the published repo is in fact what new development process with openQA and staging projects was made for. So please go ahead and submit the new systemd. The staging process will have to prove itself on packages like that. If a badly broken systemd manages to pass through we have to analyze why and improve the process.
cu Ludwig
-- (o_ Ludwig Nussel //\ V_/_ http://www.suse.de/ SUSE LINUX GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 21284 (AG Nürnberg) Maxfeldstraße 5; 90409 Nürnberg; Germany -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
-- Marcus Meissner,SUSE LINUX GmbH; Maxfeldstrasse 5; D-90409 Nuernberg; Zi. 3.1-33,+49-911-740 53-432,,serv=loki,mail=wotan,type=real <meissner@suse.de> -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
El 13/02/15 a las 06:56, Marcus Meissner escribió:
Hi,
after offline discussion with Ludwig and Coolo I have now accepted Jan Engelhardts systemd 218 submission to Base:System and forwarded it to Factory.
You need a dracut update also (it works without the update, but old dracut lacks of some unit files) I fixed a few bugs I found, they are in version 219, out later today. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Friday 2015-02-13 15:26, Cristian Rodríguez wrote:
El 13/02/15 a las 06:56, Marcus Meissner escribió:
Hi,
after offline discussion with Ludwig and Coolo I have now accepted Jan Engelhardts systemd 218 submission to Base:System and forwarded it to Factory.
You need a dracut update also (it works without the update, but old dracut lacks of some unit files)
I fixed a few bugs I found, they are in version 219, out later today.
Good to know; home:jengelh:systemd/systemd is already waiting ;-) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Fri, Feb 13, 2015 at 03:33:43PM +0100, Jan Engelhardt wrote:
On Friday 2015-02-13 15:26, Cristian Rodríguez wrote:
El 13/02/15 a las 06:56, Marcus Meissner escribió:
Hi,
after offline discussion with Ludwig and Coolo I have now accepted Jan Engelhardts systemd 218 submission to Base:System and forwarded it to Factory.
You need a dracut update also (it works without the update, but old dracut lacks of some unit files)
I fixed a few bugs I found, they are in version 219, out later today.
Good to know; home:jengelh:systemd/systemd is already waiting ;-)
Well, I had to remove the setbadness and the dbus service needs to be reviewed. Someone needs to submit a new dracut. Also I got told that you dropped necessary SUSE patches, please check and reinstate and port them. Ciao, Marcus -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
participants (9)
-
Bernhard Voelker -
Cristian Rodríguez -
Felix Miata -
Frederic Crozat -
Jan Engelhardt -
Ludwig Nussel -
Marcus Meissner -
Mustafa Muhammad -
Raymond Wooninck