debian's ongoing systemd / upstart debate
There's been some interesting discussion over the last months on lwn.net about the debian decision on what init system to use in debian jesse, next stable that is.
Here's a brief overview of the more salient points of the ongoing discussion. lwn.net/Articles/582838/ That's a comment that is very precise about some things that systemd does on a technical level, that particular posting is re embedded linux systems, but the person appears to have a good understanding. lwn.net/Articles/582410/#Comments that's in this ongoing current debian/systemd comment thread. I believe what's happening is very very simple: ubuntu started upstart in 2006 because there was no better option. Redhat gave upstart a try then quickly realized it was flawed at core, including the primary dev noting in bug reports that 'that section is a mess', possibly even unfixable. So redhat took what was better about upstart and then redid it, fixing the core glitches and logic errors along the way. This fixed init system, systemd, leveraged the power of the linux kernel to the degree that it is not portable to unix, which is not an argument against it in my opinion, it's an observation that the linux kernel is evolving more rapidly than the unix/bsd kernels, which means that if you try to be fully posix, you will by definition be compromising on functionality. That's by the way what I have found too with gnu tools like grep, sed, gawk, to use the bsd versions is a huge compromise in functionality, and is harder, and less efficient to code solutions, there is that is, nothing better about bsd sed/grep/awk in my opinion. So in a sense, this entire debate is actually a kind of tidal sea change in the nix ecosystem, where linux is actually starting to drop certain ideas because it's gone so far ahead of unix in some areas. Not of course, in simplicity etc. It's been by the way noted for the last decade plus that a lot of linux growth came not at the expense of windows, but of unix systems, and I think we're starting to see the real results of that ongoing move away from unix, in increasingly concrete ways. A situation not unlike the day that the world language group, whoever that is, made Brazilian portugese the official portugese of the world. It's noteworthy that currently only google's chromeos, which is a respin of I think either ubuntu or debian, not sure, uses upstart outside of canonical/ubuntu, and their developer camp is the only place you see anyone arguing in its favor, because it does not actually appear to have any technical merits that make it better than systemd on linux, and their bug handling is very very bad, a guy yesterday at irc told me, I think he was at the debian dev conference, that one guy from ubuntu when pushed on the need for CLA when contributing patches to upstart got all pissy and said: what is it with debian and licensing? That apparently was the nail in the coffin, because that is literally like asking: what is it with openbsd and security? And the audience got that fundamental arrogance, and apparently that is what really turned the tide against upstart, ie, you do not want to be holden to that type of pretty near closed project. Also I was told yesterday that upstart has a very low man hour allowed in the canonical corp, ie, they aren't paying enough to keep it relevant and active. lwn.net/Articles/582503/ that's a comment by jpaletta, which seems highly releavant. :: Quote :: I dont see an advantage for upstart anywhere....
On the surface, the effort to port upstart to bsd looks like a good thing for bsd. But dig deeper, and you look at the 2+ year old, deep design, bugs against upstart, and you look at what Scott has communicated with regard to the rewrite necessary to fix those bugs by using more linux kernel specific mechanisms, like cgroups. And really I'm left wondering how on earth the currently upstart team can really consider porting upstart to any other kernel when the need for a significant rewrite of upstart is just hanging out there on the dev roadmap that Scott left behind, unaddressed and untended to. Interesting. These threads are to me worth reading because they cover all the init system options, since the kfreebsd debian kernel is also under discussion, launchd, openrc etc. lwn.net/Articles/582585/ that's jpaletta providing links to all the core upstart bug reports. Back to top |
All times are GMT - 8 Hours |