5 ms·

"Even the linux die-hards have essentially grasped the necessity of systemd (Even though they're going to hate on it for awhile longer)"

by Spidler 12y ago

ausjke 12y ago
the only reason I'm re-interested in bsd is that it does not have systemd, if BSD wants systemd-alike init I will stop right here... additionally, for the vast majority linux systems that are actually embedded, BSD needs a solution for that.
- gpvos 12y ago
  OpenBSD does not seem to have any interest in systemd-lookalikes.
- ghshephard 12y ago
  The one great thing about OpenBSD, is that it provides a consistent, reliable, cleanly documented, constantly maintained, operating system. There is such an emphasis on consistency, that a solid OpenBSD administrator, who last worked on OpenBSD in 2004, and then, 20 versions later, (OpenBSD releases a new major version, along with a full set of architecture releases and binary packages every six months, like clockwork) - would have zero problem understanding all of the system concepts. A few things have changed/updated/evolved, but a couple hours with the man pages would bring them right up to date. OpenBSD is about very gradual evolution, not revolution. My sense is that OpenBSD is not designed for Watches, Tablets, or SmartPhones. It is a world class Server, Firewall, Router, networking device, networking appliance operating system, and it's unlikely to ever lose that focus. I'm pretty confident that we won't see anything like systemd confusing everyone on OpenBSD as long as Theo continues his steady management at the helm. Systemd actually serves some pretty important functions on systems that have a lot of dynamic services going up and down with dependencies, particularly on low-power devices, and on systems that require rapid parallel process initiation - so I'm not hating on systemd here, just appreciating the consistency that we find in OpenBSD.
  - clarry 12y ago
    > OpenBSD is about very gradual evolution, not revolution. To make a concrete example, look at ifconfig and ip (from iproute2 I guess?) on Linux. ifconfig was used to be the one tool to rule them all, everybody knew it, everybody used it. At some point somebody decided they need a new tool, so now there are two, and the other is said to be deprecated. The fun thing is that ifconfig still mostly works, and you find lots of documentation/tutorials/wikis telling you to use it. But at some point you might bump into some really weird issue which turns out to be because the tool you're using is.. well, deprecated. I did have such a problem myself (related to IPv6, years ago) and it wasn't much fun, because the tool looked like it's got support for the thing I was trying to do, but it would just silently and weirdly fail... In the OpenBSD land, a network interface is a network interface. You use ifconfig. IPv6 or not. Wireless or not. The same tool everybody knew has been evolved to support the new stuff. The developers are careful about needlessly changing the user facing part of a tool in a way that would require him to re-learn it or another tool. Of course, sometimes changes are inevitable, and these changes are usually documented in the upgrade guide (and the page which documents changes to the -current branch). So there are few surprises. Sometimes the underlying tool might change entirely to another implementation, yet the user facing parts are made for most part finger-compatible with the old tool. For example, mandoc(1) has replaced historical tools like man(1), apropos(1), etc. And sendmail has been replaced by smtpd, but smtpd still provides newaliases and makemap. So they don't push shiny new poorly tested tools and tell everyone to relearn everything all the time because shiny new features (and the old tool was cranky and nobody wanted to fix it).
    - emmelaich 12y ago
      ifconfig didn't support infiniband and hadn't been maintained for many years. It was irritating at first but ip is a better, more capable tool. http://serverfault.com/questions/458628/should-i-quit-using-ifconfig http://serverfault.com/questions/458628/should-i-quit-using-...
      - ghshephard 12y ago
        I've never used "ip" in a sysadmin setting - the only command I've ever used on a nix (bsd, OS X, Solars, Linux) is ifconfig - which may just indicate I'm using the wrong command, but also demonstrates the issues involved in switching to a new command. Re- ifconfig/infiniband - this is ironic - because when I google "configure infiniband linux" - every page I come up with only mentions ifconfig, and never ip. (though, ifconfig is just used to display the interfaces, and it's usually other commands like ib that seem to be used to configure the infiniband interfaces)
- justincormack 12y ago
  NetBSD is your traditional BSD for embedding. With nothing like systemd... give it a try.
digi_owl 12y ago
For me at least the problem is not the basic "trigger A on event B", but the whole jungle of systemd-*d's that have sprouted and their tight coupling to systemd-as-pid1.
marcosdumay 12y ago
Linux (and yes, FreeBSD) really needs something that provides that kind of functionality. It needs this so much that distros are wiling to put up even with systemd to get it. The only problem is that systemd is a bad piece ("piece"? Or rater "lot"?) of software.
- jacquesm 12y ago
  I can't comment on whether or not it is bad, it is badly designed and re-invents wheels that did not need re-inventing whilst re-introducing bugs from the past. The role of systemd may be a good one, the design and implementation have nothing to do with the role. But with systemds role being defined as 'anything we can grab' there is no end to what will be sucked up into the black hole and re-implemented in a half-baked manner.
vezzy-fnord 12y ago
The architecture he describes, that of an event broker, sounds a lot more like Upstart than it does like systemd, though. That said, the proposed idea doesn't sound like something that is trivial at all, but rather pretty bug-prone. There already are de facto event engines for certain areas like device hotplugging (formerly HAL and now udev on Linux, devd on FreeBSD), whereas for a lot of other events in Linux software a typical way is to subscribe on an application's D-Bus interface and wait for signals. Indeed, OS X's notify API looks precisely like a message queue from my brief observations. But then there's certain events and classes of events (such as "network is up/network is down") that could have literally over a dozen meanings. I'm not sure how a practical centralized event broker could be anything more than a glorified message bus? Otherwise, encapsulating highly multi-faceted and non-deterministic system states into generic events sounds like bugs galore.
- krakensden 12y ago
  > The architecture he describes, that of an event broker, sounds a lot more like Upstart than it does like systemd, though. Not dbus? Because that + the notify api sounds... a lot like dbus.
dredmorbius 12y ago
As much as I don't like how an entire monolithic tightly-coupled complex systemd infrastructure is being crammed down the Linux pipe, and as much as I've argued against it in the past few days here on HN, there are some useful things out of systemd. The questions are: * How much of that needs to go in PID 1? * How much is required to go together? To what extent can the system be made more modular and separable? * What allowances are going to be made for legacy operation modes? The concerns I've got are too much being put into highly critical process space while issues are shouted down and/or disparaged. My sense is that the systemd ship is going to encounter the rest-of-linux ship and both are going to be significantly changed. It's quite likely that some key personnel on both sides will get ejected in the process. I'm not happy with the present course and state of affairs.
- reality_czech 12y ago
  It's quite likely that you and the other "systemd truthers" will forget all about this in a year or two and systemd will continue just as it has. I've been using it in my Linux distro for at least a year. It works fine-- quicker startup, handles hotplug correctly, rational management of services, and easy to add new ones. I'm sorry, there is no nice way to say this, but the systemd truther crowd is a lot like the Obama birth certificate crowd, or the anti-vaxxer crowd. Repeating the same discredited arguments over and over. And for the most part, composed of people who have no experience in what they're talking about, so why should their opinion even matter? Lennart's only mistake was to be bad at public relations and rather blunt about what he said in public. (I, on the other hand, can afford to be blunt because I'm just an anonymous guy, not a project manager).
  - dredmorbius 12y ago
    Are the slurs really necessary? The "it works for me" fails to address those for whom it hasn't worked. Which includes, I'll note, any number of people who might generally be considered somewhat past beginner status with Linux: Alan Cox, Ted T'so, Bruce Perens, John Goerzen, among many others: http://www.reddit.com/r/linux/comments/2dgy45/if_systemd_is_so_controversial_why_has_it_been_so/cjrlu4v http://www.reddit.com/r/linux/comments/2dgy45/if_systemd_is_... Equating them to Donald Trump and a mob of Koch-backed disinformation shills isn't simply wrong, but trolling, of the level of Donald Trump and a mob of Koch-backed disinformation shills. I'll note once again that the focus on insults and lack of addressing specific points and issues really cheapens the case for systemd, and its proponents.

"Even the linux die-hards have essentially grasped the necessity of systemd (Even though they're going to hate on it for awhile longer)"

by Spidler 12y ago

"Even the linux die-hards have essentially grasped the necessity of systemd (Even though they're going to hate on it for awhile longer)"

ausjke 12y ago

the only reason I'm re-interested in bsd is that it does not have systemd, if BSD wants systemd-alike init I will stop right here... additionally, for the vast majority linux systems that are actually embedded, BSD needs a solution for that.

gpvos 12y ago
OpenBSD does not seem to have any interest in systemd-lookalikes.
ghshephard 12y ago
The one great thing about OpenBSD, is that it provides a consistent, reliable, cleanly documented, constantly maintained, operating system. There is such an emphasis on consistency, that a solid OpenBSD administrator, who last worked on OpenBSD in 2004, and then, 20 versions later, (OpenBSD releases a new major version, along with a full set of architecture releases and binary packages every six months, like clockwork) - would have zero problem understanding all of the system concepts. A few things have changed/updated/evolved, but a couple hours with the man pages would bring them right up to date. OpenBSD is about very gradual evolution, not revolution. My sense is that OpenBSD is not designed for Watches, Tablets, or SmartPhones. It is a world class Server, Firewall, Router, networking device, networking appliance operating system, and it's unlikely to ever lose that focus. I'm pretty confident that we won't see anything like systemd confusing everyone on OpenBSD as long as Theo continues his steady management at the helm. Systemd actually serves some pretty important functions on systems that have a lot of dynamic services going up and down with dependencies, particularly on low-power devices, and on systems that require rapid parallel process initiation - so I'm not hating on systemd here, just appreciating the consistency that we find in OpenBSD.
- clarry 12y ago
  > OpenBSD is about very gradual evolution, not revolution. To make a concrete example, look at ifconfig and ip (from iproute2 I guess?) on Linux. ifconfig was used to be the one tool to rule them all, everybody knew it, everybody used it. At some point somebody decided they need a new tool, so now there are two, and the other is said to be deprecated. The fun thing is that ifconfig still mostly works, and you find lots of documentation/tutorials/wikis telling you to use it. But at some point you might bump into some really weird issue which turns out to be because the tool you're using is.. well, deprecated. I did have such a problem myself (related to IPv6, years ago) and it wasn't much fun, because the tool looked like it's got support for the thing I was trying to do, but it would just silently and weirdly fail... In the OpenBSD land, a network interface is a network interface. You use ifconfig. IPv6 or not. Wireless or not. The same tool everybody knew has been evolved to support the new stuff. The developers are careful about needlessly changing the user facing part of a tool in a way that would require him to re-learn it or another tool. Of course, sometimes changes are inevitable, and these changes are usually documented in the upgrade guide (and the page which documents changes to the -current branch). So there are few surprises. Sometimes the underlying tool might change entirely to another implementation, yet the user facing parts are made for most part finger-compatible with the old tool. For example, mandoc(1) has replaced historical tools like man(1), apropos(1), etc. And sendmail has been replaced by smtpd, but smtpd still provides newaliases and makemap. So they don't push shiny new poorly tested tools and tell everyone to relearn everything all the time because shiny new features (and the old tool was cranky and nobody wanted to fix it).
  - emmelaich 12y ago
    ifconfig didn't support infiniband and hadn't been maintained for many years. It was irritating at first but ip is a better, more capable tool. http://serverfault.com/questions/458628/should-i-quit-using-ifconfig http://serverfault.com/questions/458628/should-i-quit-using-...
    - ghshephard 12y ago
      I've never used "ip" in a sysadmin setting - the only command I've ever used on a nix (bsd, OS X, Solars, Linux) is ifconfig - which may just indicate I'm using the wrong command, but also demonstrates the issues involved in switching to a new command. Re- ifconfig/infiniband - this is ironic - because when I google "configure infiniband linux" - every page I come up with only mentions ifconfig, and never ip. (though, ifconfig is just used to display the interfaces, and it's usually other commands like ib that seem to be used to configure the infiniband interfaces)
justincormack 12y ago
NetBSD is your traditional BSD for embedding. With nothing like systemd... give it a try.

digi_owl 12y ago

For me at least the problem is not the basic "trigger A on event B", but the whole jungle of systemd-*d's that have sprouted and their tight coupling to systemd-as-pid1.

marcosdumay 12y ago

Linux (and yes, FreeBSD) really needs something that provides that kind of functionality. It needs this so much that distros are wiling to put up even with systemd to get it. The only problem is that systemd is a bad piece ("piece"? Or rater "lot"?) of software.

jacquesm 12y ago
I can't comment on whether or not it is bad, it is badly designed and re-invents wheels that did not need re-inventing whilst re-introducing bugs from the past. The role of systemd may be a good one, the design and implementation have nothing to do with the role. But with systemds role being defined as 'anything we can grab' there is no end to what will be sucked up into the black hole and re-implemented in a half-baked manner.

vezzy-fnord 12y ago

The architecture he describes, that of an event broker, sounds a lot more like Upstart than it does like systemd, though. That said, the proposed idea doesn't sound like something that is trivial at all, but rather pretty bug-prone. There already are de facto event engines for certain areas like device hotplugging (formerly HAL and now udev on Linux, devd on FreeBSD), whereas for a lot of other events in Linux software a typical way is to subscribe on an application's D-Bus interface and wait for signals. Indeed, OS X's notify API looks precisely like a message queue from my brief observations. But then there's certain events and classes of events (such as "network is up/network is down") that could have literally over a dozen meanings. I'm not sure how a practical centralized event broker could be anything more than a glorified message bus? Otherwise, encapsulating highly multi-faceted and non-deterministic system states into generic events sounds like bugs galore.

krakensden 12y ago
> The architecture he describes, that of an event broker, sounds a lot more like Upstart than it does like systemd, though. Not dbus? Because that + the notify api sounds... a lot like dbus.

dredmorbius 12y ago

As much as I don't like how an entire monolithic tightly-coupled complex systemd infrastructure is being crammed down the Linux pipe, and as much as I've argued against it in the past few days here on HN, there are some useful things out of systemd. The questions are: * How much of that needs to go in PID 1? * How much is required to go together? To what extent can the system be made more modular and separable? * What allowances are going to be made for legacy operation modes? The concerns I've got are too much being put into highly critical process space while issues are shouted down and/or disparaged. My sense is that the systemd ship is going to encounter the rest-of-linux ship and both are going to be significantly changed. It's quite likely that some key personnel on both sides will get ejected in the process. I'm not happy with the present course and state of affairs.

reality_czech 12y ago
It's quite likely that you and the other "systemd truthers" will forget all about this in a year or two and systemd will continue just as it has. I've been using it in my Linux distro for at least a year. It works fine-- quicker startup, handles hotplug correctly, rational management of services, and easy to add new ones. I'm sorry, there is no nice way to say this, but the systemd truther crowd is a lot like the Obama birth certificate crowd, or the anti-vaxxer crowd. Repeating the same discredited arguments over and over. And for the most part, composed of people who have no experience in what they're talking about, so why should their opinion even matter? Lennart's only mistake was to be bad at public relations and rather blunt about what he said in public. (I, on the other hand, can afford to be blunt because I'm just an anonymous guy, not a project manager).
- dredmorbius 12y ago
  Are the slurs really necessary? The "it works for me" fails to address those for whom it hasn't worked. Which includes, I'll note, any number of people who might generally be considered somewhat past beginner status with Linux: Alan Cox, Ted T'so, Bruce Perens, John Goerzen, among many others: http://www.reddit.com/r/linux/comments/2dgy45/if_systemd_is_so_controversial_why_has_it_been_so/cjrlu4v http://www.reddit.com/r/linux/comments/2dgy45/if_systemd_is_... Equating them to Donald Trump and a mob of Koch-backed disinformation shills isn't simply wrong, but trolling, of the level of Donald Trump and a mob of Koch-backed disinformation shills. I'll note once again that the focus on insults and lack of addressing specific points and issues really cheapens the case for systemd, and its proponents.