From josh at rivels.org Sun Aug 2 23:26:51 2015 From: josh at rivels.org (Josh R) Date: Sun, 2 Aug 2015 23:26:51 -0400 Subject: [talk] Free hardware Message-ID: Found this in the basement while preparing to move, let me know if you're interested, pickup in Midtown East. Netopia T1 Router, model #R5331-U (With ISDN backup) There's a post it note on it that says "admin/admin" so I'm guessing you should be able to login to it. Comes with the power supply as well. Thanks! Josh From george at ceetonetechnology.com Mon Aug 3 11:38:32 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Mon, 3 Aug 2015 11:38:32 -0400 Subject: [talk] NYC*BUG Wed: What's New with OpenBSD Message-ID: <55BF8AF8.3070307@ceetonetechnology.com> Wednesday, August 5 645 PM Stone Creek Bar & Lounge: 140 E 27th St What's new with OpenBSD Another year, another two releases for OpenBSD. Even for the best of us, it can be difficult to keep track of all the development activity. This talk highlights some of the big new things over the last year of OpenBSD. Hopefully by the end of the talk you will have learned about some new feature you didn't know about before. Bio Brian is a Ph.D. student in the Science and Technology Studies department at Rensselaer Polytechnic Institute. He has been an OpenBSD developer for a few years, spending most of his time in the ports tree. One time he gave a talk at BSDCan with George. Ike was in attendance. ***** The second vBSDCon is coming up September 11-13 in Reston, Virginia. The topics have been announced: * Blacklist?d: A NetBSD project by Christos Zoulas * What is EdgeBSD by Pierre Pronchery * HardenedBSD Internals by Shawn Webb * Devio.us, the Free OpenBSD Shell Provider and Online BSD User Group: Technical and Social Lessons Learned from Half a Decade of Service by Brian Callahan and Bryce Chidester * Interesting Things You Didn?t Know You Could Do With ZFS by Allan Jude * FreeBSD Virtualization Options by Michael Dexter * Made to Measure: Network Performance Analysis in FreeBSD by George Neville-Neil and JimThompson * getdns, A New Stub Resolver by Willem Toorop * Improving MemGuard support for UMA on FreeBSD by Chang-Hsien Tsai From raulcuza at gmail.com Tue Aug 4 18:19:34 2015 From: raulcuza at gmail.com (Raul Cuza) Date: Tue, 4 Aug 2015 18:19:34 -0400 Subject: [talk] Syslog Eats Rsyslog Message-ID: Hola, I've been researching this too long and not getting headway. I'm hoping this is a "doh!" question. Unlike RFC 3195, my reading of RFC 5424 indicates that the 1024 message size is no longer in place. But when I try to tell rsyslog (v7.4.4) this I still get my long messages broken up into 1k chunks. I want to send jumbo log entries (i.e. ~4k) over the wire to a logstash server that will munch it into JSON and throw it up into elasticsearch. Am I trying to do the impossible with rsyslog? I can't run logstash on the device that is generating the logs because it is extremely resource limited. Thanks for any help you can provide. Ra?l From jschauma at netmeister.org Tue Aug 4 20:34:45 2015 From: jschauma at netmeister.org (Jan Schaumann) Date: Tue, 4 Aug 2015 20:34:45 -0400 Subject: [talk] Ethical Obligations in Internet Operations Message-ID: <20150805003445.GM18613@netmeister.org> Hello, This isn't really BSD related, but it's NY related. ;-) Also, you're all good people whose input I'd value. I'm currently preparing a talk on "Ethical Obligations in Internet Operations"[1] for Velocity NY in October (see https://www.netmeister.org/blog/velocity-ny2015-accepted.html for the proposal). In order to help me better understand our profession(s) and to prepare a better talk, I've put together a short, anonymous questionnaire for everybody involved in "Internet Operations": https://docs.google.com/forms/d/1UmJtkj1dsFVjZx2yBC6zZZi1tIuObuVZxQd3oWxGk4o/viewform If you could take the five minutes to fill out this form, that would be a great help for me. If you'd like to share this in your social or professional circles, that'd be wonderful, too. You could: - forward this email - link to the form above - link to https://www.netmeister.org/blog/velocity-ny2015-questionnaire.html for a bit more context - retweet / quote this tweet: https://twitter.com/jschauma/status/628393619196641280 Of course I also welcome comments directly to me via email, or a discussion on this mailing list, if not deemed off-topic. Many thanks in advance for your help! -Jan From bonsaime at gmail.com Tue Aug 4 22:47:32 2015 From: bonsaime at gmail.com (Jesse Callaway) Date: Tue, 4 Aug 2015 22:47:32 -0400 Subject: [talk] Syslog Eats Rsyslog In-Reply-To: References: Message-ID: The logstash "syslog input" receiver doesn't hold to either of these specifications. You'll have to set up a proper rsyslog receiver on the other end and then pipe it to a socket using the "unix input". For more info you are probably best to hit up the elasticsearch fora. On Tue, Aug 4, 2015 at 6:19 PM, Raul Cuza wrote: > Hola, > > I've been researching this too long and not getting headway. I'm > hoping this is a "doh!" question. > > Unlike RFC 3195, my reading of RFC 5424 indicates that the 1024 > message size is no longer in place. But when I try to tell rsyslog > (v7.4.4) this I still get my long messages broken up into 1k chunks. I > want to send jumbo log entries (i.e. ~4k) over the wire to a logstash > server that will munch it into JSON and throw it up into > elasticsearch. > > Am I trying to do the impossible with rsyslog? I can't run logstash on > the device that is generating the logs because it is extremely > resource limited. > > Thanks for any help you can provide. > > Ra?l > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk -- -jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From george at ceetonetechnology.com Wed Aug 5 11:01:03 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Wed, 5 Aug 2015 11:01:03 -0400 Subject: [talk] NYC*BUG Tonight: What's New with OpenBSD? Message-ID: <55C2252F.8040505@ceetonetechnology.com> August 5: What's New with OpenBSD, Brian Callahan 18:45, Stone Creek Bar & Lounge: 140 E 27th St Abstract Another year, another two releases for OpenBSD. Even for the best of us, it can be difficult to keep track of all the development activity. This talk highlights some of the big new things over the last year of OpenBSD. Hopefully by the end of the talk you will have learned about some new feature you didn't know about before. Speaker Bio Brian is a Ph.D. student in the Science and Technology Studies department at Rensselaer Polytechnic Institute. He has been an OpenBSD developer for a few years, spending most of his time in the ports tree. One time he gave a talk at BSDCan with George. Ike was in attendance. From raulcuza at gmail.com Wed Aug 5 14:21:02 2015 From: raulcuza at gmail.com (Raul Cuza) Date: Wed, 5 Aug 2015 14:21:02 -0400 Subject: [talk] Syslog Eats Rsyslog In-Reply-To: References: Message-ID: On Tue, Aug 4, 2015 at 10:47 PM, Jesse Callaway wrote: > The logstash "syslog input" receiver doesn't hold to either of these > specifications. You'll have to set up a proper rsyslog receiver on the other > end and then pipe it to a socket using the "unix input". > > For more info you are probably best to hit up the elasticsearch fora. > > > > On Tue, Aug 4, 2015 at 6:19 PM, Raul Cuza wrote: >> >> Hola, >> >> I've been researching this too long and not getting headway. I'm >> hoping this is a "doh!" question. >> >> Unlike RFC 3195, my reading of RFC 5424 indicates that the 1024 >> message size is no longer in place. But when I try to tell rsyslog >> (v7.4.4) this I still get my long messages broken up into 1k chunks. I >> want to send jumbo log entries (i.e. ~4k) over the wire to a logstash >> server that will munch it into JSON and throw it up into >> elasticsearch. >> >> Am I trying to do the impossible with rsyslog? I can't run logstash on >> the device that is generating the logs because it is extremely >> resource limited. >> >> Thanks for any help you can provide. >> >> Ra?l >> > -jesse Sorry about the OT. I exhausted my number of smart decisions for the day and just sent an email to my default $group_of_smart_people. My current guess is that `logger` is imposing the 1024 byte limit. My C skills are not strong enough for me to prove it via https://svnweb.freebsd.org/base/stable/10/usr.bin/logger/logger.c?view=markup#l89. I did prove that rsyslog does not have a 1024 byte limit. I was using `logger` in my shell script to generate syslog events. When I switched my rsyslog.conf to read from a file where my shell script wrote lines of increasing size (thanks to `curl` and the Bacon Ipsum API), I was able to add messages to `/var/log/messages` of over 5k size. Thank you, again. Ra?l From mcevoy.pat at gmail.com Thu Aug 6 13:08:44 2015 From: mcevoy.pat at gmail.com (Patrick McEvoy) Date: Thu, 06 Aug 2015 13:08:44 -0400 Subject: [talk] What's New with OpenBSD? - Did anyone grab audio? Message-ID: <55C3949C.9070208@gmail.com> Did anyone get an audio recording of Brian's talk last night? Hope it went well. P From mspitzer at gmail.com Sun Aug 9 13:44:33 2015 From: mspitzer at gmail.com (Marc Spitzer) Date: Sun, 9 Aug 2015 13:44:33 -0400 Subject: [talk] git branch visuizer Message-ID: It looked interesting and useful, https://github.com/kidpixo/git_history_visualizer What do you think? Marc -- Freedom is nothing but a chance to be better. --Albert Camus The inherent vice of capitalism is the unequal sharing of blessings; the inherent virtue of socialism is the equal sharing of miseries. -- Winston Churchill Do the arithmetic or be doomed to talk nonsense. --John McCarthy -------------- next part -------------- An HTML attachment was scrubbed... URL: From bcallah at devio.us Sun Aug 9 23:54:08 2015 From: bcallah at devio.us (Brian Callahan) Date: Sun, 9 Aug 2015 23:54:08 -0400 Subject: [talk] BSDNow 100+ episodes and T-shirt Message-ID: <55C82060.5050009@devio.us> Hi Combined New York BUG -- I'd like to point out that the crew at BSDNow recently passed 100 episodes. I think we ought to celebrate a weekly BSD news show going strong after 2 years. If you haven't seen the show, you can watch it at http://bsdnow.tv/ They're selling T-shirts to mark their two-year anniversary. I'll note the money raised is NOT going to the projects, so it's not a charity, but it is going to buy new equipment for the show. Consider this email a statement of fact and not an advertisement. ~Brian From freebsd at fongaboo.com Mon Aug 10 18:59:10 2015 From: freebsd at fongaboo.com (freebsd at fongaboo.com) Date: Mon, 10 Aug 2015 18:59:10 -0400 (EDT) Subject: [talk] [CDBUG-talk] BSDNow 100+ episodes and T-shirt In-Reply-To: <55C82060.5050009@devio.us> References: <55C82060.5050009@devio.us> Message-ID: These guys are the best! They have single-handedly done more to transform the BSD community than anyone else I know. On Sun, 9 Aug 2015, Brian Callahan wrote: > Hi Combined New York BUG -- > > I'd like to point out that the crew at BSDNow recently passed 100 > episodes. I think we ought to celebrate a weekly BSD news show going > strong after 2 years. If you haven't seen the show, you can watch it at > http://bsdnow.tv/ > > They're selling T-shirts to mark their two-year anniversary. I'll note > the money raised is NOT going to the projects, so it's not a charity, > but it is going to buy new equipment for the show. Consider this email a > statement of fact and not an advertisement. > > ~Brian > > _______________________________________________ > CDBUG-talk mailing list > CDBUG-talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/cdbug-talk > From george at ceetonetechnology.com Thu Aug 13 10:03:04 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Aug 2015 10:03:04 -0400 Subject: [talk] the next con: some thoughts (1 of 2) Message-ID: <55CCA398.20705@ceetonetechnology.com> Informally, there has been a lot of discussions about our next NYCBSDCon. Normally, this type of discussion would start on admin@, but we need to break that routine. Conferences are lot of work, and that needs to be understood. It's essentially a second (uncompensated) job to carry for a number of people, and it's exhausting. Yes, the rewards are great, and we are proud of what we've done, but that point needs to be comprehended. On that note, the last one in February 2014 was easier than in the past for a few reasons. A lot of people took on responsibility for the event. In particular, Patrick M and Mike N completely dealt with a/v to the extent that everyone else was out of the loop. Additionally, doing a day-long event focused on the NYC area, and avoiding the hassles of coordinating hotels, etc., was an enormous relief. So arrive today in August 2015. It's been a year and a half since our last con. There are a few spots I've looked at recently that could be ideal. And yesterday, I looked at a new space at 150 Broadway that just opened called LMHQ, which is one of those privately funded startup/hangout tech spaces. NYI is a founding member, and they are happy to enable us to use the space. LMHQ's main area fits up to 140 people with two screens, plus has a bunch of conference rooms. It's an ideal spot and right next to the new Fulton Street station downtown. At this point, it's worth considering something, say, at the end of February, assuming all the pieces can move into place. That includes a lot of people engaged in making it happen and taking responsibility for it as in the past. The next email will be about content... g From george at ceetonetechnology.com Thu Aug 13 10:27:45 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Aug 2015 10:27:45 -0400 Subject: [talk] the next con: content (2 of 2) Message-ID: <55CCA961.3010902@ceetonetechnology.com> February 2014's con was focused on the "BSDs in Production" and was themed broad enough to allow us to choose content while feeling consistent :) And like last con, our audience is not the BSD scene local and remote, but rather the non-BSD people in the metropolitan area. No one can take anything away from the current BSDCons' importance, but we are doing something different here, again. Our con is not an opportunity for the 'usual suspects' to meet at different cities around the world. Rather, our con is about talking to the broader community around NYC. There are two theme ideas I'm personally thinking about that have been discussed. Yes, the term "beyond" is purposeful. 1. The BSDs Beyond x86: ARM, MIPS The obvious connection for people on this topic is the Raspberry Pi, but I can imagine that will barely be mentioned. There is very significant work happening on armv7 and what is now known as aarch64 (64-bit ARM). It's not just about small hardware, but about powerful, low-energy consuming hardware that should begin creeping into data centers soon. The big firms are working on it, and even Amazon acquired an ARM hardware firm a while back. There are other angles. There is some *really* cheap hardware that is useful for testing network drivers, porting to the Chromebook, etc. Ideally, we'd get some hardware manufacturers to bring in some gear to make this a more hands-on event. 2. The BSDs and Security: Beyond the Obvious IMHO the security angle is way overplayed, and we should be angling this outside the box. There are a few topics that come to mind. OCAML being one. Capsicum/tame (fbsd/obsd, respectively). ASLR. Interesting lessons in porting Tor Browser (essentially Linux software) to OpenBSD in regards to portability, footprints (er, bloat). Upstreaming portable BSD code, specifically thinking about OBSD's arc4random and libressl (libretls now? :). Another topic might be on entropy. In light of the FBSD breakage in the fall in -current and the critiques of Linux RNG, how do we know it's working? What is good entropy? How do we know it's good? How many stupid ways do bad non-crypto developers try to replace a system's RNG? Finally, as always, we are going to work hard to keep the event as "BSD agnostic" as possible. All the BSDs should be represented, but also having more general speakers not tied to one project or another is a positive. Anyways, please feel free to jump into this topic. g From mmatalka at gmail.com Thu Aug 13 12:12:09 2015 From: mmatalka at gmail.com (Malcolm Matalka) Date: Thu, 13 Aug 2015 18:12:09 +0200 Subject: [talk] the next con: content (2 of 2) In-Reply-To: <55CCA961.3010902@ceetonetechnology.com> References: <55CCA961.3010902@ceetonetechnology.com> Message-ID: Den 13 aug 2015 16:28 skrev "George Rosamond" : > > February 2014's con was focused on the "BSDs in Production" and was > themed broad enough to allow us to choose content while feeling > consistent :) > > And like last con, our audience is not the BSD scene local and remote, > but rather the non-BSD people in the metropolitan area. No one can take > anything away from the current BSDCons' importance, but we are doing > something different here, again. Our con is not an opportunity for the > 'usual suspects' to meet at different cities around the world. Rather, > our con is about talking to the broader community around NYC. > > There are two theme ideas I'm personally thinking about that have been > discussed. Yes, the term "beyond" is purposeful. > > 1. The BSDs Beyond x86: ARM, MIPS > > The obvious connection for people on this topic is the Raspberry Pi, but > I can imagine that will barely be mentioned. > > There is very significant work happening on armv7 and what is now known > as aarch64 (64-bit ARM). It's not just about small hardware, but about > powerful, low-energy consuming hardware that should begin creeping into > data centers soon. The big firms are working on it, and even Amazon > acquired an ARM hardware firm a while back. > > There are other angles. There is some *really* cheap hardware that is > useful for testing network drivers, porting to the Chromebook, etc. > > Ideally, we'd get some hardware manufacturers to bring in some gear to > make this a more hands-on event. > > 2. The BSDs and Security: Beyond the Obvious > > IMHO the security angle is way overplayed, and we should be angling this > outside the box. > > There are a few topics that come to mind. > > OCAML being one. Capsicum/tame (fbsd/obsd, respectively). ASLR. As an ocaml fan, what are you referring to when you say ocaml here? > Interesting lessons in porting Tor Browser (essentially Linux software) > to OpenBSD in regards to portability, footprints (er, bloat). > Upstreaming portable BSD code, specifically thinking about OBSD's > arc4random and libressl (libretls now? :). > > Another topic might be on entropy. In light of the FBSD breakage in the > fall in -current and the critiques of Linux RNG, how do we know it's > working? What is good entropy? How do we know it's good? How many > stupid ways do bad non-crypto developers try to replace a system's RNG? > > Finally, as always, we are going to work hard to keep the event as "BSD > agnostic" as possible. All the BSDs should be represented, but also > having more general speakers not tied to one project or another is a > positive. > > Anyways, please feel free to jump into this topic. > > g > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk -------------- next part -------------- An HTML attachment was scrubbed... URL: From george at ceetonetechnology.com Thu Aug 13 12:26:43 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Aug 2015 12:26:43 -0400 Subject: [talk] the next con: content (2 of 2) In-Reply-To: References: <55CCA961.3010902@ceetonetechnology.com> Message-ID: <55CCC543.5000104@ceetonetechnology.com> Malcolm Matalka: > Den 13 aug 2015 16:28 skrev "George Rosamond" > : >> >> February 2014's con was focused on the "BSDs in Production" and was >> themed broad enough to allow us to choose content while feeling >> consistent :) >> >> And like last con, our audience is not the BSD scene local and remote, >> but rather the non-BSD people in the metropolitan area. No one can take >> anything away from the current BSDCons' importance, but we are doing >> something different here, again. Our con is not an opportunity for the >> 'usual suspects' to meet at different cities around the world. Rather, >> our con is about talking to the broader community around NYC. >> >> There are two theme ideas I'm personally thinking about that have been >> discussed. Yes, the term "beyond" is purposeful. >> >> 1. The BSDs Beyond x86: ARM, MIPS >> >> The obvious connection for people on this topic is the Raspberry Pi, but >> I can imagine that will barely be mentioned. >> >> There is very significant work happening on armv7 and what is now known >> as aarch64 (64-bit ARM). It's not just about small hardware, but about >> powerful, low-energy consuming hardware that should begin creeping into >> data centers soon. The big firms are working on it, and even Amazon >> acquired an ARM hardware firm a while back. >> >> There are other angles. There is some *really* cheap hardware that is >> useful for testing network drivers, porting to the Chromebook, etc. >> >> Ideally, we'd get some hardware manufacturers to bring in some gear to >> make this a more hands-on event. >> >> 2. The BSDs and Security: Beyond the Obvious >> >> IMHO the security angle is way overplayed, and we should be angling this >> outside the box. >> >> There are a few topics that come to mind. >> >> OCAML being one. Capsicum/tame (fbsd/obsd, respectively). ASLR. > > As an ocaml fan, what are you referring to when you say ocaml here? > The language in the context of building secure services is one idea. And what would your idea be? >> Interesting lessons in porting Tor Browser (essentially Linux software) >> to OpenBSD in regards to portability, footprints (er, bloat). >> Upstreaming portable BSD code, specifically thinking about OBSD's >> arc4random and libressl (libretls now? :). >> >> Another topic might be on entropy. In light of the FBSD breakage in the >> fall in -current and the critiques of Linux RNG, how do we know it's >> working? What is good entropy? How do we know it's good? How many >> stupid ways do bad non-crypto developers try to replace a system's RNG? >> >> Finally, as always, we are going to work hard to keep the event as "BSD >> agnostic" as possible. All the BSDs should be represented, but also >> having more general speakers not tied to one project or another is a >> positive. >> >> Anyways, please feel free to jump into this topic. g From mmatalka at gmail.com Thu Aug 13 12:57:04 2015 From: mmatalka at gmail.com (Malcolm Matalka) Date: Thu, 13 Aug 2015 16:57:04 +0000 Subject: [talk] the next con: content (2 of 2) In-Reply-To: <55CCC543.5000104@ceetonetechnology.com> (George Rosamond's message of "Thu, 13 Aug 2015 12:26:43 -0400") References: <55CCA961.3010902@ceetonetechnology.com> <55CCC543.5000104@ceetonetechnology.com> Message-ID: <86egj7kty7.fsf@gmail.com> George Rosamond writes: > Malcolm Matalka: >> Den 13 aug 2015 16:28 skrev "George Rosamond" >> : >>> >>> February 2014's con was focused on the "BSDs in Production" and was >>> themed broad enough to allow us to choose content while feeling >>> consistent :) >>> >>> And like last con, our audience is not the BSD scene local and remote, >>> but rather the non-BSD people in the metropolitan area. No one can take >>> anything away from the current BSDCons' importance, but we are doing >>> something different here, again. Our con is not an opportunity for the >>> 'usual suspects' to meet at different cities around the world. Rather, >>> our con is about talking to the broader community around NYC. >>> >>> There are two theme ideas I'm personally thinking about that have been >>> discussed. Yes, the term "beyond" is purposeful. >>> >>> 1. The BSDs Beyond x86: ARM, MIPS >>> >>> The obvious connection for people on this topic is the Raspberry Pi, but >>> I can imagine that will barely be mentioned. >>> >>> There is very significant work happening on armv7 and what is now known >>> as aarch64 (64-bit ARM). It's not just about small hardware, but about >>> powerful, low-energy consuming hardware that should begin creeping into >>> data centers soon. The big firms are working on it, and even Amazon >>> acquired an ARM hardware firm a while back. >>> >>> There are other angles. There is some *really* cheap hardware that is >>> useful for testing network drivers, porting to the Chromebook, etc. >>> >>> Ideally, we'd get some hardware manufacturers to bring in some gear to >>> make this a more hands-on event. >>> >>> 2. The BSDs and Security: Beyond the Obvious >>> >>> IMHO the security angle is way overplayed, and we should be angling this >>> outside the box. >>> >>> There are a few topics that come to mind. >>> >>> OCAML being one. Capsicum/tame (fbsd/obsd, respectively). ASLR. >> >> As an ocaml fan, what are you referring to when you say ocaml here? >> > > The language in the context of building secure services is one idea. > > And what would your idea be? I don't have any idea per se, more of just curious to see it explicitly mentioned. I actually think Ocaml is great for low-levelish-user-land things and spend most of my time developing tools and services in Ocaml so I'd be interested to see it become more prominent in the BSD community. > >>> Interesting lessons in porting Tor Browser (essentially Linux software) >>> to OpenBSD in regards to portability, footprints (er, bloat). >>> Upstreaming portable BSD code, specifically thinking about OBSD's >>> arc4random and libressl (libretls now? :). >>> >>> Another topic might be on entropy. In light of the FBSD breakage in the >>> fall in -current and the critiques of Linux RNG, how do we know it's >>> working? What is good entropy? How do we know it's good? How many >>> stupid ways do bad non-crypto developers try to replace a system's RNG? >>> >>> Finally, as always, we are going to work hard to keep the event as "BSD >>> agnostic" as possible. All the BSDs should be represented, but also >>> having more general speakers not tied to one project or another is a >>> positive. >>> >>> Anyways, please feel free to jump into this topic. > > g From george at ceetonetechnology.com Thu Aug 13 13:05:11 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Aug 2015 13:05:11 -0400 Subject: [talk] the next con: content (2 of 2) In-Reply-To: <86egj7kty7.fsf@gmail.com> References: <55CCA961.3010902@ceetonetechnology.com> <55CCC543.5000104@ceetonetechnology.com> <86egj7kty7.fsf@gmail.com> Message-ID: <55CCCE47.2030806@ceetonetechnology.com> Malcolm Matalka: > George Rosamond writes: > >> Malcolm Matalka: >>> Den 13 aug 2015 16:28 skrev "George Rosamond" >>> : >>>> >>>> February 2014's con was focused on the "BSDs in Production" and was >>>> themed broad enough to allow us to choose content while feeling >>>> consistent :) >>>> >>>> And like last con, our audience is not the BSD scene local and remote, >>>> but rather the non-BSD people in the metropolitan area. No one can take >>>> anything away from the current BSDCons' importance, but we are doing >>>> something different here, again. Our con is not an opportunity for the >>>> 'usual suspects' to meet at different cities around the world. Rather, >>>> our con is about talking to the broader community around NYC. >>>> >>>> There are two theme ideas I'm personally thinking about that have been >>>> discussed. Yes, the term "beyond" is purposeful. >>>> >>>> 1. The BSDs Beyond x86: ARM, MIPS >>>> >>>> The obvious connection for people on this topic is the Raspberry Pi, but >>>> I can imagine that will barely be mentioned. >>>> >>>> There is very significant work happening on armv7 and what is now known >>>> as aarch64 (64-bit ARM). It's not just about small hardware, but about >>>> powerful, low-energy consuming hardware that should begin creeping into >>>> data centers soon. The big firms are working on it, and even Amazon >>>> acquired an ARM hardware firm a while back. >>>> >>>> There are other angles. There is some *really* cheap hardware that is >>>> useful for testing network drivers, porting to the Chromebook, etc. >>>> >>>> Ideally, we'd get some hardware manufacturers to bring in some gear to >>>> make this a more hands-on event. >>>> >>>> 2. The BSDs and Security: Beyond the Obvious >>>> >>>> IMHO the security angle is way overplayed, and we should be angling this >>>> outside the box. >>>> >>>> There are a few topics that come to mind. >>>> >>>> OCAML being one. Capsicum/tame (fbsd/obsd, respectively). ASLR. >>> >>> As an ocaml fan, what are you referring to when you say ocaml here? >>> >> >> The language in the context of building secure services is one idea. >> >> And what would your idea be? > > I don't have any idea per se, more of just curious to see it explicitly > mentioned. I actually think Ocaml is great for low-levelish-user-land > things and spend most of my time developing tools and services in Ocaml > so I'd be interested to see it become more prominent in the BSD > community. > I don't know a whole lot about ocaml... but I do know that one of prominent devs is/was an OpenBSD dev, and a remote NYC*BUG person who also used to be an OpenBSD dev uses it daily. It was actually my conversation with him that sparked the thought. But I do think your last line is a decent start... "Building OCAML Prominence in BSD Land", or "OCAML: What is it and Why you Need to Pay Attention" g >> >>>> Interesting lessons in porting Tor Browser (essentially Linux software) >>>> to OpenBSD in regards to portability, footprints (er, bloat). >>>> Upstreaming portable BSD code, specifically thinking about OBSD's >>>> arc4random and libressl (libretls now? :). >>>> >>>> Another topic might be on entropy. In light of the FBSD breakage in the >>>> fall in -current and the critiques of Linux RNG, how do we know it's >>>> working? What is good entropy? How do we know it's good? How many >>>> stupid ways do bad non-crypto developers try to replace a system's RNG? >>>> >>>> Finally, as always, we are going to work hard to keep the event as "BSD >>>> agnostic" as possible. All the BSDs should be represented, but also >>>> having more general speakers not tied to one project or another is a >>>> positive. >>>> >>>> Anyways, please feel free to jump into this topic. >> >> g From jkeen at verizon.net Thu Aug 13 12:13:47 2015 From: jkeen at verizon.net (James E Keenan) Date: Thu, 13 Aug 2015 12:13:47 -0400 Subject: [talk] the next con: content (2 of 2) In-Reply-To: <55CCA961.3010902@ceetonetechnology.com> References: <55CCA961.3010902@ceetonetechnology.com> Message-ID: <55CCC23B.6090508@verizon.net> On 08/13/2015 10:27 AM, George Rosamond wrote:> > 2. The BSDs and Security: Beyond the Obvious > > IMHO the security angle is way overplayed, and we should be angling this > outside the box. > Can you elaborate on what you mean by "overplayed" and "outside the box"? (I don't know enough about security issues to guess what you're referring to.) Thank you very much. Jim Keenan From attila at stalphonsos.com Thu Aug 13 15:20:12 2015 From: attila at stalphonsos.com (attila) Date: Thu, 13 Aug 2015 14:20:12 -0500 Subject: [talk] Anyone have a USB TRNG they want to work under OpenBSD? Message-ID: <87y4hfq9lf.fsf@rotfl.l.stalphonsos.net> Back in April I wrote an OpenBSD driver for the Alea II USB TRNG: http://www.araneus.fi/products/alea2/en/ It was committed to the tree 16 April: http://marc.info/?l=openbsd-cvs&m=142917454025402&w=2 As of now if you plug one of those into an OpenBSD box running a reasonably recent version of OpenBSD the system entropy pool will be seeded by the TRNG. I would like to continue to do this, so that eventually you can plug any cheap(?), reasonably available USB TRNG into an OpenBSD box (say, a busy Tor relay) and have Good Things Happen. Does anyone out there in NYC*BUG land have a TrueRNG, OneRNG or other TRNG that you can actually procure without too much hassle who would be willing to help me do this? I did the Alea II first because the developer/vendor is a friend of mine and he sent me one for free. I live in a place where it's difficult to get this kind of thing, but if you want to send it to me I promise to send it back. OTOH, I was thinking I could also do it remotely if someone with a device was willing to collaborate... Pax, -A -- http://trac.haqistan.net | attila at stalphonsos.com | 0xE6CC1EDB From pete at nomadlogic.org Thu Aug 13 15:55:40 2015 From: pete at nomadlogic.org (Pete Wright) Date: Thu, 13 Aug 2015 12:55:40 -0700 Subject: [talk] the next con: content (2 of 2) In-Reply-To: <55CCA961.3010902@ceetonetechnology.com> References: <55CCA961.3010902@ceetonetechnology.com> Message-ID: <55CCF63C.8000207@nomadlogic.org> On 08/13/15 07:27, George Rosamond wrote: > February 2014's con was focused on the "BSDs in Production" and was > themed broad enough to allow us to choose content while feeling > consistent :) > > And like last con, our audience is not the BSD scene local and remote, > but rather the non-BSD people in the metropolitan area. No one can take > anything away from the current BSDCons' importance, but we are doing > something different here, again. Our con is not an opportunity for the > 'usual suspects' to meet at different cities around the world. Rather, > our con is about talking to the broader community around NYC. > > > Finally, as always, we are going to work hard to keep the event as "BSD > agnostic" as possible. All the BSDs should be represented, but also > having more general speakers not tied to one project or another is a > positive. > > Anyways, please feel free to jump into this topic. > i think having a discussion about visualization in general may be a good topic. there is ton's of exciting *BSD specific and non-*BSD activity going on in this area currently. of the top of my head: 1) bhyve 2) netbsd and xen still doing tons of interesting stuff (see recent articles about netbsd rump kernels triggering peoples interest in using it on VM stacks) 3) openstack (self hosted AWS style clouds) 4) docker, or how i pretended to invent chroot and jails without any of the security benefits and now all i got is this hipster mustache and cool logo and i'm not really bitter... -- in all seriousness - docker can't really be ignored and there is some dev work going into supporting freebsd in the docker ecosystem for better or worse. i'm sure there is much more on this topic as well that can be added.... gman - is the idea to have one theme rule the day, or try to get enough speakers to have two tracks? -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From george at ceetonetechnology.com Thu Aug 13 18:21:52 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Aug 2015 18:21:52 -0400 Subject: [talk] the next con: content (2 of 2) In-Reply-To: <55CCC23B.6090508@verizon.net> References: <55CCA961.3010902@ceetonetechnology.com> <55CCC23B.6090508@verizon.net> Message-ID: <55CD1880.5020104@ceetonetechnology.com> James E Keenan: > On 08/13/2015 10:27 AM, George Rosamond wrote:> >> 2. The BSDs and Security: Beyond the Obvious >> >> IMHO the security angle is way overplayed, and we should be angling this >> outside the box. >> > > Can you elaborate on what you mean by "overplayed" and "outside the > box"? (I don't know enough about security issues to guess what you're > referring to.) Valid question. We all know it's a buzzword, and instead of doing the standard sec conference topics, we figure out how to show the legitimacy of BSD code in the scene. I hate to just pick on the OBSD stuff, but something like LibreSSL, as portable code, can become an option for a port that requires OpenSSL. And a portable version means other OSs can benefit. Same with arc4random. When you need good clean and cheap entropy in an application, devs who are not cryptographers do stupid things like write RNGs in, say, Java. arc4random can be that piece that enables those devs to not just bypass playing cryptographers on TV, but also deal with crappy /dev/urandom clunkiness and ugliness. Make sense? g From george at ceetonetechnology.com Thu Aug 13 18:29:41 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Aug 2015 18:29:41 -0400 Subject: [talk] the next con: content (2 of 2) In-Reply-To: <55CCF63C.8000207@nomadlogic.org> References: <55CCA961.3010902@ceetonetechnology.com> <55CCF63C.8000207@nomadlogic.org> Message-ID: <55CD1A55.9030100@ceetonetechnology.com> Pete Wright: > > > On 08/13/15 07:27, George Rosamond wrote: >> February 2014's con was focused on the "BSDs in Production" and was >> themed broad enough to allow us to choose content while feeling >> consistent :) >> >> And like last con, our audience is not the BSD scene local and remote, >> but rather the non-BSD people in the metropolitan area. No one can take >> anything away from the current BSDCons' importance, but we are doing >> something different here, again. Our con is not an opportunity for the >> 'usual suspects' to meet at different cities around the world. Rather, >> our con is about talking to the broader community around NYC. >> > >> >> Finally, as always, we are going to work hard to keep the event as "BSD >> agnostic" as possible. All the BSDs should be represented, but also >> having more general speakers not tied to one project or another is a >> positive. >> >> Anyways, please feel free to jump into this topic. >> > > i think having a discussion about visualization in general may be a good > topic. there is ton's of exciting *BSD specific and non-*BSD activity > going on in this area currently. of the top of my head: > Definitely... this has come up before as a con theme, and it means we talk to that broad 20-something audience that's missing what an OS even is, much less Unix. > 1) bhyve > > 2) netbsd and xen still doing tons of interesting stuff (see recent > articles about netbsd rump kernels triggering peoples interest in using > it on VM stacks) yes... we tried to get a rump talk a while ago. > > 3) openstack (self hosted AWS style clouds) > > 4) docker, or how i pretended to invent chroot and jails without any of > the security benefits and now all i got is this hipster mustache and > cool logo and i'm not really bitter... > -- in all seriousness - docker can't really be ignored and there is some > dev work going into supporting freebsd in the docker ecosystem for > better or worse. OMFG. You should write publicity materials... for like, Anonymous. > > i'm sure there is much more on this topic as well that can be added.... > > > gman - is the idea to have one theme rule the day, or try to get enough > speakers to have two tracks? We kept to one track to not only make it easier to manage, but also to force people to hear talks they wouldn't necessarily attend. And it allows us to keep one clear theme. (These are all my opinions, and not some consensus on admin@, nyc*bug, or any of our parent investors, holding companies, and whatevers. I may or may not own stock in OpenSSL or LibreSSL if it was even offered. It's all open to debate.) g From pete at nomadlogic.org Thu Aug 13 18:44:33 2015 From: pete at nomadlogic.org (Pete Wright) Date: Thu, 13 Aug 2015 15:44:33 -0700 Subject: [talk] the next con: content (2 of 2) In-Reply-To: <55CD1A55.9030100@ceetonetechnology.com> References: <55CCA961.3010902@ceetonetechnology.com> <55CCF63C.8000207@nomadlogic.org> <55CD1A55.9030100@ceetonetechnology.com> Message-ID: <55CD1DD1.50202@nomadlogic.org> On 08/13/15 15:29, George Rosamond wrote: > Pete Wright: >> >> >> On 08/13/15 07:27, George Rosamond wrote: >>> February 2014's con was focused on the "BSDs in Production" and was >>> themed broad enough to allow us to choose content while feeling >>> consistent :) >>> >>> And like last con, our audience is not the BSD scene local and remote, >>> but rather the non-BSD people in the metropolitan area. No one can take >>> anything away from the current BSDCons' importance, but we are doing >>> something different here, again. Our con is not an opportunity for the >>> 'usual suspects' to meet at different cities around the world. Rather, >>> our con is about talking to the broader community around NYC. >>> >> >>> >>> Finally, as always, we are going to work hard to keep the event as "BSD >>> agnostic" as possible. All the BSDs should be represented, but also >>> having more general speakers not tied to one project or another is a >>> positive. >>> >>> Anyways, please feel free to jump into this topic. >>> >> >> i think having a discussion about visualization in general may be a good >> topic. there is ton's of exciting *BSD specific and non-*BSD activity >> going on in this area currently. of the top of my head: >> > > Definitely... this has come up before as a con theme, and it means we > talk to that broad 20-something audience that's missing what an OS even > is, much less Unix. > >> 1) bhyve >> >> 2) netbsd and xen still doing tons of interesting stuff (see recent >> articles about netbsd rump kernels triggering peoples interest in using >> it on VM stacks) > > yes... we tried to get a rump talk a while ago. > >> >> 3) openstack (self hosted AWS style clouds) >> >> 4) docker, or how i pretended to invent chroot and jails without any of >> the security benefits and now all i got is this hipster mustache and >> cool logo and i'm not really bitter... >> -- in all seriousness - docker can't really be ignored and there is some >> dev work going into supporting freebsd in the docker ecosystem for >> better or worse. > > OMFG. You should write publicity materials... for like, Anonymous. > >> >> i'm sure there is much more on this topic as well that can be added.... >> >> >> gman - is the idea to have one theme rule the day, or try to get enough >> speakers to have two tracks? > > We kept to one track to not only make it easier to manage, but also to > force people to hear talks they wouldn't necessarily attend. And it > allows us to keep one clear theme. > > (These are all my opinions, and not some consensus on admin@, nyc*bug, > or any of our parent investors, holding companies, and whatevers. I may > or may not own stock in OpenSSL or LibreSSL if it was even offered. > It's all open to debate.) sweet - thanks for the clarification gman! like you i have a semi-vested interest in the virtualization stuff as i'm a beastie in a sea of penguins at my $JOB. having said that i am making inroads via various virtualization efforts to get wider use of non-linux systems (as both guests and hypervisors). cheers, -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From mmatalka at gmail.com Fri Aug 14 11:24:05 2015 From: mmatalka at gmail.com (Malcolm Matalka) Date: Fri, 14 Aug 2015 15:24:05 +0000 Subject: [talk] Idea: NixBSD Message-ID: <864mk1lwq2.fsf@gmail.com> Hello everyone, I'd like to throw out an idea and get peoples reactions. This is not a concrete plan or anything I'm actively working on, more of a long term vision I'd love to see happen. Idea: A distribution of BSD that leverages the Nix package manager down to the OS configuration. Roughly a BSD version of NixOS. Motivation: The effort put into the ports is fantastic but the ports system is rigid in that supporting multiple versions of a single package requires a fair amount of effort, installing/removing ports is limited to the administrator of the machine, each port has to produce a unique set of output names, source based vs binary based might have a similar source but are distinct package systems. Solution: The Nix package manager is a purely functional package manager that supports source-based, and binary-based packages. It allows multiple versions of a port to be installed simultaneously and by non-admin users. It supports a binary cache on source-based ports. I believe that Nix packages are easier to create and maintain than ports, however that is clearly subjective. The goal would be to have a BSD distribution based on FreeBSD or OpenBSD that migrates the ports collection to nix and then eventually replaces the OS configuration with Nix as well. This puts all of the machine configuration in a single place, simplifying the configuration of the operating system. This would be a lot of work, but I believe worth it in the long run. In the future I hope to have some time to add FreeBSD support back to Nix (it was not maintained and currently does not compile AFAIK) then getting ports to work under it. I'm curious, what technical problems do people see with this idea? What social problems? IMO, Nix, while having issues, is vastly superior to any package manager I have used, but I'm not sure if it fixes a problem that anyone else in the BSD community believes exists. Thanks, /Malcolm From pete at nomadlogic.org Fri Aug 14 14:52:39 2015 From: pete at nomadlogic.org (Pete Wright) Date: Fri, 14 Aug 2015 11:52:39 -0700 Subject: [talk] vxlan interfaces Message-ID: <55CE38F7.3030506@nomadlogic.org> noticed this addition to FreeBSD in the 10.2-RELEASE notes yesterday: "The vxlan(4) driver has been added, which creates a virtual Layer 2 (Ethernet) network overlaid in a Layer 3 (IP/UDP) network. The vxlan(4) driver is analogous to vlan(4), but is designed to be better suited for large, multiple-tenant datacenter environments. [r284365]" https://www.freebsd.org/cgi/man.cgi?query=vxlan&apropos=0&sektion=4&manpath=FreeBSD+10.2-RELEASE&arch=default&format=html This looks pretty interesting to me and it looks like this is supported by Cisco (as well as other switch vendors probably too). Has anyone on the list used this in production? I'd love to hear some use-cases - I can think of it being useful in the virutalization world (amazon VPC for example). cheers! -pete -- Pete Wright pete at nomadlogic.org From bonsaime at gmail.com Sat Aug 15 08:50:05 2015 From: bonsaime at gmail.com (Jesse Callaway) Date: Sat, 15 Aug 2015 08:50:05 -0400 Subject: [talk] vxlan interfaces In-Reply-To: <55CE38F7.3030506@nomadlogic.org> References: <55CE38F7.3030506@nomadlogic.org> Message-ID: On Fri, Aug 14, 2015 at 2:52 PM, Pete Wright wrote: > noticed this addition to FreeBSD in the 10.2-RELEASE notes yesterday: > > "The vxlan(4) driver has been added, which creates a virtual Layer 2 > (Ethernet) network overlaid in a Layer 3 (IP/UDP) network. The vxlan(4) > driver is analogous to vlan(4), but is designed to be better suited for > large, multiple-tenant datacenter environments. [r284365]" > > > https://www.freebsd.org/cgi/man.cgi?query=vxlan&apropos=0&sektion=4&manpath=FreeBSD+10.2-RELEASE&arch=default&format=html > > This looks pretty interesting to me and it looks like this is supported by > Cisco (as well as other switch vendors probably too). Has anyone on the > list used this in production? > > I'd love to hear some use-cases - I can think of it being useful in the > virutalization world (amazon VPC for example). > > cheers! > -pete > > > -- > Pete Wright > pete at nomadlogic.org > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > I wonder if this is the sauce AWS uses/invented. -- -jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From bonsaime at gmail.com Sat Aug 15 10:05:22 2015 From: bonsaime at gmail.com (Jesse Callaway) Date: Sat, 15 Aug 2015 10:05:22 -0400 Subject: [talk] vxlan interfaces In-Reply-To: References: <55CE38F7.3030506@nomadlogic.org> Message-ID: On Sat, Aug 15, 2015 at 8:50 AM, Jesse Callaway wrote: > > On Fri, Aug 14, 2015 at 2:52 PM, Pete Wright wrote: > >> noticed this addition to FreeBSD in the 10.2-RELEASE notes yesterday: >> >> "The vxlan(4) driver has been added, which creates a virtual Layer 2 >> (Ethernet) network overlaid in a Layer 3 (IP/UDP) network. The vxlan(4) >> driver is analogous to vlan(4), but is designed to be better suited for >> large, multiple-tenant datacenter environments. [r284365]" >> >> >> https://www.freebsd.org/cgi/man.cgi?query=vxlan&apropos=0&sektion=4&manpath=FreeBSD+10.2-RELEASE&arch=default&format=html >> >> This looks pretty interesting to me and it looks like this is supported >> by Cisco (as well as other switch vendors probably too). Has anyone on the >> list used this in production? >> >> I'd love to hear some use-cases - I can think of it being useful in the >> virutalization world (amazon VPC for example). >> >> cheers! >> -pete >> >> >> -- >> Pete Wright >> pete at nomadlogic.org >> >> _______________________________________________ >> talk mailing list >> talk at lists.nycbug.org >> http://lists.nycbug.org/mailman/listinfo/talk >> > > > I wonder if this is the sauce AWS uses/invented. > > -- > -jesse > Looking over it.. .and I don't know how I missed your last sentence there. The RFC was drafted by engineers employed at Cisco, Arista, Broadcom, RedHat, Vmware, and Citrix. I'm pretty sure that this is what the VPC's use from how the Amazon sales engineers explained it. This is also how I got the impression that they invented it, but what was said was actually that "we had to write our own code for a custom solution". -- -jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From spork at bway.net Sat Aug 15 14:16:17 2015 From: spork at bway.net (Charles Sprickman) Date: Sat, 15 Aug 2015 14:16:17 -0400 Subject: [talk] vxlan interfaces In-Reply-To: References: <55CE38F7.3030506@nomadlogic.org> Message-ID: <9F6F4921-6272-42A6-A670-7E3F77CCFE4D@bway.net> On Aug 15, 2015, at 10:05 AM, Jesse Callaway wrote: > > On Sat, Aug 15, 2015 at 8:50 AM, Jesse Callaway wrote: > > On Fri, Aug 14, 2015 at 2:52 PM, Pete Wright wrote: > noticed this addition to FreeBSD in the 10.2-RELEASE notes yesterday: > > "The vxlan(4) driver has been added, which creates a virtual Layer 2 (Ethernet) network overlaid in a Layer 3 (IP/UDP) network. The vxlan(4) driver is analogous to vlan(4), but is designed to be better suited for large, multiple-tenant datacenter environments. [r284365]" > > https://www.freebsd.org/cgi/man.cgi?query=vxlan&apropos=0&sektion=4&manpath=FreeBSD+10.2-RELEASE&arch=default&format=html > > This looks pretty interesting to me and it looks like this is supported by Cisco (as well as other switch vendors probably too). Has anyone on the list used this in production? > > I'd love to hear some use-cases - I can think of it being useful in the virutalization world (amazon VPC for example). > > cheers! > -pete > > > -- > Pete Wright > pete at nomadlogic.org > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > > > I wonder if this is the sauce AWS uses/invented. > > -- > -jesse > > Looking over it.. .and I don't know how I missed your last sentence there. The RFC was drafted by engineers employed at Cisco, Arista, Broadcom, RedHat, Vmware, and Citrix. I'm pretty sure that this is what the VPC's use from how the Amazon sales engineers explained it. This is also how I got the impression that they invented it, but what was said was actually that "we had to write our own code for a custom solution?. It?s crazy stuff: "The vxlan interface encapsulates the Ethernet frame by prepending IP/UDP and vxlan headers. Thus, the encapsulated (inner) frame is able to transmitted over a routed, Layer 3 network to the remote host.? Sounds very similar to Ethernet over IP found in Mikrotik routers. The bummer is that even though vxlan tunnels over IP, you can?t make use of it outside the datacenter since it wants (requires?) jumbo frames to fit all the extra prepends on the encapsulated ethernet frame. Regardless, it?s certainly nifty. Charles > > -- > -jesse > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk -------------- next part -------------- An HTML attachment was scrubbed... URL: From pete at nomadlogic.org Sat Aug 15 21:53:41 2015 From: pete at nomadlogic.org (Pete Wright) Date: Sat, 15 Aug 2015 18:53:41 -0700 Subject: [talk] vxlan interfaces In-Reply-To: References: <55CE38F7.3030506@nomadlogic.org> Message-ID: <55CFED25.9030606@nomadlogic.org> On 08/15/2015 07:05 AM, Jesse Callaway wrote: > On Sat, Aug 15, 2015 at 8:50 AM, Jesse Callaway wrote: > >> >> On Fri, Aug 14, 2015 at 2:52 PM, Pete Wright wrote: >> >>> noticed this addition to FreeBSD in the 10.2-RELEASE notes yesterday: >>> >>> "The vxlan(4) driver has been added, which creates a virtual Layer 2 >>> (Ethernet) network overlaid in a Layer 3 (IP/UDP) network. The vxlan(4) >>> driver is analogous to vlan(4), but is designed to be better suited for >>> large, multiple-tenant datacenter environments. [r284365]" >>> >>> >>> https://www.freebsd.org/cgi/man.cgi?query=vxlan&apropos=0&sektion=4&manpath=FreeBSD+10.2-RELEASE&arch=default&format=html >>> >>> This looks pretty interesting to me and it looks like this is supported >>> by Cisco (as well as other switch vendors probably too). Has anyone on the >>> list used this in production? >>> >>> I'd love to hear some use-cases - I can think of it being useful in the >>> virutalization world (amazon VPC for example). >>> >>> cheers! >>> -pete >>> >>> >>> -- >>> Pete Wright >>> pete at nomadlogic.org >>> >>> _______________________________________________ >>> talk mailing list >>> talk at lists.nycbug.org >>> http://lists.nycbug.org/mailman/listinfo/talk >>> >> >> >> I wonder if this is the sauce AWS uses/invented. >> >> -- >> -jesse >> > > Looking over it.. .and I don't know how I missed your last sentence there. > The RFC was drafted by engineers employed at Cisco, Arista, Broadcom, > RedHat, Vmware, and Citrix. I'm pretty sure that this is what the VPC's use > from how the Amazon sales engineers explained it. This is also how I got > the impression that they invented it, but what was said was actually that > "we had to write our own code for a custom solution". > yea i agree with you jesse - def seems like an ideal fit for VPC type deployments. okan pointed me towards a talk at asia bsdcon about vxlan in openbsd: https://www.youtube.com/watch?v=ufeEP_hzFN0 -pete -- Pete Wright pete at nomadlogic.org From george at ceetonetechnology.com Mon Aug 17 12:01:05 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Mon, 17 Aug 2015 12:01:05 -0400 Subject: [talk] Kirk McKusick interview Message-ID: <55D20541.3070505@ceetonetechnology.com> If anyone didn't catch this: http://www.infoq.com/articles/freebsd-design-implementation-review From john at netpurgatory.com Mon Aug 17 09:24:12 2015 From: john at netpurgatory.com (John C. Vernaleo) Date: Mon, 17 Aug 2015 09:24:12 -0400 (EDT) Subject: [talk] Idea: NixBSD In-Reply-To: <864mk1lwq2.fsf@gmail.com> References: <864mk1lwq2.fsf@gmail.com> Message-ID: Internet gremlins ate my reply the first time, trying again. On Fri, 14 Aug 2015, Malcolm Matalka wrote: > Hello everyone, > > I'd like to throw out an idea and get peoples reactions. This is not a > concrete plan or anything I'm actively working on, more of a long term > vision I'd love to see happen. > > Idea: > > A distribution of BSD that leverages the Nix package manager down to the > OS configuration. Roughly a BSD version of NixOS. > > Motivation: > > The effort put into the ports is fantastic but the ports system is rigid > in that supporting multiple versions of a single package requires a fair > amount of effort, installing/removing ports is limited to the > administrator of the machine, each port has to produce a unique set of > output names, source based vs binary based might have a similar source > but are distinct package systems. > > Solution: > > The Nix package manager is a purely functional package manager that > supports source-based, and binary-based packages. It allows multiple > versions of a port to be installed simultaneously and by non-admin > users. It supports a binary cache on source-based ports. I believe > that Nix packages are easier to create and maintain than ports, however > that is clearly subjective. > > The goal would be to have a BSD distribution based on FreeBSD or OpenBSD > that migrates the ports collection to nix and then eventually replaces > the OS configuration with Nix as well. This puts all of the machine > configuration in a single place, simplifying the configuration of the > operating system. > > This would be a lot of work, but I believe worth it in the long run. In > the future I hope to have some time to add FreeBSD support back to Nix > (it was not maintained and currently does not compile AFAIK) then > getting ports to work under it. > > I'm curious, what technical problems do people see with this idea? What > social problems? IMO, Nix, while having issues, is vastly superior to > any package manager I have used, but I'm not sure if it fixes a problem > that anyone else in the BSD community believes exists. > > Thanks, > /Malcolm > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > It's funny you should mention that since I've been playing around with NixOS recently and it has some pretty cool ideas. I do a lot of work on Bitrig ports and there are definitely features from nix that I would love to have. It would also be interesting to be able to use the same package manager on both linux and bsd (the number of times I've typed apt-get install on the bitrig box and port install on the linux box is startling high). That being said, from Bitrig I've seen that keep any number of ports up-to-date is a huge amount of work unless you have an army of people to help out (and I'm sure even then it is work). If you feel like poking around with it on openbsd and/or bitrig, feel free to ping me offlist or on irc cause I'd be interested in poking around a bit as well. John ------------------------------------------------------- John C. Vernaleo, Ph.D. www.netpurgatory.com john at netpurgatory.com ------------------------------------------------------- From ike at blackskyresearch.net Tue Aug 18 08:40:00 2015 From: ike at blackskyresearch.net (Isaac (.ike) Levy) Date: Tue, 18 Aug 2015 08:40:00 -0400 Subject: [talk] Idea: NixBSD In-Reply-To: <864mk1lwq2.fsf@gmail.com> References: <864mk1lwq2.fsf@gmail.com> Message-ID: <1439901603-2454813.06857236.ft7ICdxQ9013880@rs149.luxsci.com> Hi Malcom, Constructively, I'd like to address your thought from what I believe to be a general *BSD perspective, > On Aug 14, 2015, at 11:24 AM, Malcolm Matalka wrote: > > Hello everyone, > > I'd like to throw out an idea and get peoples reactions. This is not a > concrete plan or anything I'm actively working on, more of a long term > vision I'd love to see happen. > > Idea: > > A distribution of BSD that leverages the Nix package manager down to the > OS configuration. Roughly a BSD version of NixOS. The BSD's are not distros, that has been very important to the longevity and success of the *BSD's. A BSD is a full operating system, end to end. Here is a great presentation on just that topic, https://www.youtube.com/watch?v=wwbO4eTieQY > > Motivation: > > The effort put into the ports is fantastic but the ports system is rigid > in that supporting multiple versions of a single package requires a fair > amount of effort, installing/removing ports is limited to the > administrator of the machine, each port has to produce a unique set of > output names, source based vs binary based might have a similar source > but are distinct package systems. > > Solution: > > The Nix package manager is a purely functional package manager that > supports source-based, and binary-based packages. It allows multiple > versions of a port to be installed simultaneously and by non-admin > users. It supports a binary cache on source-based ports. I believe > that Nix packages are easier to create and maintain than ports, however > that is clearly subjective. > > The goal would be to have a BSD distribution based on FreeBSD or OpenBSD > that migrates the ports collection to nix and then eventually replaces > the OS configuration with Nix as well. Building your idea 'based on FreeBSD or OpenBSD', well, they are very different beasts- and again, are full, complete operating systems- not just a kernel. You could perhaps get one working release together based on either OS, but if you were to pick and choose parts of the OS base to use, you'd have a very hard time keeping your derived OS up to date with "upstream", to use Linux vocabulary. Conversely, if you tweaked your idea just a tad- and worked to make the Nix package manager work well on any *BSD system, that is a far simpler proposition. No matter what you think of the pkgsrc project, they have achieved truly remarkable portability, and longevity. > This puts all of the machine > configuration in a single place, simplifying the configuration of the > operating system. It is worth noting here, that the *BSD's do not suffer this config confusion problem in the same way that many Linux distributions do. In the base OS, /etc and rc are simple, and 'in an single place'. BSD users and developers are typically much more proactive and involved with the various 3rd party software they use, and often focus on making *that* software less '/etc spammy'- instead of adding another layer of abstraction in the middle. This isn't a hard rule in the *BSD's, but it's certainly a world apart from any Linux distros- in implementation and mindset. > > This would be a lot of work, but I believe worth it in the long run. In > the future I hope to have some time to add FreeBSD support back to Nix > (it was not maintained and currently does not compile AFAIK) then > getting ports to work under it. Just getting Nix to run on say, FreeBSD or OpenBSD, would be a huge start. I'm betting most of the problems are in bash-ism's in the build of it, I'd love to hear how this goes for you- and certainly feel free to ask questions about the *BSD end of things in here! (If people can't help, they certainly can know where to point you). > > I'm curious, what technical problems do people see with this idea? What > social problems? IMO, Nix, while having issues, is vastly superior to > any package manager I have used, but I'm not sure if it fixes a problem > that anyone else in the BSD community believes exists. That, I'm afraid, is indeed insightful- I'm not sure people believe these problems exist. Additionally, speaking for myself, the install is a bit of a non-starter from a security and best practices perspective, "curl https://nixos.org/nix/install | sh" A lot can certainly go wrong there. Yet, who knows- this could finally be the light, clean, portable package manager that people could really use! Good luck, and do tell us if you get it to build! Best, .ike > > Thanks, > /Malcolm > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > From raulcuza at gmail.com Tue Aug 18 21:45:57 2015 From: raulcuza at gmail.com (Raul Cuza) Date: Tue, 18 Aug 2015 21:45:57 -0400 Subject: [talk] the next con: some thoughts (1 of 2) In-Reply-To: <55CCA398.20705@ceetonetechnology.com> References: <55CCA398.20705@ceetonetechnology.com> Message-ID: On Thu, Aug 13, 2015 at 10:03 AM, George Rosamond < george at ceetonetechnology.com> wrote: > Informally, there has been a lot of discussions about our next > NYCBSDCon. Normally, this type of discussion would start on admin@, but > we need to break that routine. > > Conferences are lot of work, and that needs to be understood. It's > essentially a second (uncompensated) job to carry for a number of > people, and it's exhausting. Yes, the rewards are great, and we are > proud of what we've done, but that point needs to be comprehended. > > On that note, the last one in February 2014 was easier than in the past > for a few reasons. A lot of people took on responsibility for the event. > In particular, Patrick M and Mike N completely dealt with a/v to the > extent that everyone else was out of the loop. Additionally, doing a > day-long event focused on the NYC area, and avoiding the hassles of > coordinating hotels, etc., was an enormous relief. > > So arrive today in August 2015. It's been a year and a half since our > last con. > > There are a few spots I've looked at recently that could be ideal. And > yesterday, I looked at a new space at 150 Broadway that just opened > called LMHQ, which is one of those privately funded startup/hangout tech > spaces. NYI is a founding member, and they are happy to enable us to > use the space. > > LMHQ's main area fits up to 140 people with two screens, plus has a > bunch of conference rooms. It's an ideal spot and right next to the new > Fulton Street station downtown. > > At this point, it's worth considering something, say, at the end of > February, assuming all the pieces can move into place. That includes a > lot of people engaged in making it happen and taking responsibility for > it as in the past. > > The next email will be about content... > > g > > > At the last code-a-thon I co-ran, I meet someone from Grand Central Tech [ http://www.grandcentraltech.com/ ] who said their space was open for big events. I'm not sure if they have one space that can hold 140 people, but I don't mind asking. Let me know if you want me to reach out. If not, let me know when the kickoff cool-beverage session, I mean planning session will be, and I'll commit to something else that needs to be done to make this happen. ra?l -------------- next part -------------- An HTML attachment was scrubbed... URL: From freebsd at fongaboo.com Wed Aug 19 11:59:25 2015 From: freebsd at fongaboo.com (freebsd at fongaboo.com) Date: Wed, 19 Aug 2015 11:59:25 -0400 (EDT) Subject: [talk] insert line after matched pattern (fwd) Message-ID: Anyone know how to insert a line after a matched pattern using sed, awk or any other method that gets the job done? Any tutorials I find seem not to work in BSD due to GNU sed vs. non-GNU? ------------------------------------------------------------------------- shot through the heart ooh baby do you know what that's worth and you're to blame ooh heaven is a place on earth darling you give love they say in heaven love comes first a bad name we'll make heaven a place on earth ORBITAL "Halcyon Live" _______________________________________________ CDBUG-talk mailing list CDBUG-talk at lists.nycbug.org http://lists.nycbug.org/mailman/listinfo/cdbug-talk From franco at opnsense.org Wed Aug 19 12:05:26 2015 From: franco at opnsense.org (Franco Fichtner) Date: Wed, 19 Aug 2015 18:05:26 +0200 Subject: [talk] insert line after matched pattern (fwd) In-Reply-To: References: Message-ID: <943EE361-DB96-4F02-894D-5B83C86E34E0@opnsense.org> Hi, I had a similar issue once, maybe this code snippet can help you (basically going the route of an extra file with multiple lines): https://github.com/opnsense/tools/blob/544cb95fda/build/core.sh#L78-L79 Not a beauty and not in production code anymore, but it did work. :) Cheers, Franco > On 19 Aug 2015, at 17:59, freebsd at fongaboo.com wrote: > > > Anyone know how to insert a line after a matched pattern using sed, awk or any other method that gets the job done? Any tutorials I find seem not to work in BSD due to GNU sed vs. non-GNU? > > ------------------------------------------------------------------------- > shot through the heart ooh baby do you know what that's worth > and you're to blame ooh heaven is a place on earth > darling you give love they say in heaven love comes first > a bad name we'll make heaven a place on earth > ORBITAL "Halcyon Live" > > _______________________________________________ > CDBUG-talk mailing list > CDBUG-talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/cdbug-talk > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk From skreuzer at exit2shell.com Wed Aug 19 12:18:22 2015 From: skreuzer at exit2shell.com (Steven Kreuzer) Date: Wed, 19 Aug 2015 12:18:22 -0400 Subject: [talk] insert line after matched pattern (fwd) In-Reply-To: References: Message-ID: > On Aug 19, 2015, at 11:59 AM, freebsd at fongaboo.com wrote: > > > Anyone know how to insert a line after a matched pattern using sed, awk or any other method that gets the job done? Any tutorials I find seem not to work in BSD due to GNU sed vs. non-GNU? With awk: $ echo 'FIRST_LINE\nLAST_LINE' | awk '1;/FIRST_LINE/{print "NEWLINE"}' FIRST_LINE NEWLINE LAST_LINE The '1' is a shortcut for {print} so it will always print the line. However, if FIRST_LINE gets matched, it will also print NEWLINE From njt at ayvali.org Wed Aug 19 12:30:01 2015 From: njt at ayvali.org (N.J. Thomas) Date: Wed, 19 Aug 2015 12:30:01 -0400 Subject: [talk] insert line after matched pattern (fwd) In-Reply-To: References: Message-ID: <20150819163000.GS88312@zaph.org> * [2015-08-19 11:59:25-0400]: > Anyone know how to insert a line after a matched pattern using sed, > awk or any other method that gets the job done? Any tutorials I find > seem not to work in BSD due to GNU sed vs. non-GNU? The sed command to use is (a)ppend: $ cat foo line1 line2 line3 $ cat foo|sed '/^line2/a\ foo ' line1 line2 foo line3 Note that BSD version above (I'm running this on FreeBSD 10.1) requires a new line after the backslash delimeter, whereas GNU sed doesn't: $ cat foo|gsed '/^line2/a\foo' line1 line2 foo line3 You can even leave out the \ after the a with GNU sed, and it won't complain (unless you run with --posix). hth, Thomas From freebsd at fongaboo.com Wed Aug 19 12:48:06 2015 From: freebsd at fongaboo.com (freebsd at fongaboo.com) Date: Wed, 19 Aug 2015 12:48:06 -0400 (EDT) Subject: [talk] insert line after matched pattern (fwd) In-Reply-To: References: Message-ID: This worked well. Thank you! On Wed, 19 Aug 2015, Steven Kreuzer wrote: > >> On Aug 19, 2015, at 11:59 AM, freebsd at fongaboo.com wrote: >> >> >> Anyone know how to insert a line after a matched pattern using sed, awk or any other method that gets the job done? Any tutorials I find seem not to work in BSD due to GNU sed vs. non-GNU? > > > With awk: > > $ echo 'FIRST_LINE\nLAST_LINE' | awk '1;/FIRST_LINE/{print "NEWLINE"}' > FIRST_LINE > NEWLINE > LAST_LINE > > The '1' is a shortcut for {print} so it will always print the line. However, if FIRST_LINE gets matched, it will also print NEWLINE From ike at blackskyresearch.net Wed Aug 19 12:50:22 2015 From: ike at blackskyresearch.net (Isaac (.ike) Levy) Date: Wed, 19 Aug 2015 12:50:22 -0400 Subject: [talk] insert line after matched pattern (fwd) In-Reply-To: <943EE361-DB96-4F02-894D-5B83C86E34E0@opnsense.org> References: <943EE361-DB96-4F02-894D-5B83C86E34E0@opnsense.org> Message-ID: <1440003063-6606809.41685154.ft7JGoN88022462@rs149.luxsci.com> Fun! I'll also add: # after your pattern match, inserting on a new line: # sed '/pattern/a \ line to insert after line with pattern\ ' file # before your pattern match, inserting on a new line: $ sed '/pattern/a \ line to insert after line with pattern\ ' file -- Notable fun, Infamous sed1line.txt and awk1line.txt docs, (can be found all over the net), http://blackskyresearch.net/sed1line.txt http://blackskyresearch.net/awk1line.txt Best, .ike > On Aug 19, 2015, at 12:05 PM, Franco Fichtner wrote: > > Hi, > > I had a similar issue once, maybe this code snippet > can help you (basically going the route of an extra file > with multiple lines): > > https://github.com/opnsense/tools/blob/544cb95fda/build/core.sh#L78-L79 > > Not a beauty and not in production code anymore, but it > did work. :) > > > Cheers, > Franco > >> On 19 Aug 2015, at 17:59, freebsd at fongaboo.com wrote: >> >> >> Anyone know how to insert a line after a matched pattern using sed, awk or any other method that gets the job done? Any tutorials I find seem not to work in BSD due to GNU sed vs. non-GNU? >> >> ------------------------------------------------------------------------- >> shot through the heart ooh baby do you know what that's worth >> and you're to blame ooh heaven is a place on earth >> darling you give love they say in heaven love comes first >> a bad name we'll make heaven a place on earth >> ORBITAL "Halcyon Live" >> >> _______________________________________________ >> CDBUG-talk mailing list >> CDBUG-talk at lists.nycbug.org >> http://lists.nycbug.org/mailman/listinfo/cdbug-talk >> >> _______________________________________________ >> talk mailing list >> talk at lists.nycbug.org >> http://lists.nycbug.org/mailman/listinfo/talk > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > From george at ceetonetechnology.com Sun Aug 23 21:04:09 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Sun, 23 Aug 2015 21:04:09 -0400 Subject: [talk] NYC*BUG Upcoming Message-ID: <55DA6D89.3080109@ceetonetechnology.com> We have a good line-up for NYC... Aug 24: Classical Code Reading Group on true(1) and false(1) Sept 16: Ike Levy on OPNsense Oct 7: TBA Nov 19: Stephen R Bourne And note upcoming BSDCons vBSDCon September 11-13 http://vbsdcon.com/ EuroBSDCon October 3-4 https://2015.eurobsdcon.org/ BSDCon Brasil: October 9-10 http://2015.bsdcon.br/ ***** Some new BSD hackers are in town, and they import an event they have held in Scandinavia. We were excited to hear about the meeting content and form, and are happy to get the word out to the NYC*BUG lists: The search for truth: the `true` and `false` programs August 24, 7:00 PM thoughtbot, 1384 Broadway 20th Floor, New York, NY (map) This meetup will concentrate on simple and common commands: true and false. We will start with the OpenBSD true program and compare it to FreeBSD's, Solaris', GNU bash's, and GNU's. They all have different complexity, and some even have different features, which should provide for an interesting discussion. See the Meetup page for more details: http://www.meetup.com/Classical-Code-Reading-Group-of-New-York/events/224744308/ ***************** September 16: OPNsense: On the Shoulders of Giants, Isaac (.ike) Levy 18:45, Stone Creek Bar & Lounge: 140 E 27th St Abstract OPNsense is a BSD-licensed, easy-to-use and easy-to-build FreeBSD-based firewall and routing platform. This presentation is a hands-on preview of OPNsense, and should appeal to a wide range of people looking for BSD based router and firewall platforms. Speaker Bio Isaac (.ike) Levy is a crusty UNIX Hacker. ike, a long-time pfSense user, has moved on to become a contributor to the OPNSense project. Ike has been focused on i18n work, and Japanese translations, and for his sins, has been hacking on AWS AMI builds: http://dotike.github.io/opnsense.core.ja_JP.UTF8/ In 2006, ike gave an overview on pfSense and it`s mother project m0n0wall, which were new and exciting router platforms back then, "throw your Linksys/SoHo/WiFi router in the garbage where it belongs" In 2010, ike gave an overview of life with pfSense in Datacenter/Large deployments, "you might wanna` put your Sonicwall/Juniper/Cisco routers up on Ebay." A long-time community contributor to the *BSD's, ike is obsessed with high-availability and redundant networked servers systems, mostly because he likes to sleep at night. Standing on the shoulders of giants, his background includes partnering to run a Virtual Server ISP before anyone called it a cloud, as well as having a long history building internet-facing infrastructure with UNIX systems. .ike has been a part of NYC*BUG since it was first launched in January 2004. He was a long-time member of the Lower East Side Mac Unix User Group, and is still in denial that this group no longer exists. He has spoken frequently on a number of UNIX and internet security topics at various venues, particularly on the topic of FreeBSD's jail(8). ***************** * October 7: TBA ***************** * November 19: Special Meeting, Stephen R. Bourne 18:45, TBA Notice: special meeting, not regular date Abstract my history and background how and why we had to re write the shell why I wrote my own memory management key language design decisions where those ideas came from what was hard to get right system changes we made to accommodate sh what the rules were in UNIX group what would I do differently today Speaker Bio Steve Bourne is computer scientist who is internationally known for his work on the UNIX operating system. While at Bell Laboratories, Steve designed the UNIX Command Language known as the "Bourne Shell". It is the standard command line interface to UNIX and is widely used today in scripting in the UNIX programming environment. Steve has a Bachelor's degree in mathematics from King's College London, England. He has a Diploma (or Master's degree) in Computer Science and a Ph.D. in mathematics from Trinity College, Cambridge. While at the University of Cambridge Computer Laboratory he worked on an ALGOL 68 compiler and CAMAL an early algebra system. After Cambridge, Steve spent nine years at Bell Labs with the Seventh Edition Unix team. As well as the Bourne shell, he wrote the adb debugger and published /The UNIX System/, the second book on the UNIX system, intended for a general readership. This book is recognized as a text for the effective use of UNIX. After Bell Labs, he spent 20 years in senior engineering management positions. At Cisco Systems, he was director of engineering for enterprise network management; at Sun Microsystems, he managed the Solaris 2.0 program; at Digital Equipment Corporation, he developed DEC's first RISC-based workstation; and at Silicon Graphics, he was Director of Software Engineering responsible for the introduction of the IRIS, the company's first graphics workstation. >From 2000 to 2002 he was President of the Association for Computing Machinery. For his work on computing he was made a Fellow of the ACM in 2005. He is also a Fellow of the Royal Astronomical Society. At present Steve is chief technology officer at Rally Venture Partners, a Menlo Park-based venture capital group in California. He is also the chair of the Editorial Advisory Board for /ACM Queue/, a magazine he started when he was President of the ACM. From ike at blackskyresearch.net Mon Aug 24 23:13:24 2015 From: ike at blackskyresearch.net (Isaac (.ike) Levy) Date: Mon, 24 Aug 2015 23:13:24 -0400 Subject: [talk] NYC*BUG Upcoming In-Reply-To: <55DA6D89.3080109@ceetonetechnology.com> References: <55DA6D89.3080109@ceetonetechnology.com> Message-ID: <1440472442-2440290.30105767.ft7P3DRX2025557@rs149.luxsci.com> Wow, On 08/23/15 21:04, George Rosamond wrote: > Some new BSD hackers are in town, and they import an event they have > held in Scandinavia. We were excited to hear about the meeting content > and form, and are happy to get the word out to the NYC*BUG lists: > > The search for truth: the `true` and `false` programs > > August 24, 7:00 PM > thoughtbot, 1384 Broadway 20th Floor, New York, NY (map) > > This meetup will concentrate on simple and common commands: true and > false. We will start with the OpenBSD true program and compare it to > FreeBSD's, Solaris', GNU bash's, and GNU's. They all have different > complexity, and some even have different features, which should provide > for an interesting discussion. > > See the Meetup page for more details: > OMG folks come out next month... > http://www.meetup.com/Classical-Code-Reading-Group-of-New-York/events/224744308/ So I just got home from this and was blown away, Mike B. (with the help of George B.) put together an excellent evening. I was the only NYC*BUG regular in attendance, (though some folks there had been to NYC*BUG meetings before). That was seriously a great time- come out next month folks! For fun, you can read what we read tonight as well - true and false... -- While reading the code consider the following discussion points in addition to any you think of: What is the code boilerplate and why is it there? This is a small program; how did the different implementations demonstrate this? Why does this program exist? What shortcuts did they take and how do those make it easier to read? For those who don't yet have five variants of true.c on your hard disk, you can find them online: OpenBSD: - http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.bin/true/true.sh?rev=1.2&content-type=text/x-cvsweb-markup - http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.bin/false/false.sh?rev=1.2&content-type=text/x-cvsweb-markup FreeBSD: - http://svnweb.freebsd.org/base/head/usr.bin/true/true.c?revision=216370&view=markup - http://svnweb.freebsd.org/base/head/usr.bin/false/false.c?revision=216370&view=markup Solaris: - https://github.com/illumos/illumos-gate/blob/master/usr/src/cmd/true/true.c - https://github.com/illumos/illumos-gate/blob/master/usr/src/cmd/false/false.c GNU Bash: - https://raw.githubusercontent.com/bminor/bash/master/examples/loadables/truefalse.c GNU: - http://git.savannah.gnu.org/cgit/coreutils.git/tree/src/true.c - http://git.savannah.gnu.org/cgit/coreutils.git/tree/src/false.c -- Best, .ike From mark.saad at ymail.com Wed Aug 26 10:05:52 2015 From: mark.saad at ymail.com (Mark Saad) Date: Wed, 26 Aug 2015 10:05:52 -0400 Subject: [talk] NeXTBSD Message-ID: <55DDC7C0.1060208@ymail.com> Hi Talk Has anyone watched this intresting talk by Kip and Jordan on a new fork/branch of FreeBSD . The short story is they are adding Mach IPC and some other NeXT/OSX bits to a stock FreeBSD to get something new. https://www.youtube.com/watch?t=222&v=49sPYHh473U -- Mark Saad| mark.saad at ymail.com From briancoca+nycbug at gmail.com Sun Aug 30 12:31:26 2015 From: briancoca+nycbug at gmail.com (Brian Coca) Date: Sun, 30 Aug 2015 12:31:26 -0400 Subject: [talk] classifying BSD init Message-ID: Hi all, As part of my work on Ansible I wanted to revamp init system detection and break the current monolithic service plugin into smaller more targeted ones (systemd, upstart, sysv, etc). Since the BSD support is important to me (and hopefully others) I wanted to ask the community how I should break this down. What makes more sense? A unified bsdinit plugin that accounts for the differences on each system or independent plugins? Are There are enough differences in options that the user interface might become too complicated? 99% of my BSD usage has been FreeBSD with some OpenBSD, so I'm not sure on how much the daemon control systems differ across these and other BSD distributions. The following is my detection code (python). Currently I'm using this (which can also just be the fallback): elif self.facts['system'].endswith('BSD'): self.facts['service_mgr'] = 'bsdinit' But I was thinking in the lines of this: elif self.facts['system'] == 'FreeBSD': self.facts['service_mgr'] = 'rcconf' elif self.facts['system'] == 'NetBSD': self.facts['service_mgr'] = 'rcd' elif self.facts['system'] == 'OpenBSD': self.facts['service_mgr'] = 'rcctl' elif self.facts['system'].endswith('BSD'): self.facts['service_mgr'] = 'bsdinit' This will also impact on how I break down the current service plugin. You can look at the current BSD support starting here: https://github.com/ansible/ansible-modules-core/blob/devel/system/service.py#L942 Any info about other BSD init systems is also welcome. Thanks in advance, ------------ Brian Coca From justin at shiningsilence.com Sun Aug 30 13:13:20 2015 From: justin at shiningsilence.com (Justin Sherrill) Date: Sun, 30 Aug 2015 13:13:20 -0400 Subject: [talk] classifying BSD init In-Reply-To: References: Message-ID: Don't forget DragonFly, which does not end with "BSD". (And is rcconf) On Aug 30, 2015 12:33 PM, "Brian Coca" wrote: > Hi all, > > As part of my work on Ansible I wanted to revamp init system detection > and break the current monolithic service plugin into smaller more > targeted ones (systemd, upstart, sysv, etc). Since the BSD support is > important to me (and hopefully others) I wanted to ask the community > how I should break this down. > > What makes more sense? A unified bsdinit plugin that accounts for the > differences on each system or independent plugins? Are There are > enough differences in options that the user interface might become too > complicated? > > 99% of my BSD usage has been FreeBSD with some OpenBSD, so I'm not > sure on how much the daemon control systems differ across these and > other BSD distributions. > > The following is my detection code (python). > > Currently I'm using this (which can also just be the fallback): > > elif self.facts['system'].endswith('BSD'): > self.facts['service_mgr'] = 'bsdinit' > > But I was thinking in the lines of this: > > elif self.facts['system'] == 'FreeBSD': > self.facts['service_mgr'] = 'rcconf' > elif self.facts['system'] == 'NetBSD': > self.facts['service_mgr'] = 'rcd' > elif self.facts['system'] == 'OpenBSD': > self.facts['service_mgr'] = 'rcctl' > elif self.facts['system'].endswith('BSD'): > self.facts['service_mgr'] = 'bsdinit' > > > This will also impact on how I break down the current service plugin. > > You can look at the current BSD support starting here: > > https://github.com/ansible/ansible-modules-core/blob/devel/system/service.py#L942 > > Any info about other BSD init systems is also welcome. > > > Thanks in advance, > > ------------ > Brian Coca > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > -------------- next part -------------- An HTML attachment was scrubbed... URL: From briancoca+nycbug at gmail.com Sun Aug 30 13:28:43 2015 From: briancoca+nycbug at gmail.com (Brian Coca) Date: Sun, 30 Aug 2015 13:28:43 -0400 Subject: [talk] classifying BSD init In-Reply-To: References: Message-ID: Thanks, I did not know this, the service plugin is probably not currently working with DragonFly, I'll try to fix that also. On Sun, Aug 30, 2015 at 1:13 PM, Justin Sherrill wrote: > Fly, which does not end with "BSD". (And is rcconf) ------------ Brian Coca From patrik.lundin.swe at gmail.com Sun Aug 30 15:02:19 2015 From: patrik.lundin.swe at gmail.com (Patrik Lundin) Date: Sun, 30 Aug 2015 21:02:19 +0200 Subject: [talk] classifying BSD init In-Reply-To: References: Message-ID: Hello, Being the one who refactored the service module into the classes format it has today, as well as being the initial author of OpenBSD support my thoughts are these: I generally like the idea of a "one module per project" approach, there are pros and cons in both directions though. In a general sense one drawback I see is the potential for code duplication: * There is currently some code that is shared among FreeBSD and NetBSD, specifically service_enable_rcconf(). * There is also all of the state calculation, command execution code etc, which is currently shared among all supported platforms and not just the BSDs. Is there a plan to manage this in some way? Maby move these functions into some central location instead so they can be called by separate modules? This is a relevant question both for a unified bsdinit module, as well as having one module per project. On a more specific OpenBSD note: Calling the OpenBSD module "rcctl" might be misleading, given that this only applies to versions of OpenBSD new enough to actually have that utility (older versions use the /etc/rc.d scripts directly). Given that the package manager is called "openbsd_pkg", maby calling it "openbsd_rc" or "openbsd_service" would be better (as well as not dependant on the "service manager of the day")? I guess the main drawback of this naming convention would be if another OS decided to port the OpenBSD rcctl utility, however, this begs the question if the "rcctl" module would be separate from a "openbsd_rc.d" utility for the older OpenBSD systems.... ugh... :). -- Patrik Lundin On Sun, Aug 30, 2015 at 7:28 PM, Brian Coca wrote: > Thanks, I did not know this, the service plugin is probably not > currently working with DragonFly, I'll try to fix that also. > > On Sun, Aug 30, 2015 at 1:13 PM, Justin Sherrill > wrote: >> Fly, which does not end with "BSD". (And is rcconf) > > ------------ > Brian Coca > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk From briancoca+nycbug at gmail.com Sun Aug 30 15:30:55 2015 From: briancoca+nycbug at gmail.com (Brian Coca) Date: Sun, 30 Aug 2015 15:30:55 -0400 Subject: [talk] classifying BSD init In-Reply-To: References: Message-ID: Patrick, Thanks for your input, responding inline: On Sun, Aug 30, 2015 at 3:02 PM, Patrik Lundin wrote: > Hello, > > Being the one who refactored the service module into the classes > format it has today, as well as being the initial author of OpenBSD > support my thoughts are these: > I generally like the idea of a "one module per project" approach, > there are pros and cons in both directions though. I'm very thankful for this, you've corrected many of my assumptions and mistakes and improved BSD support greatly. > In a general sense one drawback I see is the potential for code duplication: > * There is currently some code that is shared among FreeBSD and > NetBSD, specifically service_enable_rcconf(). > * There is also all of the state calculation, command execution code > etc, which is currently shared among all supported platforms and not > just the BSDs. > > Is there a plan to manage this in some way? Maby move these functions > into some central location instead so they can be called by separate > modules? This is a relevant question both for a unified bsdinit > module, as well as having one module per project. shared module code lives in lib/ansible/module_utils, this is where i was planning to move shared code from the current service module. > On a more specific OpenBSD note: > Calling the OpenBSD module "rcctl" might be misleading, given that > this only applies to versions of OpenBSD new enough to actually have > that utility (older versions use the /etc/rc.d scripts directly). > Given that the package manager is called "openbsd_pkg", maby calling > it "openbsd_rc" or "openbsd_service" would be better (as well as not > dependant on the "service manager of the day")? > I guess the main drawback of this naming convention would be if > another OS decided to port the OpenBSD rcctl utility, however, this > begs the question if the "rcctl" module would be separate from a > "openbsd_rc.d" utility for the older OpenBSD systems.... ugh... :). > This also depends a lot on how they port, if it is 100% compatible my guess is that having an rcctl package will work for both OpenBSD and whatever system it was ported to. As for detection, I'll amend it to look for the rcctl binary and default to rcconf if otherwise. elif self.facts['system'] == 'OpenBSD': if module.find_bin('rcctl'): self.facts['service_mgr'] = 'rcctl' else: self.facts['service_mgr'] = 'rcconf' > -- > Patrik Lundin > > On Sun, Aug 30, 2015 at 7:28 PM, Brian Coca wrote: >> Thanks, I did not know this, the service plugin is probably not >> currently working with DragonFly, I'll try to fix that also. >> >> On Sun, Aug 30, 2015 at 1:13 PM, Justin Sherrill >> wrote: >>> Fly, which does not end with "BSD". (And is rcconf) >> >> ------------ >> Brian Coca >> >> _______________________________________________ >> talk mailing list >> talk at lists.nycbug.org >> http://lists.nycbug.org/mailman/listinfo/talk From patrik at sigterm.se Sun Aug 30 17:29:44 2015 From: patrik at sigterm.se (Patrik Lundin) Date: Sun, 30 Aug 2015 23:29:44 +0200 Subject: [talk] classifying BSD init In-Reply-To: References: Message-ID: <20150830212944.GA25742@major.strace.se> On Sun, Aug 30, 2015 at 03:30:55PM -0400, Brian Coca wrote: > > > > Is there a plan to manage this in some way? Maby move these functions > > into some central location instead so they can be called by separate > > modules? This is a relevant question both for a unified bsdinit > > module, as well as having one module per project. > > shared module code lives in lib/ansible/module_utils, this is where i > was planning to move shared code from the current service module. > This sounds like a nice way forward :). > > On a more specific OpenBSD note: > > Calling the OpenBSD module "rcctl" might be misleading, given that > > this only applies to versions of OpenBSD new enough to actually have > > that utility (older versions use the /etc/rc.d scripts directly). > > Given that the package manager is called "openbsd_pkg", maby calling > > it "openbsd_rc" or "openbsd_service" would be better (as well as not > > dependant on the "service manager of the day")? > > I guess the main drawback of this naming convention would be if > > another OS decided to port the OpenBSD rcctl utility, however, this > > begs the question if the "rcctl" module would be separate from a > > "openbsd_rc.d" utility for the older OpenBSD systems.... ugh... :). > > > > This also depends a lot on how they port, if it is 100% compatible my > guess is that having an rcctl package will work for both OpenBSD and > whatever system it was ported to. > As for detection, I'll amend it to look for the rcctl binary and > default to rcconf if otherwise. > > elif self.facts['system'] == 'OpenBSD': > if module.find_bin('rcctl'): > self.facts['service_mgr'] = 'rcctl' > else: > self.facts['service_mgr'] = 'rcconf' > This can work, rcconf then needs to be aware of what OS it is running on since OpenBSD uses a different approach than FreeBSD and NetBSD (it does not use the YES/NO boolean flags). For example, enabling a service (configuring that it should start at boot) is not possible on OpenBSD when the rcctl utility is missing because of this. This basically means "rcconf" will not actually be able to modify rc.conf.local on OpenBSD, but only call rc.d scripts directly to start/stop services. Under any circumstances, I would be happy to help out on the OpenBSD side of things if/when this gets going. Seems the cleanest way forward would be to remove classes from the service module in a step-by-step fashion as new modules are ready to take over. -- Patrik Lundin From john at netpurgatory.com Sun Aug 30 17:50:10 2015 From: john at netpurgatory.com (John C. Vernaleo) Date: Sun, 30 Aug 2015 17:50:10 -0400 (EDT) Subject: [talk] classifying BSD init In-Reply-To: References: Message-ID: Bitrig also doesn't end with *BSD (which has been a huge source of trouble for me with ports) but is likely to work with things OpenBSD works with. John ------------------------------------------------------- John C. Vernaleo, Ph.D. www.netpurgatory.com john at netpurgatory.com ------------------------------------------------------- On Sun, 30 Aug 2015, Justin Sherrill wrote: > > Don't forget DragonFly, which does not end with "BSD".? (And is rcconf) > > On Aug 30, 2015 12:33 PM, "Brian Coca" wrote: > Hi all, > > As part of my work on Ansible I wanted to revamp init system > detection > and break the current monolithic service plugin into smaller > more > targeted ones (systemd, upstart, sysv, etc). Since the BSD > support is > important to me (and hopefully others) I wanted to ask the > community > how I should break this down. > > What makes more sense? A unified bsdinit plugin that accounts > for the > differences on each system or independent plugins? Are There are > enough differences in options that the user interface might > become too > complicated? > > 99% of my BSD usage has been FreeBSD with some OpenBSD, so I'm > not > sure on how much the daemon control systems differ across these > and > other BSD distributions. > > The following is my detection code (python). > > Currently I'm using this (which can also just be the fallback): > > ? ? ? ? elif? self.facts['system'].endswith('BSD'): > ? ? ? ? ? ? ? ? self.facts['service_mgr'] = 'bsdinit' > > But I was thinking in the lines of this: > > ? ? ? ? elif self.facts['system'] == 'FreeBSD': > ? ? ? ? ? ? ? ? self.facts['service_mgr'] = 'rcconf' > ? ? ? ? elif self.facts['system'] == 'NetBSD': > ? ? ? ? ? ? ? ? self.facts['service_mgr'] = 'rcd' > ? ? ? ? elif self.facts['system'] == 'OpenBSD': > ? ? ? ? ? ? ? ? self.facts['service_mgr'] = 'rcctl' > ? ? ? ? elif? self.facts['system'].endswith('BSD'): > ? ? ? ? ? ? ? ? self.facts['service_mgr'] = 'bsdinit' > > > This will also impact on how I break down the current service > plugin. > > You can look at the current BSD support starting here: > https://github.com/ansible/ansible-modules-core/blob/devel/system/service.p > y#L942 > > Any info about other BSD init systems is also welcome. > > > Thanks in advance, > > ------------ > Brian Coca > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > > > From sjt.kar at gmail.com Mon Aug 31 10:18:23 2015 From: sjt.kar at gmail.com (Sujit K M) Date: Mon, 31 Aug 2015 19:48:23 +0530 Subject: [talk] classifying BSD init In-Reply-To: References: Message-ID: On Mon, Aug 31, 2015 at 12:32 AM, Patrik Lundin wrote: > I generally like the idea of a "one module per project" approach, > there are pros and cons in both directions though. Just out of curiosity, what is the difference between one module per project and the present scheme. IMHO there is a lot difference in *BSD as Project. Though This might be true in the case Linux Distros. You have X11 Dbus it is functioning the same way in Linux Distros, But not in the case of BSD. > > In a general sense one drawback I see is the potential for code duplication: > * There is currently some code that is shared among FreeBSD and > NetBSD, specifically service_enable_rcconf(). > * There is also all of the state calculation, command execution code > etc, which is currently shared among all supported platforms and not > just the BSDs. This is plainly depends on the refactoring you do. > > Is there a plan to manage this in some way? Maby move these functions > into some central location instead so they can be called by separate > modules? This is a relevant question both for a unified bsdinit > module, as well as having one module per project. This is purely refactoring. > > On a more specific OpenBSD note: > Calling the OpenBSD module "rcctl" might be misleading, given that > this only applies to versions of OpenBSD new enough to actually have > that utility (older versions use the /etc/rc.d scripts directly). > Given that the package manager is called "openbsd_pkg", maby calling > it "openbsd_rc" or "openbsd_service" would be better (as well as not > dependant on the "service manager of the day")? > I guess the main drawback of this naming convention would be if > another OS decided to port the OpenBSD rcctl utility, however, this > begs the question if the "rcctl" module would be separate from a > "openbsd_rc.d" utility for the older OpenBSD systems.... ugh... :). I agree with this. > > -- > Patrik Lundin From pete at nomadlogic.org Mon Aug 31 13:23:42 2015 From: pete at nomadlogic.org (Pete Wright) Date: Mon, 31 Aug 2015 10:23:42 -0700 Subject: [talk] classifying BSD init In-Reply-To: References: Message-ID: <55E48D9E.6030105@nomadlogic.org> On 08/30/15 09:31, Brian Coca wrote: > Hi all, > > As part of my work on Ansible I wanted to revamp init system detection > and break the current monolithic service plugin into smaller more > targeted ones (systemd, upstart, sysv, etc). Since the BSD support is > important to me (and hopefully others) I wanted to ask the community > how I should break this down. > > What makes more sense? A unified bsdinit plugin that accounts for the > differences on each system or independent plugins? Are There are > enough differences in options that the user interface might become too > complicated? > IMHO - and this may be unpopular - trying to create one unified plugin for multiple OS variants is not a good Unix design pattern. It certainly moves away from the writting one utility that is flexible approach towards the gnu/linux kitchen sink approach. having said that - i could see some benefit from having a main bsdinit class that inherits freebsd,openbsd,netbsd,etc. classes. this would allow one to put common methods and interfaces in the bsdinit class, then OS specific methods in bsdinit.freebsd for example. just my two bits, -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From pvarga at pvrg.net Mon Aug 31 14:12:15 2015 From: pvarga at pvrg.net (Peter Varga) Date: Mon, 31 Aug 2015 14:12:15 -0400 Subject: [talk] virtualization support in openbsd Message-ID: <1441044735.984651.370858017.12B21498@webmail.messagingengine.com> http://blog.gmane.org/gmane.os.openbsd.tech Thank you, Mike Larkin. From mmatalka at gmail.com Mon Aug 31 14:44:01 2015 From: mmatalka at gmail.com (Malcolm Matalka) Date: Mon, 31 Aug 2015 20:44:01 +0200 Subject: [talk] virtualization support in openbsd In-Reply-To: <1441044735.984651.370858017.12B21498@webmail.messagingengine.com> References: <1441044735.984651.370858017.12B21498@webmail.messagingengine.com> Message-ID: I don't know much about virtualization, how come not port kvm or bhyve? Den 31 aug 2015 8:13 em skrev "Peter Varga" : > http://blog.gmane.org/gmane.os.openbsd.tech > > Thank you, Mike Larkin. > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > -------------- next part -------------- An HTML attachment was scrubbed... URL: From justin at shiningsilence.com Mon Aug 31 14:55:22 2015 From: justin at shiningsilence.com (Justin Sherrill) Date: Mon, 31 Aug 2015 14:55:22 -0400 Subject: [talk] virtualization support in openbsd In-Reply-To: References: <1441044735.984651.370858017.12B21498@webmail.messagingengine.com> Message-ID: On Mon, Aug 31, 2015 at 2:44 PM, Malcolm Matalka wrote: > I don't know much about virtualization, how come not port kvm or bhyve? > Or Xen (which I think NetBSD is the only one supporting) or vkernels (DragonFly-only) or hey, jails, which has been around forever but everyone has a slightly different version of. It's frustrating to have so many projects working on the same problem in different ways, and not only every one with a case of NIH, but not even looking for a compatible methodology. At this point, Linux virtualization is more consistent than BSD, which goes against the one-system-with-predictable-results selling point that's been around for so long. From george at ceetonetechnology.com Mon Aug 31 15:03:20 2015 From: george at ceetonetechnology.com (George Rosamond) Date: Mon, 31 Aug 2015 15:03:20 -0400 Subject: [talk] virtualization support in openbsd In-Reply-To: References: <1441044735.984651.370858017.12B21498@webmail.messagingengine.com> Message-ID: <55E4A4F8.70107@ceetonetechnology.com> Justin Sherrill: > On Mon, Aug 31, 2015 at 2:44 PM, Malcolm Matalka wrote: >> I don't know much about virtualization, how come not port kvm or bhyve? >> > > Or Xen (which I think NetBSD is the only one supporting) or vkernels > (DragonFly-only) or hey, jails, which has been around forever but > everyone has a slightly different version of. > > It's frustrating to have so many projects working on the same problem > in different ways, and not only every one with a case of NIH, but not > even looking for a compatible methodology. At this point, Linux > virtualization is more consistent than BSD, which goes against the > one-system-with-predictable-results selling point that's been around > for so long. FreeBSD jail provides a very different functionality to Xen. Although jail(8) is a very different system than it was ten years ago. I think NetBSD was the first to support Xen (Johnny Lam meeting... 2005?), but I think more of them have support as guests. Although maybe only NetBSD can do dom0, if that's what you meant. And to keep up with the hip Linux virtualization "default choice" seems like a difficult task.. so I'm not sure what you mean by "consistent." Consistently moving with the breeze? ;` g From pete at nomadlogic.org Mon Aug 31 15:08:05 2015 From: pete at nomadlogic.org (Pete Wright) Date: Mon, 31 Aug 2015 12:08:05 -0700 Subject: [talk] virtualization support in openbsd In-Reply-To: References: <1441044735.984651.370858017.12B21498@webmail.messagingengine.com> Message-ID: <55E4A615.3010805@nomadlogic.org> On 08/31/15 11:55, Justin Sherrill wrote: > On Mon, Aug 31, 2015 at 2:44 PM, Malcolm Matalka wrote: >> I don't know much about virtualization, how come not port kvm or bhyve? >> > > Or Xen (which I think NetBSD is the only one supporting) or vkernels > (DragonFly-only) or hey, jails, which has been around forever but > everyone has a slightly different version of. > freebsd also has continuing support being developed to Xen (as well as support for virtual box). > It's frustrating to have so many projects working on the same problem > in different ways, and not only every one with a case of NIH, but not > even looking for a compatible methodology. At this point, Linux > virtualization is more consistent than BSD, which goes against the > one-system-with-predictable-results selling point that's been around > for so long. hah - i wish you were correct re: linux vitrualization being consistent. libvirt is *sorta* consistent if you stick to one distro - but once you start mixing centos, unbuntu and debian hypervisors proceed at your own risk. and Qemu/KVM also has interesting delta's between distro's and kernel revisions I have found. i'm all for having support for multiple hypervisors in the BSD ecosystem. i am willing to bet that the approach OpenBSD will take will be quite a bit different than the bhyve team. cheers, -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From justin at shiningsilence.com Mon Aug 31 15:34:26 2015 From: justin at shiningsilence.com (Justin Sherrill) Date: Mon, 31 Aug 2015 15:34:26 -0400 Subject: [talk] virtualization support in openbsd In-Reply-To: <55E4A615.3010805@nomadlogic.org> References: <1441044735.984651.370858017.12B21498@webmail.messagingengine.com> <55E4A615.3010805@nomadlogic.org> Message-ID: On Mon, Aug 31, 2015 at 3:08 PM, Pete Wright wrote: > hah - i wish you were correct re: linux vitrualization being consistent. > libvirt is *sorta* consistent if you stick to one distro - but once you > start mixing centos, unbuntu and debian hypervisors proceed at your own > risk. and Qemu/KVM also has interesting delta's between distro's and > kernel revisions I have found. Yes, but! It's the same thing between distros. It's not libvirtng on one and libvirthyve on another and rcvirt on another, none of which work in the same way. Differences in tooling is not the same as inconsistency in implementation. I realize my blanket complaint can be picked at in different ways; my frustration comes from the circles always being drawn between BSDs.