Interview with Lennart Pottering on Linux Piter about Linux changes, systemd, and why attend conferences
Lennart Pottering is one of the legends of the Linux community. Since the 1990s, he has been working on the Linux kernel. Lennart launched such projects as PulseAudio , Avahi , kdbus , systemd and became their main driving force. He currently works for Red Hat in Germany. Last year, Lennart attended the Linux Piter 2017 conference with a report today, on the eve of Linux Piter 2018, we publish an interview with this eminent open source developer, in which he tells why systemd was needed, how the Linux architecture changed and changed, how he personally responded to numerous criticisms, why he needed to attend conferences, and that he personally was given such events like Linux Piter.
An interview with Lennart Pottering is taken by Vasily Tolstoy, member of the Linux Piter Program Program Committee , DELL EMC Lead Programmer, and an active member of the St. Petersburg Linux community .
- Lennart, today you gave a talk at the Linux Piter conference, tell very briefly what your talk was about.
- Perhaps, “containers” is one of the most fashionable words in the Linux world. There are different points of view on what it is, but there are several concepts that are interesting not only for specific modern container technologies. The main idea of my report is to take some of the ideas underlying modern containers, for example, isolating processes and assembling resources into packages, and use them in the classic management of operating system services. After all, until now, classical control systems are used much more widely than containers. In general, the report was about this; In addition, I went into detail in it, how exactly we can isolate processes, which resource packages can be used, about systemd, and how does this relate to how docker and other container systems support this functionality.
Report of Lennart Pottering on Linux Piter 2017 (original version in English):
The same report with simultaneous translation into Russian
- A question for thought. What do you think is interesting in the Linux world over the past year?
- In general, I think LINUX is now so stable a product that it no longer makes major leaps forward, but the existing mechanisms are only improving. Specifically in LINUX nothing radical happened. For example, docker and similar things that have noticeably changed everything that is generally done in IT, they are already about three years old. I like the fact that today there is a constant and steady improvement in Linux. In any case, the basic components that make up the ecosystem, clearly stabilized, and gradually moving for the better. This, in my opinion, is a sign of maturity and professionalism, and this is probably a good thing.
- And what technologies are now on the wave of popularity, if there are any?
- I think it is automation, artificial intelligence. I always liked how different buzzwords in the industry make everyone “electrify” for a while and then disappear. For example, “containers” is actually a magic word already three years ago, but it is still popular. And then IOT became such, and so far no one really knows what it really is and how it differs from ordinary embedded systems. I think this year the trend is artificial intelligence. Even the Linux Foundation is now creating an artificial intelligence project. I think that this is the very “wave” that large companies now want to catch. However, I am not very closely following this topic, although I am watching from the sidelines.
- The Linux community and its key figures are getting older. What are your thoughts on this? Does it have consequences?
- Of course it does. In general, I believe that when Linux appeared in the 90s, it certainly was a revolution in many areas at once. Now this stage is completed, and each project is developing in a very specific direction. And the developers themselves, as a rule, are looking for stability in life in order to calmly manage their project. I do not know whether this is good or not. But, I think, a bad sign is when people get stuck in that way of thinking, which they got used to back in the 90s. I look at it and I wonder how I will think in 10 years? Will my head and thoughts get stuck in the past 10 years ago? Or do I really manage to keep up with the times and in time to understand that the things I knew before are no longer related to the modern IT world? I can see that some Linux developers are better at this than others.
For example, when we were doing systemd, there was a lot of noise. We removed a lot of old concepts that were fixed in Unix on a really religious level. Many of the old stuff in Unix wasn't well thought out from the start, we needed something more modern and dynamic.
We fought a great war, even at the simplest level of control of Linux devices. Now you connect some equipment, something happens, and it becomes available. UNIX was not designed that way. In the best case, it was necessary to turn off the computer, plug the wires, plug in something, run it, and then something would change in the system, but certainly there was no hot plug at all.
It was a big step forward and a real fight - to push this approach through all layers of the system and give life to this project. Old people always say: “Old schemes are much simpler.” Of course, they are simpler, but no longer relevant. Maybe one day I will be the one who says: “No, all these new things are no good,” but I hope that this will not happen.
There is always a conflict between the “new” and the “old”. I think that in general, most Linux projects are stable enough to develop further, even if their main developers leave. Take the core: if at some point Linus decides that all that interests him now is to dive with a mask and look at the fish, almost no one will notice, and nothing will happen to the core.
In the end, there is something beautiful in the very idea of “open source software”: the responsibility here lies on the shoulders of so many people that if someone leaves, someone else will certainly take his place and give his shoulder. But if this did not happen, then this most likely means that the project was not really as stable, popular and in demand as everyone thought. I think that this is a great advantage of open source software - there is no one company behind it that owns everything, and when a company or a person for some reason leaves the project, all the good will remain with the project. It makes you believe that standing projects will survive anyway.
- Should we, as a community around Linux, consciously make efforts to attract new people to open source projects?
- Undoubtedly. If you support such a big product as ours, the most difficult thing is to attract developers and motivate them to stay. Honestly, I do not know how to do it. Several times we succeeded. There are so many people who send us patches, we review their code, “merzhim”. But to take such developers and make them become permanent participants in the project, accompany the project, and eventually become the ones who review other developers' code - it is really very difficult. I repeat, I do not really understand how to do this.
I think we need to make the community so friendly and interesting that people stay and are not only focused on writing code, but also interested in reviewing the code - and this is something that most people don’t even think about. Reviewing a code is at least as important as writing the same code, because for each code fragment there should be a reviewer able to analyze this fragment and merge it with the rest. Most companies tend to care only about writing code. It seems to me that in the minds of IT managers, this concept is simply absent - that someone also needs to do a review of what has been written. Maybe this is my preconceived opinion, because I spend a lot of time on the review and it seems to me that this work remains undervalued. But I seriously think that there is a significant complexity: we invite young people to contribute creatively to our projects, and ultimately we would like them to be the ones who just read someone else's code. This is a difficult task.
I have no idea how to build a truly friendly community. For example, I believe that the Linux kernel community has a reputation not as the friendliest community, but on the other hand, they are very successful and they manage to do something new. There is obviously some contradiction.
We need developers in systemd - we have a lot of them, but it would be great if there were even more! We are trying to create a much friendlier community. For example, that which takes place in the mailing lists of the kernel is not allowed here, such as particularly bright epithets and the transition to personalities.
When you start working with open source software, the attitude towards you is pretty cool at first. Therefore, I do not know how to make people stay in the project. Probably just try to be benevolent and publicly recognize the importance of all the fixes we receive. Somehow it becomes interesting to some, and they remain. If I knew how to make the developers and their attendants grow on trees - oh, then our community would have a completely different level of magical power.
- Lennart, tell me, please, what is the reason or reasons for such success systemd? Everyone knows how much controversy and controversy it caused.
- I believe that the main reason why we managed to break through, while others did not have is that people who did systemd, including me, are able to overcome strong external pressure. This does not mean that outside pressure did not affect us at all, we just found ways to cope with it. I very quickly learned to move away from all sorts of attacks. When I received messages from the Internet from some regular “clever man” that I was stupid, I proceeded from the fact that if this person did not deserve my respect as a specialist, I just ignore these words.
Honestly, I believe that if you are lucky to be born “thick-skinned”, then you have a bright future in open source. If you're not, then you have a problem. I think this is the main reason. And I am absolutely proud of myself that we managed to get through, while others did not.
Another reason is probably the documentation. Documentation is never written well enough. But people say that systemd is not bad. Everything, or almost everything that we have done, we have documented from the very beginning ... I meet a lot of technically advanced projects, where documentation is obviously not enough. Therefore, it is sometimes difficult to even discuss their brainchild with the developers.
And here we come to the conferences and contacts of developers among themselves. I want to say that when we come to the conference, we would like to receive feedback and understand what people really think. At the same time, we can show people that it is important for us to hear their opinions. Developers need to hear and understand the answer to what question people are actually looking for. Some questions make you think: “Okay, this is a different point of view on what I do. What does it mean for my work, and can I decide this use case in order to make the product more useful in the general case? ”Of course, we do not fulfill all the wishes, but only those that we consider to be the key ones, we try to set priorities correctly.
- Speaking of newbies, can you give them any recommendations? Where to start to enter the world of open source?
“When I got into open-source, the thing that hooked me was the GNOME conference in Barcelona.” I did not know anyone in the industry and did not understand that open-source developers are good guys. So it’s not the technical part that hooked me, but how great it is to go to conferences and communicate with different people. I would never have thought that the human factor and such desires as “going to a conference” can make you understand that hacking is cool. For example, I would recommend beginners to attend a FOSDEM conference or something similar. This conference is held in Brussels, it is annual and free, and this is a great opportunity to start. In fact, it is interesting for everyone, regardless of what you do - you should still visit FOSDEM. Just walk around and see what is there.
Learn to program. In general, it is not bad to be able to do something with a computer, but this is not a mandatory entrance ticket in open source. Open source is more than hacking: it is documentation, artistic creation, and much more. This will be my answer - go to conferences and try to become “yours” at them. Especially since open source projects are open for study. Yes, sometimes the documentation is not perfect, and the code is clearer than the documentation. Sometimes it is difficult to read the code, especially if you are not used to it, because it makes you think like another person thought, and you have to “put your head in the head of another person”.
- Very interesting point of view. This is an interesting and important part in hacking - not computers and code, but people and the connection of your mind with the mind of another engineer.
- Yes, everyone thinks differently, and will think differently. Give ten people the same problem, which they can translate into code, and get fifty different answers “how it is actually done”. And I am sure that thirteen of these options will be unexpected for you. Active code reading is very evolving. For me, and for others who are developing free software or want to get into this area, this is a great opportunity for self-development.
- And finally, what else would you like to say to our young readers who think - should they go to programmers and be involved in open source projects?
- Well, first is the money. Participating in open source projects is not only an exciting job - at least if you like hacking - but also a good income.
- More freedom. In most open source projects, at least, as far as I know, there is much more freedom than in ordinary software companies.
- International community. For example, in this way I was invited to a conference in St. Petersburg. I love to travel, communicate with different people from different countries.
I compare myself with my classmates - here I definitely won. We are international thanks to the Internet, for example in the systemd a few people are Germans, but all the others are not from Germany: Americans, of course, are Russians, guys from India, and from all over the world. Very few spheres where there is such an international. Where else would you meet such different cultures and meet different people.
- Lennart, we had a great conversation. Thank you!
- Yeah, I love my job! Thank!
Video recording of an interview with Lennart Pottering (English):
Video recordings of reports of 3 past Linux Piter conferences , as well as interviews with speakers of the Linux Piter conference as part of the #DevTalkRu project, are available on the conference's YouTube channel .
Lennart Pottering this year will again attend the Linux Piter Conference ( LinuxPiter.com ), which will be held on November 2-3, 2018.
Like last year, all English-language presentations will be accompanied by simultaneous translation into Russian. By the way, and vice versa too.
As a bonus, participants are given the right to attend all the events of the friendly conference, PiterPy , which is held at the same time on the same site as theLinux Piter .
For those who do not have the opportunity to attend the conference in person - there is an opportunity to buy access to online video broadcasting.
See you on Linux Piter 2018 !
An interview with Lennart Pottering is taken by Vasily Tolstoy, member of the Linux Piter Program Program Committee , DELL EMC Lead Programmer, and an active member of the St. Petersburg Linux community .
- Lennart, today you gave a talk at the Linux Piter conference, tell very briefly what your talk was about.
- Perhaps, “containers” is one of the most fashionable words in the Linux world. There are different points of view on what it is, but there are several concepts that are interesting not only for specific modern container technologies. The main idea of my report is to take some of the ideas underlying modern containers, for example, isolating processes and assembling resources into packages, and use them in the classic management of operating system services. After all, until now, classical control systems are used much more widely than containers. In general, the report was about this; In addition, I went into detail in it, how exactly we can isolate processes, which resource packages can be used, about systemd, and how does this relate to how docker and other container systems support this functionality.
Report of Lennart Pottering on Linux Piter 2017 (original version in English):
The same report with simultaneous translation into Russian
- A question for thought. What do you think is interesting in the Linux world over the past year?
- In general, I think LINUX is now so stable a product that it no longer makes major leaps forward, but the existing mechanisms are only improving. Specifically in LINUX nothing radical happened. For example, docker and similar things that have noticeably changed everything that is generally done in IT, they are already about three years old. I like the fact that today there is a constant and steady improvement in Linux. In any case, the basic components that make up the ecosystem, clearly stabilized, and gradually moving for the better. This, in my opinion, is a sign of maturity and professionalism, and this is probably a good thing.
- And what technologies are now on the wave of popularity, if there are any?
- I think it is automation, artificial intelligence. I always liked how different buzzwords in the industry make everyone “electrify” for a while and then disappear. For example, “containers” is actually a magic word already three years ago, but it is still popular. And then IOT became such, and so far no one really knows what it really is and how it differs from ordinary embedded systems. I think this year the trend is artificial intelligence. Even the Linux Foundation is now creating an artificial intelligence project. I think that this is the very “wave” that large companies now want to catch. However, I am not very closely following this topic, although I am watching from the sidelines.
- The Linux community and its key figures are getting older. What are your thoughts on this? Does it have consequences?
- Of course it does. In general, I believe that when Linux appeared in the 90s, it certainly was a revolution in many areas at once. Now this stage is completed, and each project is developing in a very specific direction. And the developers themselves, as a rule, are looking for stability in life in order to calmly manage their project. I do not know whether this is good or not. But, I think, a bad sign is when people get stuck in that way of thinking, which they got used to back in the 90s. I look at it and I wonder how I will think in 10 years? Will my head and thoughts get stuck in the past 10 years ago? Or do I really manage to keep up with the times and in time to understand that the things I knew before are no longer related to the modern IT world? I can see that some Linux developers are better at this than others.
For example, when we were doing systemd, there was a lot of noise. We removed a lot of old concepts that were fixed in Unix on a really religious level. Many of the old stuff in Unix wasn't well thought out from the start, we needed something more modern and dynamic.
We fought a great war, even at the simplest level of control of Linux devices. Now you connect some equipment, something happens, and it becomes available. UNIX was not designed that way. In the best case, it was necessary to turn off the computer, plug the wires, plug in something, run it, and then something would change in the system, but certainly there was no hot plug at all.
It was a big step forward and a real fight - to push this approach through all layers of the system and give life to this project. Old people always say: “Old schemes are much simpler.” Of course, they are simpler, but no longer relevant. Maybe one day I will be the one who says: “No, all these new things are no good,” but I hope that this will not happen.
There is always a conflict between the “new” and the “old”. I think that in general, most Linux projects are stable enough to develop further, even if their main developers leave. Take the core: if at some point Linus decides that all that interests him now is to dive with a mask and look at the fish, almost no one will notice, and nothing will happen to the core.
In the end, there is something beautiful in the very idea of “open source software”: the responsibility here lies on the shoulders of so many people that if someone leaves, someone else will certainly take his place and give his shoulder. But if this did not happen, then this most likely means that the project was not really as stable, popular and in demand as everyone thought. I think that this is a great advantage of open source software - there is no one company behind it that owns everything, and when a company or a person for some reason leaves the project, all the good will remain with the project. It makes you believe that standing projects will survive anyway.
- Should we, as a community around Linux, consciously make efforts to attract new people to open source projects?
- Undoubtedly. If you support such a big product as ours, the most difficult thing is to attract developers and motivate them to stay. Honestly, I do not know how to do it. Several times we succeeded. There are so many people who send us patches, we review their code, “merzhim”. But to take such developers and make them become permanent participants in the project, accompany the project, and eventually become the ones who review other developers' code - it is really very difficult. I repeat, I do not really understand how to do this.
I think we need to make the community so friendly and interesting that people stay and are not only focused on writing code, but also interested in reviewing the code - and this is something that most people don’t even think about. Reviewing a code is at least as important as writing the same code, because for each code fragment there should be a reviewer able to analyze this fragment and merge it with the rest. Most companies tend to care only about writing code. It seems to me that in the minds of IT managers, this concept is simply absent - that someone also needs to do a review of what has been written. Maybe this is my preconceived opinion, because I spend a lot of time on the review and it seems to me that this work remains undervalued. But I seriously think that there is a significant complexity: we invite young people to contribute creatively to our projects, and ultimately we would like them to be the ones who just read someone else's code. This is a difficult task.
I have no idea how to build a truly friendly community. For example, I believe that the Linux kernel community has a reputation not as the friendliest community, but on the other hand, they are very successful and they manage to do something new. There is obviously some contradiction.
We need developers in systemd - we have a lot of them, but it would be great if there were even more! We are trying to create a much friendlier community. For example, that which takes place in the mailing lists of the kernel is not allowed here, such as particularly bright epithets and the transition to personalities.
When you start working with open source software, the attitude towards you is pretty cool at first. Therefore, I do not know how to make people stay in the project. Probably just try to be benevolent and publicly recognize the importance of all the fixes we receive. Somehow it becomes interesting to some, and they remain. If I knew how to make the developers and their attendants grow on trees - oh, then our community would have a completely different level of magical power.
- Lennart, tell me, please, what is the reason or reasons for such success systemd? Everyone knows how much controversy and controversy it caused.
- I believe that the main reason why we managed to break through, while others did not have is that people who did systemd, including me, are able to overcome strong external pressure. This does not mean that outside pressure did not affect us at all, we just found ways to cope with it. I very quickly learned to move away from all sorts of attacks. When I received messages from the Internet from some regular “clever man” that I was stupid, I proceeded from the fact that if this person did not deserve my respect as a specialist, I just ignore these words.
Honestly, I believe that if you are lucky to be born “thick-skinned”, then you have a bright future in open source. If you're not, then you have a problem. I think this is the main reason. And I am absolutely proud of myself that we managed to get through, while others did not.
Another reason is probably the documentation. Documentation is never written well enough. But people say that systemd is not bad. Everything, or almost everything that we have done, we have documented from the very beginning ... I meet a lot of technically advanced projects, where documentation is obviously not enough. Therefore, it is sometimes difficult to even discuss their brainchild with the developers.
And here we come to the conferences and contacts of developers among themselves. I want to say that when we come to the conference, we would like to receive feedback and understand what people really think. At the same time, we can show people that it is important for us to hear their opinions. Developers need to hear and understand the answer to what question people are actually looking for. Some questions make you think: “Okay, this is a different point of view on what I do. What does it mean for my work, and can I decide this use case in order to make the product more useful in the general case? ”Of course, we do not fulfill all the wishes, but only those that we consider to be the key ones, we try to set priorities correctly.
- Speaking of newbies, can you give them any recommendations? Where to start to enter the world of open source?
“When I got into open-source, the thing that hooked me was the GNOME conference in Barcelona.” I did not know anyone in the industry and did not understand that open-source developers are good guys. So it’s not the technical part that hooked me, but how great it is to go to conferences and communicate with different people. I would never have thought that the human factor and such desires as “going to a conference” can make you understand that hacking is cool. For example, I would recommend beginners to attend a FOSDEM conference or something similar. This conference is held in Brussels, it is annual and free, and this is a great opportunity to start. In fact, it is interesting for everyone, regardless of what you do - you should still visit FOSDEM. Just walk around and see what is there.
Learn to program. In general, it is not bad to be able to do something with a computer, but this is not a mandatory entrance ticket in open source. Open source is more than hacking: it is documentation, artistic creation, and much more. This will be my answer - go to conferences and try to become “yours” at them. Especially since open source projects are open for study. Yes, sometimes the documentation is not perfect, and the code is clearer than the documentation. Sometimes it is difficult to read the code, especially if you are not used to it, because it makes you think like another person thought, and you have to “put your head in the head of another person”.
- Very interesting point of view. This is an interesting and important part in hacking - not computers and code, but people and the connection of your mind with the mind of another engineer.
- Yes, everyone thinks differently, and will think differently. Give ten people the same problem, which they can translate into code, and get fifty different answers “how it is actually done”. And I am sure that thirteen of these options will be unexpected for you. Active code reading is very evolving. For me, and for others who are developing free software or want to get into this area, this is a great opportunity for self-development.
- And finally, what else would you like to say to our young readers who think - should they go to programmers and be involved in open source projects?
- Well, first is the money. Participating in open source projects is not only an exciting job - at least if you like hacking - but also a good income.
- More freedom. In most open source projects, at least, as far as I know, there is much more freedom than in ordinary software companies.
- International community. For example, in this way I was invited to a conference in St. Petersburg. I love to travel, communicate with different people from different countries.
I compare myself with my classmates - here I definitely won. We are international thanks to the Internet, for example in the systemd a few people are Germans, but all the others are not from Germany: Americans, of course, are Russians, guys from India, and from all over the world. Very few spheres where there is such an international. Where else would you meet such different cultures and meet different people.
- Lennart, we had a great conversation. Thank you!
- Yeah, I love my job! Thank!
Video recording of an interview with Lennart Pottering (English):
Video recordings of reports of 3 past Linux Piter conferences , as well as interviews with speakers of the Linux Piter conference as part of the #DevTalkRu project, are available on the conference's YouTube channel .
Lennart Pottering this year will again attend the Linux Piter Conference ( LinuxPiter.com ), which will be held on November 2-3, 2018.
Like last year, all English-language presentations will be accompanied by simultaneous translation into Russian. By the way, and vice versa too.
As a bonus, participants are given the right to attend all the events of the friendly conference, PiterPy , which is held at the same time on the same site as theLinux Piter .
For those who do not have the opportunity to attend the conference in person - there is an opportunity to buy access to online video broadcasting.
See you on Linux Piter 2018 !