Upstart Now Available In Debian Unstable

[peppepz]

why bother with Upstart when you have systemd? doesn the former has *any* advantage over the latter one?

It's much simpler, and therefore maintainable, it's well documented, it does its job and only that, it compiles, it doesn't require userland to be changed to conform to its model. As a bonus, it's not developed by people who like to force everyone into using their products.

[AdamW]

It's much simpler, and therefore maintainable

Hm.

, it's well documented

So is systemd.

http://www.freedesktop.org/wiki/Software/systemd
http://www.freedesktop.org/software/systemd/man/
http://www.freedesktop.org/wiki/Soft.../TipsAndTricks
http://www.freedesktop.org/wiki/Soft...AskedQuestions
http://freedesktop.org/wiki/Software/systemd/Debugging
http://www.freedesktop.org/wiki/Soft...ompatibilities

and the entire 'systemd for Administrators' and 'Documentation for Developers' sections at http://www.freedesktop.org/wiki/Software/systemd .

it does its job and only that

This is a very short-sighted perspective. It is unrealistic to assume the precise correct job descriptions for all components of a computer were perfected in the 1970s, and all it is now possible to improve is the precise implementations of each defined job. There is no stone tablet anywhere which says 'There Shalt Be An Init Daemon Which Handles Nothing But Process Creation'. It's just a design choice.

As gregkh perceptively pointed out in that whole eudev kerfuffle, when udev first showed up, there were plenty of people who accused it of being too complex and doing too much compared to devfs or static /dev trees (remember those?) Now udev is the paragon of virtue, I don't see too many people clamoring to be allowed to return to maintaining a static /dev tree any more.

it compiles

Er...seems to go out without saying, but so does systemd...

it doesn't require userland to be changed to conform to its model

Neither does systemd. Both upstart and systemd can function as transparent drop-in replacements for sysv if so desired. But if you want to use most of the extended features of *either*, you have to write native services - both upstart and systemd have their own native formats that are different from and incompatible with sysv.

[carrkevinu]

Canonical should really push for adoption of Upstart in order to get mindshare and make people change to it instead of that systemd.

[peppepz]

So is systemd.

The problem is that systemd's documentation is scattered all over many places, many of them informal or unofficial. Also, much of it is bitrotten as systemd changes the way it works quite often, pushing the burden of keeping up onto the users.

and the entire 'systemd for Administrators' and 'Documentation for Developers' sections at http://www.freedesktop.org/wiki/Software/systemd

I wouldn't call an init daemon which is documented through an unordered series of 18 blog posts by its author "well documented".

This is a very short-sighted perspective. It is unrealistic to assume the precise correct job descriptions for all components of a computer were perfected in the 1970s, and all it is now possible to improve is the precise implementations of each defined job. There is no stone tablet anywhere which says 'There Shalt Be An Init Daemon Which Handles Nothing But Process Creation'. It's just a design choice.

I never said that. Upstart, for instance, is much more recent than the 1970s. The value of a system where the boundaries and the functions of each of its components are well defined and well documented, on the other hand, has no age. An administrator can obtain complete knowledge of such a system, reasonably foresee its behaviour, and diagnose faults when things don't work as expected. A kitchen sink that leaves you with a dead console and a wall of text made up of unrelated diagnostic messages because something went wrong is the opposite of that.

As gregkh perceptively pointed out in that whole eudev kerfuffle, when udev first showed up, there were plenty of people who accused it of being too complex and doing too much compared to devfs or static /dev trees (remember those?) Now udev is the paragon of virtue, I don't see too many people clamoring to be allowed to return to maintaining a static /dev tree any more.

I remember well when udev appeared. It replaced devfs, which was a simple approach, of the kind that I like, to the problem of populating /dev. The kernel would automatically make device nodes appear when the related drivers were loaded. ( devfsd on the other hand was broken, but let's ignore that for simplicity. ) The udev design replaced that with an userspace daemon, with certain requirements of its own, which had to be invoked in a special way at a special time during the system boot process. Things that were automatic with devfs (such as booting without an initrd in all the cases where this was previously possible) became no longer directly available with udev, and the protests were quickly dismissed by the udev proponents in the usual way: "fix that in userspace". This meant that, for instance, obtaining a working early userspace changed from being fully automatic into something that required dances of pivot_root and file descriptor voodoo.

Recently, after going out from the door, devfs has come back through the window in the form of devtmpfs, so in the end the "simple" approach won somehow. And the attitude of systemd developers, and their propension to ignore certain problems of their users, are leading the kernel developers to kill udev altogether and have the kernel load firmware on its own again.

Er...seems to go out without saying, but so does systemd...

That's the funny thing, it should go out without saying, but it doesn't. One of the last systemd releases that I tried just didn't compile on 32 bit architectures because it contained a mismatch in a function signature between a header and the implementation. This means that systemd gets released without even checking if it compiles on x86, which I find quite unsettling when we're talking about one of the critical pieces of a Linux installation.

Neither does systemd. Both upstart and systemd can function as transparent drop-in replacements for sysv if so desired. But if you want to use most of the extended features of *either*, you have to write native services - both upstart and systemd have their own native formats that are different from and incompatible with sysv.

Just do a Google search and see for yourself how many third-party packages are being modified to respond to systemd's automatic activation rules. Do the same search for upstart and see which of the two init systems is more invasive. The fact that one should modify a server in order to match the behaviour of a superserver, of which the former shouldn't even know the existence, is a violation of the principle of separation of software components.

That said, in my personal experience, it is very easy to set up upstart as a replacement for sysV. Systemd - not so much. Systemd is even less a "transparent drop-in" for sysvinit than PulseAudio was for plain ALSA.

[kigurai]

I wouldn't call an init daemon which is documented through an unordered series of 18 blog posts by its author "well documented".

Those are not so much documentation as helpful tutorials. I like how you spin extra documentation as something negative.

I thought that "man systemd" looked quite substantial. What is missing?

[storma]

That said, in my personal experience, it is very easy to set up upstart as a replacement for sysV. Systemd - not so much. Systemd is even less a "transparent drop-in" for sysvinit than PulseAudio was for plain ALSA.

I just dropped in two systems to systemd, from Debian Wheezy, wasn't that hard. Follow the documentation. I'll have to brush up on the new way of doing things but that's expected.

[Artemis3]

and: "Debian GNU/Linux can now handle either SysVinit, systemd, and Upstart to handle a head-to-head system booting battle."

Just, please, Phoronix, don't run some benchmarks and Declare a Victor. The extended capabilities of both upstart and systemd versus sysv are far more significant and interesting than any potential improvement in boot times they offer.

And the test should include a system loaded with services (typical desktop, typical server) instead of only a base system.
It would be fair if the starting services include both systemd and upstart files, as explained in the article most things lack systemd and perhaps none has upstart (in Debian).

I think Debian should go with upstart, because systemd is very intrusive and the linux requirement is a big negative (there is kFreebsd, hurd, etc). Of course systemd might actually address those things with time.

Its a fact that upstart is the more mature choice.

Here is Canonical clearly contributing a project developed by them back to the parent distro, so its about time people stop complaining about the "not giving back" excuse, and support this and encourage them to continue.

[ryao]

As gregkh perceptively pointed out in that whole eudev kerfuffle, when udev first showed up, there were plenty of people who accused it of being too complex and doing too much compared to devfs or static /dev trees (remember those?) Now udev is the paragon of virtue, I don't see too many people clamoring to be allowed to return to maintaining a static /dev tree any more.

I hope that I did not contribute to this perception. udev is far from perfect and there are better ways of doing things. However, many packages now depend on udev and moving away from it would be a burden on maintainers, so we appear to be stuck with it.

With that said, there have been suggestions that FreeBSD's devd be ported as an alternative. If udev were not so well established, porting devd would probably work extremely well.

[Teho]

I think Debian should go with upstart, because systemd is very intrusive and the linux requirement is a big negative (there is kFreebsd, hurd, etc).

The planned addition of support for user sessions in Upstart states that:

By making use of a Linux-specific prctl(2) call, we effectively tie Upstart to systems running with a Linux kernel. This is a major restriction, but porting to other systems is already complicated by the fact that even the BSDs do not provide a full POSIX environment (missing "waitid(2)" for example).

So it seems that Upstart isn't properly portable even now and in the future it's even less. It's clearly not any priority either.

Its a fact that upstart is the more mature choice.

It's also a project developed by single company without really any community participation and that has very little developement activity. It's no longer widely used out side of Ubuntu either. It's developed under contributor licence agreement using Bazaar in Launchpad. It seems very Ubuntu/Canonical specific project to me whereas systemd is developed by various companies (Red Hat, Intel, ProFusion, SUSE...) and communites like Arch Linux. It also doesn't have CLA is developed by using git in freedesktop.org. It seems that RHEL7 (Oracle Linux, CentOS, Scientific Linux...) and SLES12 will also be using it.

Here is Canonical clearly contributing a project developed by them back to the parent distro, so its about time people stop complaining about the "not giving back" excuse, and support this and encourage them to continue.

Well they are also doing a lot more harm than good by fragmenting the Linux userspace by using their techically inferior solution.

[JanC]

It's also a project developed by single company without really any community participation and that has very little developement activity. It's no longer widely used out side of Ubuntu either.

Upstart is also part of ChromeOS and used in a number of embedded devices.

(And there is activity in some development branches at least.)