From mark.saad at ymail.com Mon Mar 3 20:14:03 2014 From: mark.saad at ymail.com (Mark Saad) Date: Mon, 3 Mar 2014 20:14:03 -0500 Subject: Cryptome article on fortuna Message-ID: <7D07DAB5-A411-4583-8D44-07A759E41E95@ymail.com> All Cryptome has a long and detailed reposted article on rng's in particular fortuna . It's by some SUNY students ; It's math heavy and a bit beyond casual reading . But worth the read . http://cryptome.org/2014/03/eat-entropy-have-it.pdf --- Mark -------------- next part -------------- An HTML attachment was scrubbed... URL: From george at ceetonetechnology.com Mon Mar 3 20:23:13 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Mon, 03 Mar 2014 20:23:13 -0500 Subject: Cryptome article on fortuna In-Reply-To: <7D07DAB5-A411-4583-8D44-07A759E41E95@ymail.com> References: <7D07DAB5-A411-4583-8D44-07A759E41E95@ymail.com> Message-ID: <53152B01.80009@ceetonetechnology.com> Mark Saad: > All > Cryptome has a long and detailed reposted article on rng's in particular fortuna . It's by some SUNY students ; It's math heavy and a bit beyond casual reading . But worth the read . > > http://cryptome.org/2014/03/eat-entropy-have-it.pdf Yes... and one of those authors is doing a meeting for us in either April or May, Yevgeniy Dodis. He spoke at Real World Crypto two months ago, and bashed the Linux RNG pretty hard. This is the link he sent me an hour ago... same paper. http://eprint.iacr.org/2014/167 Definitely will be worth the time to at least browse the article. It should be a very good (and intense) meeting. g From george at ceetonetechnology.com Wed Mar 5 09:14:23 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Wed, 05 Mar 2014 09:14:23 -0500 Subject: pfsense-dev discussion Message-ID: <5317313F.1030200@ceetonetechnology.com> So there is a long (and sometimes tiring) discussion on the pfsense-dev list: http://lists.pfsense.org/pipermail/dev/2014-March/date.html You could even go back to February if you are not exhausted by March :) It's a bit perplexing, but some points shout out loud and clear to me. Chris makes what is probably an offhand comment about how they should have licensed as GPLv2. Outside the question of how you take a FreeBSD system with pf and make it GPL is outside my legal areas of expertise (which even more nil outside of that), the license is clearly a side-casualty in the conversation. Chris is by all means "one of us." Some of us have known him since they launched the project, and have enormous respect for him. Keep that in mind. While I wouldn't take the whole thread and focus on the license comment, I also think the BSD license, and more importantly, the enormous user base of pfSense, protects them *in practice* against some (likely) kid taking the github repo and rebranding it. They have the trademark on pfSense, and attribution is a core part of the BSD license. Are they really threatened by cases such as that? I find the license very much secondary. Could someone really take a well-known BSD-licensed project, rerelease as their own with the same name, sucking down changes from their public repo, and really be a threat? Outside of the clear trademark issue, I don't get it. Seems like yet another test case for the relation between an open source project and the related business services more than anything to me. There's lots of GPL examples (in which the related business often ignores the license), but it would seem the discussions aren't open and generalized enough in our scene. g From george at ceetonetechnology.com Wed Mar 5 19:08:23 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Wed, 05 Mar 2014 19:08:23 -0500 Subject: tonight's meeting Message-ID: <5317BC77.7020702@ceetonetechnology.com> It's streaming at: nycbsdcon.org/2014/streaming.html First stream. And join us on irc freenode #nycbug g From schmonz at schmonz.com Wed Mar 5 22:04:36 2014 From: schmonz at schmonz.com (Amitai Schlair) Date: Wed, 5 Mar 2014 22:04:36 -0500 Subject: tonight's meeting In-Reply-To: <5317BC77.7020702@ceetonetechnology.com> References: <5317BC77.7020702@ceetonetechnology.com> Message-ID: On Mar 5, 2014, at 7:08 PM, George Rosamond wrote: > > It's streaming at: > > nycbsdcon.org/2014/streaming.html > > First stream. Dang, I missed it. Also I left my brother-in-law's dongle (seriously, the Apple-to-VGA one). Did anyone grab it? Thanks for being a fun and involved audience! From ericshane at eradman.com Thu Mar 6 14:47:42 2014 From: ericshane at eradman.com (Eric Radman) Date: Thu, 6 Mar 2014 14:47:42 -0500 Subject: pfsense-dev discussion In-Reply-To: <5317313F.1030200@ceetonetechnology.com> References: <5317313F.1030200@ceetonetechnology.com> Message-ID: <20140306194742.GA24077@vm.eradman.com> On Wed, Mar 05, 2014 at 09:14:23AM -0500, George Rosamond wrote: > So there is a long (and sometimes tiring) discussion on the > pfsense-dev list: > > http://lists.pfsense.org/pipermail/dev/2014-March/date.html > ... > While I wouldn't take the whole thread and focus on the license > comment, > I also think the BSD license, and more importantly, the enormous user > base of pfSense, protects them *in practice* against some (likely) kid > taking the github repo and rebranding it. > > They have the trademark on pfSense, and attribution is a core part of > the BSD license. Right on. The license is vital, but it doesn't _do_ anything; it allows the rest of the ecosystem to operate and build a community under well-understood terms. Rob Landley, the lead on toybox (formerly a busybox maintainer) noted in his talk at the Embedded Linux Conference 2013 that toybox uses a BSD-like license now because GPLv3 introduced confusion. The end result of all their time enforcing the GPL was a pile of patches none of which were written well enough to be considered a contribution. The topic is how to structure their company so that that it remains a viable business. Jim Thompson framed the problem clearly in 2014-March/000537.html > Are they really threatened by cases such as that? I find the license > very much secondary. I don't recall hearing of such a situation, so the answer appears to be "not likely". If someone did manage to provide _better_ release engineering and support than the people who built the product then the founders need to face down some hard questions; or if they're any good hire the buggers! From george at ceetonetechnology.com Thu Mar 6 19:12:21 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 06 Mar 2014 19:12:21 -0500 Subject: tmpfs Message-ID: <53190EE5.4080309@ceetonetechnology.com> I brought up the inclusion of the option of tmpfs(5) in various FreeBSD kernels where it doesn't currently (and oddly) exist such as for the Raspberry Pi and now Alix boards. I've been using tmpfs for a long while instead of md(4). I haven't had any issues when using for /var/log, /tmp or /var/tmp, but then I haven't done any real stress-testing outside of 'cleaning' databases and related stuff. FYI, Julio (who was there last night) ported it originally for NetBSD as a GSoC project a long while ago. My question is this, has anyone seen any drawbacks? Also, I have tended to use it with a defined size instead of letting it grow dynamically. Does anyone know the differences in the impact of setting it statically (like in /etc/fstab) or letting it grow dynamically? thanks g From justin at shiningsilence.com Thu Mar 6 19:33:33 2014 From: justin at shiningsilence.com (Justin Sherrill) Date: Thu, 6 Mar 2014 19:33:33 -0500 Subject: tmpfs In-Reply-To: <53190EE5.4080309@ceetonetechnology.com> References: <53190EE5.4080309@ceetonetechnology.com> Message-ID: Looks like it's getting wider testing now: http://svnweb.freebsd.org/base?view=revision&revision=262877 I've had several machines using tmpfs... but those were DragonFly, and I don't know how close the tmpfs implementations are. On Thu, Mar 6, 2014 at 7:12 PM, George Rosamond < george at ceetonetechnology.com> wrote: > I brought up the inclusion of the option of tmpfs(5) in various FreeBSD > kernels where it doesn't currently (and oddly) exist such as for the > Raspberry Pi and now Alix boards. > > I've been using tmpfs for a long while instead of md(4). I haven't had > any issues when using for /var/log, /tmp or /var/tmp, but then I haven't > done any real stress-testing outside of 'cleaning' databases and related > stuff. > > FYI, Julio (who was there last night) ported it originally for NetBSD as > a GSoC project a long while ago. > > My question is this, has anyone seen any drawbacks? > > Also, I have tended to use it with a defined size instead of letting it > grow dynamically. Does anyone know the differences in the impact of > setting it statically (like in /etc/fstab) or letting it grow dynamically? > > thanks > > g > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > -------------- next part -------------- An HTML attachment was scrubbed... URL: From edlinuxguru at gmail.com Thu Mar 6 20:00:35 2014 From: edlinuxguru at gmail.com (Edward Capriolo) Date: Thu, 6 Mar 2014 20:00:35 -0500 Subject: What is a hackathon In-Reply-To: <9B550292-1176-4914-99A6-1BD76AAD2CC8@gmail.com> References: <1391055640.6400.YahooMailBasic@web140102.mail.bf1.yahoo.com> <52E9D8D9.6000302@devio.us> <9B550292-1176-4914-99A6-1BD76AAD2CC8@gmail.com> Message-ID: On Friday, January 31, 2014, Ra?l Cuza wrote: > On Jan 29, 2014, at 23:45, Brian Callahan wrote: >> >> >> Slightly off-topic history lesson: OpenBSD held the first ever hackathon and invented the term. Have a trophy? > http://www.openbsd.org/hackathons.html >> > > I totally read that in Pavel Chekov's voice. > > Sent without help from A.I. | ' L ' | > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > -- Sorry this was sent from mobile. Will do less grammar and spell check than usual. -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: CAM00221.jpg Type: image/jpeg Size: 3127324 bytes Desc: not available URL: From george at ceetonetechnology.com Thu Mar 6 20:10:37 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 06 Mar 2014 20:10:37 -0500 Subject: tmpfs In-Reply-To: References: <53190EE5.4080309@ceetonetechnology.com> Message-ID: <53191C8D.4070200@ceetonetechnology.com> Justin Sherrill: > Looks like it's getting wider testing now: > > http://svnweb.freebsd.org/base?view=revision&revision=262877 Yup.. and I spawned a scrappy little battle on freebsd-arm about it. Oops! Should have just kept adding quietly to the ALIX and RPI-B kernels on my own hardware. Should have just whined about svn instead! > > I've had several machines using tmpfs... but those were DragonFly, and I > don't know how close the tmpfs implementations are. Well, Wikipedia also reflects my memory on this.. FreeBSD ported from NetBSD, and DFly ported NetBSDs also. g From george at ceetonetechnology.com Thu Mar 6 20:30:06 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 06 Mar 2014 20:30:06 -0500 Subject: What is a hackathon In-Reply-To: References: <1391055640.6400.YahooMailBasic@web140102.mail.bf1.yahoo.com> <52E9D8D9.6000302@devio.us> <9B550292-1176-4914-99A6-1BD76AAD2CC8@gmail.com> Message-ID: <5319211E.8070206@ceetonetechnology.com> Edward Capriolo: > On Friday, January 31, 2014, Ra?l Cuza wrote: >> On Jan 29, 2014, at 23:45, Brian Callahan wrote: >>> >>> >>> Slightly off-topic history lesson: OpenBSD held the first ever hackathon > and invented the term. > Have a trophy? > OT, although not nearly as OT as Ed. . . limit is now hard set at 500K for postings to this list. Sorry about that. Must have been set temporarily higher due to our Director of Artistic Endeavors, Mr. Levy, submitting clunky blobby images. If you get a bounce message for sending large attachments, please read the bounce instead of asking the list owners why. Ed: please continue on your unique brand of trolling. But 3M images are unfriendly to those without quad core phones. g From edlinuxguru at gmail.com Thu Mar 6 22:06:04 2014 From: edlinuxguru at gmail.com (Edward Capriolo) Date: Thu, 6 Mar 2014 22:06:04 -0500 Subject: What is a hackathon In-Reply-To: <5319211E.8070206@ceetonetechnology.com> References: <1391055640.6400.YahooMailBasic@web140102.mail.bf1.yahoo.com> <52E9D8D9.6000302@devio.us> <9B550292-1176-4914-99A6-1BD76AAD2CC8@gmail.com> <5319211E.8070206@ceetonetechnology.com> Message-ID: On Thursday, March 6, 2014, George Rosamond wrote: > Edward Capriolo: >> On Friday, January 31, 2014, Ra?l Cuza wrote: >>> On Jan 29, 2014, at 23:45, Brian Callahan wrote: >>>> >>>> >>>> Slightly off-topic history lesson: OpenBSD held the first ever hackathon >> and invented the term. >> Have a trophy? >> > > > OT, although not nearly as OT as Ed. . . limit is now hard set at 500K > for postings to this list. Sorry about that. Must have been set > temporarily higher due to our Director of Artistic Endeavors, Mr. Levy, > submitting clunky blobby images. > > If you get a bounce message for sending large attachments, please read > the bounce instead of asking the list owners why. > > Ed: please continue on your unique brand of trolling. But 3M images are > unfriendly to those without quad core phones. > > g > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > Sorry i was not trying to troll. I was at a meetup and that was a picture of the company hack day trophy. I instantly thought of this email chain and hack day prizes. What could be a more glorious prize then your name on a trophy, there for years after you leave the company. Btw i do have a quad core phone. Were you able to determine this from image meta data, email headers, or was it just a lucky guess? Either way i am impressed and at the same time scared. -- Sorry this was sent from mobile. Will do less grammar and spell check than usual. -------------- next part -------------- An HTML attachment was scrubbed... URL: From john at netpurgatory.com Fri Mar 7 08:20:52 2014 From: john at netpurgatory.com (John C. Vernaleo) Date: Fri, 7 Mar 2014 08:20:52 -0500 (EST) Subject: tmpfs In-Reply-To: <53191C8D.4070200@ceetonetechnology.com> References: <53190EE5.4080309@ceetonetechnology.com> <53191C8D.4070200@ceetonetechnology.com> Message-ID: I'm not using tmpfs myself, but I know at least one person who has been using it pretty heavily on bitrig and it seems to be working out well. This is on amd64. I really doubt he has tried on arm. ------------------------------------------------------- John C. Vernaleo, Ph.D. www.netpurgatory.com john at netpurgatory.com ------------------------------------------------------- On Thu, 6 Mar 2014, George Rosamond wrote: > Justin Sherrill: >> Looks like it's getting wider testing now: >> >> http://svnweb.freebsd.org/base?view=revision&revision=262877 > > Yup.. and I spawned a scrappy little battle on freebsd-arm about it. > Oops! Should have just kept adding quietly to the ALIX and RPI-B > kernels on my own hardware. > > Should have just whined about svn instead! > >> >> I've had several machines using tmpfs... but those were DragonFly, and I >> don't know how close the tmpfs implementations are. > > Well, Wikipedia also reflects my memory on this.. FreeBSD ported from > NetBSD, and DFly ported NetBSDs also. > > g > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > From mark.saad at ymail.com Fri Mar 7 15:15:45 2014 From: mark.saad at ymail.com (Mark Saad) Date: Fri, 07 Mar 2014 15:15:45 -0500 Subject: tmpfs In-Reply-To: References: <53190EE5.4080309@ceetonetechnology.com> <53191C8D.4070200@ceetonetechnology.com> Message-ID: <531A28F1.2060606@ymail.com> On 03/07/14 08:20, John C. Vernaleo wrote: > I'm not using tmpfs myself, but I know at least one person who has been > using it pretty heavily on bitrig and it seems to be working out well. > This is on amd64. I really doubt he has tried on arm. > > ------------------------------------------------------- > John C. Vernaleo, Ph.D. > www.netpurgatory.com > john at netpurgatory.com > ------------------------------------------------------- > > On Thu, 6 Mar 2014, George Rosamond wrote: > >> Justin Sherrill: >>> Looks like it's getting wider testing now: >>> >>> http://svnweb.freebsd.org/base?view=revision&revision=262877 >> >> Yup.. and I spawned a scrappy little battle on freebsd-arm about it. >> Oops! Should have just kept adding quietly to the ALIX and RPI-B >> kernels on my own hardware. >> >> Should have just whined about svn instead! >> >>> >>> I've had several machines using tmpfs... but those were DragonFly, and I >>> don't know how close the tmpfs implementations are. >> >> Well, Wikipedia also reflects my memory on this.. FreeBSD ported from >> NetBSD, and DFly ported NetBSDs also. >> >> g >> _________ Gman I used tmpfs in production FreeBSD 7.4 and 9.1 with out much issue. My use cases were as follows. A mysql tmp table space , a web session storage space, a mail server queue storage space , and for /usr/obj. In each case except /usr/obj I set a size on the mounts. My rational was that if I gave the user no set size someone would fill the tmpfs and potentially crash the system wile it was under load. Limiting them to 200M or 4G gave them working bounds that keep everything happy . Mark ______________________________________ >> talk mailing list >> talk at lists.nycbug.org >> http://www.nycbug.org/mailman/listinfo/talk >> > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk From mark.saad at ymail.com Fri Mar 7 18:10:22 2014 From: mark.saad at ymail.com (Mark Saad) Date: Fri, 7 Mar 2014 15:10:22 -0800 (PST) Subject: What is a hackathon In-Reply-To: Message-ID: <1394233822.4594.YahooMailMobile@web140104.mail.bf1.yahoo.com> Sorry for the top post . Yahoo mail hates me today . To clarify ed and I worked together , there is an inside joke in this thread ; I can not go into details . I totally expect a trophy to be created for the hacksathon winner . The jackass in charge of the hacksathon believes it's more of a "I did it better and I win" then let's work together and make something better challenge . I liked the discussion On This thread . Keep it up people Mark -------------- next part -------------- An HTML attachment was scrubbed... URL: From ike at blackskyresearch.net Wed Mar 12 10:36:18 2014 From: ike at blackskyresearch.net (Isaac (.ike) Levy) Date: Wed, 12 Mar 2014 14:36:18 +0000 Subject: FreeBSD 10Gbs Networking Working Message-ID: <201403121436.s2CEaIBO017200@rs102.luxsci.com> Hi All, Just a quick excited post to share: Using commodity hardware and 10Gbit Intel nics, and FreeBSD 10.0, I'm pushing an actual 9.1-9.3Gbps (netperf tests), across 4x nics, no tuning. -- TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 172.16.0.1 () port 0 AF_INET : histogram : interval : dirty data : demo Recv Send Send Socket Socket Message Elapsed Size Size Size Time Throughput bytes bytes bytes secs. 10^6bits/sec 65536 32768 32768 10.01 9388.17 -- Not a joke. I've been running these in loops simultaneously on all 4 ports. Game changed, mind blown. Best, .ike ------------ more details The OS is stock FreeBSD 10, dmesg posted here: http://nycbug.org/?action=dmesgd&do=view&dmesgid=2527 That HW costs about 3200 bucks. (Ponder that.) The only tuning I did was in bios- to disable hyperthreading (pci interrupts), and disable cpu virtualization. The two pci cards each have 2x SFP slots, with fiber SFP's. The 1u chassis puts them both on a X16 pcie3 bus, (each becomming X8 pci). The driver is ix, http://downloadmirror.intel.com/14688/eng/README.txt I think the man page for it is ixgbe(4), (but I may be wrong here?) http://www.freebsd.org/cgi/man.cgi?query=ixgbe&apropos=0&sektion=0&manpath=FreeBSD+10.0-RELEASE&arch=default&format=html General Notes: Apparently I'm CPU bound here, but using somewhat "typical" CPU's here, could increase the number cores and see what happens next. It's delightful to finally have a problem we can throw CPU at in the world again :) -- Additionally, I'm running tests using pfSense, (2.1.1-prerelease has the driver fixes to support these cards.) Since pfSense is based on FreeBSD 8.3, the performance is not obviously as stunning- but out the gate, with minimal fuss, I'm able to reliably push 4+Gbps across 2 ports (4Gbps each direction for total 8Gbps throughput), and 2+Gbps each direction when saturating all 4 ports (total 8Gbps throughput again). CPU is pegged, so I'm pretty certain that dropping in more cores will help, as well as more tuning. We've got some bigger boxes in the works, (8x total 10Gbe ports, 4x GigE ports on the motherboard, and a dedicated PCI slot for each card), so we'll be dropping more CPU cores in the box here too... We'll see what improvements this brings. While I'm personally looking forward to a bright shiny 10.x based pfSense in the future, I'm likewise impressed with the overall stability and responsiveness with this hardware. -- Some urls on tuning: I don't know much of this project, but these tuning urls have been fascinating, http://bsdrp.net/documentation/technical_docs/performance Other interesting 10Gbps nics, http://www.freebsd.org/cgi/man.cgi?query=cxgb&sektion=4&apropos=0&manpath=FreeBSD+10.0-RELEASE http://www.freebsd.org/cgi/man.cgi?query=cxgbe&sektion=4&apropos=0&manpath=FreeBSD+10.0-RELEASE Yeah yeah yeah Calomel, great notes on PCI calculations: https://calomel.org/network_performance.html From scottro at nyc.rr.com Wed Mar 12 17:03:17 2014 From: scottro at nyc.rr.com (Scott Robbins) Date: Wed, 12 Mar 2014 17:03:17 -0400 Subject: Skype Message-ID: <20140312210317.GA16971@scott1.scottro.net> Just thought I'd mention that for those interested, thanks to the efforts of xmj on FreeBSD forums, it is now almost trivial to get skype4 working. Some others have had success prior to this, but for me at least, it began taking more time than it was worth. (Especially as I only use it when my wife goes on business trips.) http://forums.freebsd.org/viewtopic.php?f=22&t=44800 is the thread where it's discussed, but for what it's worth, it was actually rather trivial. git clone http://github.com/xmj/linux-ports, then run a cp -r linux-ports/* /usr/ports, and follow the instructions in the README for loading linprocfs and linsysfs and it all just worked for me, tested on two different laptops. (In addition, one can use the CentOS 6 based Linux port which is somewhat newer than the current Fedora 10 one. There's also a working port for the CentOS 6 version of flash--and much as I suspect we all hate flash--well, Ok, much as I dislike flash--unfortunately, it's still a fact of life on far too many sites. Hopefully, this is of interest to a few people. -- Scott Robbins PGP keyID EB3467D6 ( 1B48 077D 66F6 9DB0 FDC2 A409 FA54 EB34 67D6 ) gpg --keyserver pgp.mit.edu --recv-keys EB3467D6 From justin at shiningsilence.com Thu Mar 13 10:54:28 2014 From: justin at shiningsilence.com (Justin Sherrill) Date: Thu, 13 Mar 2014 10:54:28 -0400 Subject: FreeBSD 10Gbs Networking Working In-Reply-To: <201403121436.s2CEaIBO017200@rs102.luxsci.com> References: <201403121436.s2CEaIBO017200@rs102.luxsci.com> Message-ID: On Wed, Mar 12, 2014 at 10:36 AM, Isaac (.ike) Levy < ike at blackskyresearch.net> wrote: > Using commodity hardware and 10Gbit Intel nics, and FreeBSD 10.0, I'm > pushing an actual 9.1-9.3Gbps (netperf tests), across 4x nics, no tuning. > For what it's worth: http://lists.dragonflybsd.org/pipermail/users/2014-March/128240.html Not exactly the same hardware, but comparable. Something I've thought would be useful is some sort of test setup that would automatically run test cases like this as new BSD versions come in - not necessarily for comparison between types, but to show advancement or regressions between releases. That's wandering off to a whole new topic... -------------- next part -------------- An HTML attachment was scrubbed... URL: From george at ceetonetechnology.com Thu Mar 13 12:45:21 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Mar 2014 12:45:21 -0400 Subject: next meeting... Message-ID: <5321E0A1.9050208@ceetonetechnology.com> ...just hit the announce@ list. Definitely spread the word as it should be a fascinating meeting. Tweet, friend it, whatever you kids do today. Note that we will be meeting: at NYU.. not Suspenders on Tuesday, April 1... not Wednesday April 2 at 7:15 PM... not 6:45 PM. g From ike at blackskyresearch.net Thu Mar 13 12:52:22 2014 From: ike at blackskyresearch.net (Isaac Levy) Date: Thu, 13 Mar 2014 12:52:22 -0400 Subject: FreeBSD 10Gbs Networking Working In-Reply-To: References: <201403121436.s2CEaIBO017200@rs102.luxsci.com> Message-ID: <1394729583-6431728.8271511.fs2DGqMmC017310@rs149.luxsci.com> > > On Mar 13, 2014, at 10:54 AM, Justin Sherrill wrote: > > >> On Wed, Mar 12, 2014 at 10:36 AM, Isaac (.ike) Levy wrote: >> Using commodity hardware and 10Gbit Intel nics, and FreeBSD 10.0, I'm pushing an actual 9.1-9.3Gbps (netperf tests), across 4x nics, no tuning. > > For what it's worth: > > http://lists.dragonflybsd.org/pipermail/users/2014-March/128240.html > > Not exactly the same hardware, but comparable. > > Something I've thought would be useful is some sort of test setup that would automatically run test cases like this as new BSD versions come in - not necessarily for comparison between types, but to show advancement or regressions between releases. That's wandering off to a whole new topic... Slammin' post, thanks- Monday sometime I'll followup with the exact same tests... -- Benchmarks are obviously complicated by context, but, I firmly agree that common bench-tests are immensely useful. Related, Pete Wright and others on list have discussed disk testing on list with similar aims... Best, .ike -------------- next part -------------- An HTML attachment was scrubbed... URL: From lists at eitanadler.com Thu Mar 13 13:02:02 2014 From: lists at eitanadler.com (Eitan Adler) Date: Thu, 13 Mar 2014 13:02:02 -0400 Subject: next meeting... In-Reply-To: <5321E0A1.9050208@ceetonetechnology.com> References: <5321E0A1.9050208@ceetonetechnology.com> Message-ID: On 13 March 2014 12:45, George Rosamond wrote: > ...just hit the announce@ list. > > Definitely spread the word as it should be a fascinating meeting. > Tweet, friend it, whatever you kids do today. > Will it be streamed as well? -- Eitan Adler From george at ceetonetechnology.com Thu Mar 13 13:07:17 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Mar 2014 13:07:17 -0400 Subject: next meeting... In-Reply-To: References: <5321E0A1.9050208@ceetonetechnology.com> Message-ID: <5321E5C5.7030806@ceetonetechnology.com> Eitan Adler: > On 13 March 2014 12:45, George Rosamond wrote: >> ...just hit the announce@ list. >> >> Definitely spread the word as it should be a fascinating meeting. >> Tweet, friend it, whatever you kids do today. >> > Will it be streamed as well? Hopefully... and video should be up. Patrick did a great job for the last meeting... hopefully we'll get the slides in a bit clearer. g From pvarga at pvrg.net Thu Mar 13 13:17:13 2014 From: pvarga at pvrg.net (pvarga at pvrg.net) Date: Thu, 13 Mar 2014 13:17:13 -0400 Subject: next meeting... In-Reply-To: <5321E0A1.9050208@ceetonetechnology.com> References: <5321E0A1.9050208@ceetonetechnology.com> Message-ID: <20140313171712.GA20842@pvrg.net> On Thu, Mar 13, 2014 at 12:45:21PM -0400, George Rosamond wrote: > ...just hit the announce@ list. > > Definitely spread the word as it should be a fascinating meeting. > Tweet, friend it, whatever you kids do today. > > Note that we will be meeting: > > at NYU.. not Suspenders > on Tuesday, April 1... not Wednesday April 2 > at 7:15 PM... not 6:45 PM. Is the subject the 'eating enthropy and have it too' paper? http://cryptome.org/2014/03/eat-entropy-have-it.pdf Please help me where and which campus of NYU and rsvp needed. > > g > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk From george at ceetonetechnology.com Thu Mar 13 13:27:32 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Mar 2014 13:27:32 -0400 Subject: next meeting... In-Reply-To: <20140313171712.GA20842@pvrg.net> References: <5321E0A1.9050208@ceetonetechnology.com> <20140313171712.GA20842@pvrg.net> Message-ID: <5321EA84.1010509@ceetonetechnology.com> pvarga at pvrg.net: > On Thu, Mar 13, 2014 at 12:45:21PM -0400, George Rosamond wrote: >> ...just hit the announce@ list. >> >> Definitely spread the word as it should be a fascinating meeting. >> Tweet, friend it, whatever you kids do today. >> >> Note that we will be meeting: >> >> at NYU.. not Suspenders >> on Tuesday, April 1... not Wednesday April 2 >> at 7:15 PM... not 6:45 PM. > > Is the subject the 'eating enthropy and have it too' paper? > > http://cryptome.org/2014/03/eat-entropy-have-it.pdf Hey Peter! Go ahead and compare to links for his papers on nycbug.org. > > Please help me where and which campus of NYU and rsvp needed. AFAIK right now, no RSVP needed. 251 Mercer street is address. . . g From george at ceetonetechnology.com Thu Mar 13 13:36:42 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Thu, 13 Mar 2014 13:36:42 -0400 Subject: announce@ list Message-ID: <5321ECAA.8090705@ceetonetechnology.com> If you're not on it for the (low volume) announces, you should be: http://www.nycbug.org/mailman/listinfo/announce And the archives are here: http://www.nycbug.org/pipermail/announce/ mailman is user-managed for accounts... don't both the list-owners! g From pvarga at pvrg.net Thu Mar 13 18:35:24 2014 From: pvarga at pvrg.net (pvarga at pvrg.net) Date: Thu, 13 Mar 2014 18:35:24 -0400 Subject: next meeting... In-Reply-To: <5321EA84.1010509@ceetonetechnology.com> References: <5321E0A1.9050208@ceetonetechnology.com> <20140313171712.GA20842@pvrg.net> <5321EA84.1010509@ceetonetechnology.com> Message-ID: <20140313223524.GA32493@pvrg.net> On Thu, Mar 13, 2014 at 01:27:32PM -0400, George Rosamond wrote: > pvarga at pvrg.net: > > On Thu, Mar 13, 2014 at 12:45:21PM -0400, George Rosamond wrote: > >> ...just hit the announce@ list. > >> > >> Definitely spread the word as it should be a fascinating meeting. > >> Tweet, friend it, whatever you kids do today. > >> > >> Note that we will be meeting: > >> > >> at NYU.. not Suspenders > >> on Tuesday, April 1... not Wednesday April 2 > >> at 7:15 PM... not 6:45 PM. > > > > Is the subject the 'eating enthropy and have it too' paper? > > > > http://cryptome.org/2014/03/eat-entropy-have-it.pdf > > Hey Peter! Go ahead and compare to links for his papers on nycbug.org. Thanks for the cluestick and pardon me for the noise. > > > > > Please help me where and which campus of NYU and rsvp needed. > > AFAIK right now, no RSVP needed. > > 251 Mercer street is address. . . > > g > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk From george at ceetonetechnology.com Fri Mar 14 10:52:59 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Fri, 14 Mar 2014 10:52:59 -0400 Subject: mid-town connectivity Message-ID: <532317CB.2080507@ceetonetechnology.com> Here's my problem: mid-town crappy connectivity. We have full control of building, roof, etc. TWC "business" class is supposed to be around 20m down, which only reflects in testing on the TWC test site. So add the usual lies and latency that their network bundles (for free :) and it's 5m with any other test site, wget, etc. The uptime is horrific for a 24/7 business. FIOS not available. The copper is crappy in the area. Verizon ethernet has a long lead time for installs, and is expensive. Towerstream WiMax is very expensive for comparable bandwidth. What are others using in midtown, between 34th and 42nd? Custom ATT fiber, etc? Why is this not a dead thread in 2014 NYC? If someone says ISDN I'll realize my clock is broken. g From mspitzer at gmail.com Fri Mar 14 11:29:38 2014 From: mspitzer at gmail.com (Marc Spitzer) Date: Fri, 14 Mar 2014 11:29:38 -0400 Subject: I declare not so stupid unix tricks thread open Message-ID: We need to share our not so stupid unix tricks. Here is one: ssh key access audit script: for i in $(seq 254) ; do x="192.168.49.$i"; echo -n "$x: " ; ssh -o PasswordAuthentication=no -o NumberOfPasswordPrompts=0 $x hostname ; this will give you one of 3 things: 1: hostname 2: failed login 3: unreachable error Easy way to find out where your keys work, who's next? 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 george at ceetonetechnology.com Fri Mar 14 11:35:13 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Fri, 14 Mar 2014 11:35:13 -0400 Subject: I declare not so stupid unix tricks thread open In-Reply-To: References: Message-ID: <532321B1.1010802@ceetonetechnology.com> Marc Spitzer: > We need to share our not so stupid unix tricks. Here is one: > > ssh key access audit script: > for i in $(seq 254) ; do x="192.168.49.$i"; echo -n "$x: " ; ssh -o > PasswordAuthentication=no -o NumberOfPasswordPrompts=0 $x hostname ; > > this will give you one of 3 things: > 1: hostname > 2: failed login > 3: unreachable error > > Easy way to find out where your keys work, who's next? Bravo. It shouldn't have taken us meeting in the data center for you to post ;P This was the idea behind the "share you tips with the community" meetings in the past. Simple stupid /bin/sh sh-it is best. Portable, interoperable, solving one problem. More please, sir. I choose Spork to share some knowledge now! g From mike at myownsoho.net Fri Mar 14 12:16:56 2014 From: mike at myownsoho.net (Mike N.) Date: Fri, 14 Mar 2014 12:16:56 -0400 Subject: mid-town connectivity In-Reply-To: <532317CB.2080507@ceetonetechnology.com> References: <532317CB.2080507@ceetonetechnology.com> Message-ID: On 2014-03-14 10:52, George Rosamond wrote: > Here's my problem: > > mid-town crappy connectivity. We have full control of building, roof, etc. > > TWC "business" class is supposed to be around 20m down, which only > reflects in testing on the TWC test site. So add the usual lies and > latency that their network bundles (for free :) and it's 5m with any > other test site, wget, etc. > > The uptime is horrific for a 24/7 business. > > FIOS not available. The copper is crappy in the area. Verizon ethernet > has a long lead time for installs, and is expensive. > > Towerstream WiMax is very expensive for comparable bandwidth. > > What are others using in midtown, between 34th and 42nd? Custom ATT > fiber, etc? > > Why is this not a dead thread in 2014 NYC? If someone says ISDN I'll > realize my clock is broken. > > g > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk [1] Go fer RCN Fiber. I had good experience working with them shortly after they took over Sidera 5-6 yrs ago. It's "custom" fiber, but pretty good rates. I remember looking at 50 over 15 at around $1500/mo. I can't imagine prices going up since then. Maybe other wireless providers are available to you? Rainbow pricing maybe better? Good luck! Links: ------ [1] http://www.nycbug.org/mailman/listinfo/talk -------------- next part -------------- An HTML attachment was scrubbed... URL: From spork at bway.net Fri Mar 14 12:43:14 2014 From: spork at bway.net (Charles Sprickman) Date: Fri, 14 Mar 2014 12:43:14 -0400 Subject: mid-town connectivity In-Reply-To: <532317CB.2080507@ceetonetechnology.com> References: <532317CB.2080507@ceetonetechnology.com> Message-ID: >From phone, apologies for top-post. We have 3 metro Ethernet providers as well as "EOC" (bonded symmetric dsl w/real cpe). Very flexible options as far as provisioning and supporting services. And of course, native IPv6. And incentives if you help get building mgmt onboard with dropping equipment in the telco closet. C Sent from my iDong On Mar 14, 2014, at 10:52 AM, George Rosamond wrote: > Here's my problem: > > mid-town crappy connectivity. We have full control of building, roof, etc. > > TWC "business" class is supposed to be around 20m down, which only > reflects in testing on the TWC test site. So add the usual lies and > latency that their network bundles (for free :) and it's 5m with any > other test site, wget, etc. > > The uptime is horrific for a 24/7 business. > > FIOS not available. The copper is crappy in the area. Verizon ethernet > has a long lead time for installs, and is expensive. > > Towerstream WiMax is very expensive for comparable bandwidth. > > What are others using in midtown, between 34th and 42nd? Custom ATT > fiber, etc? > > Why is this not a dead thread in 2014 NYC? If someone says ISDN I'll > realize my clock is broken. > > g > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk From pete at nomadlogic.org Fri Mar 14 12:57:23 2014 From: pete at nomadlogic.org (Pete Wright) Date: Fri, 14 Mar 2014 09:57:23 -0700 Subject: I declare not so stupid unix tricks thread open In-Reply-To: References: Message-ID: <532334F3.4030909@nomadlogic.org> On 03/14/14 08:29, Marc Spitzer wrote: > We need to share our not so stupid unix tricks. Here is one: > > ssh key access audit script: > for i in $(seq 254) ; do x="192.168.49.$i"; echo -n "$x: " ; ssh -o > PasswordAuthentication=no -o NumberOfPasswordPrompts=0 $x hostname ; > > this will give you one of 3 things: > 1: hostname > 2: failed login > 3: unreachable error > > Easy way to find out where your keys work, who's next? here's a stupid bash aliases i use to print out all hosts in a given DC/sub-domain: function h() { host -la "$@".MY.DOMAIN \ DNS0."$@".MY.DOMAIN|grep ^f|awk -F" " '{print $1}' \ | sed -e 's/\.$//g' ;} so if i want to see all my "web" nodes in my iad1 subdamon i'd run: $ h iad | grep web -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From jschauma at netmeister.org Fri Mar 14 13:29:32 2014 From: jschauma at netmeister.org (Jan Schaumann) Date: Fri, 14 Mar 2014 13:29:32 -0400 Subject: I declare not so stupid unix tricks thread open In-Reply-To: References: Message-ID: <20140314172932.GB4653@netmeister.org> Marc Spitzer wrote: > We need to share our not so stupid unix tricks. Some time ago I had started to collect a few of these here: https://twitter.com/rtfmsh Explanations here: https://www.netmeister.org/rtfmsh/ The one I use most regularly and find most useful is probably: https://www.netmeister.org/rtfmsh/302788758137352193.html ssh hostA "tar cf - foo" | ssh hostB "tar xf -" I should go back and collect a few more... :-) -Jan -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 478 bytes Desc: not available URL: From raulcuza at gmail.com Fri Mar 14 13:37:59 2014 From: raulcuza at gmail.com (Raul Cuza) Date: Fri, 14 Mar 2014 13:37:59 -0400 Subject: mid-town connectivity In-Reply-To: References: <532317CB.2080507@ceetonetechnology.com> Message-ID: On Fri, Mar 14, 2014 at 12:16 PM, Mike N. wrote: > On 2014-03-14 10:52, George Rosamond wrote: > > Here's my problem: > > mid-town crappy connectivity. We have full control of building, roof, etc. > > TWC "business" class is supposed to be around 20m down, which only > reflects in testing on the TWC test site. So add the usual lies and > latency that their network bundles (for free :) and it's 5m with any > other test site, wget, etc. > > The uptime is horrific for a 24/7 business. > > FIOS not available. The copper is crappy in the area. Verizon ethernet > has a long lead time for installs, and is expensive. > > Towerstream WiMax is very expensive for comparable bandwidth. > > What are others using in midtown, between 34th and 42nd? Custom ATT > fiber, etc? > > Why is this not a dead thread in 2014 NYC? If someone says ISDN I'll > realize my clock is broken. > > g > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > > Go fer RCN Fiber. I had good experience working with them shortly after > they took over Sidera 5-6 yrs ago. It's "custom" fiber, but pretty good > rates. I remember looking at 50 over 15 at around $1500/mo. I can't > imagine prices going up since then. > > Maybe other wireless providers are available to you? Rainbow pricing maybe > better? > > Good luck! > > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk My $work has RCN with great uptime (if you ignore having to yell at them to give us more than 4 hours notice on maintenance windows when we started with them). 250Mb symmetric for $2700/mon. Only caveat is we are by the water, so not technically midtown. Anyone use Panix in midtown? They don't service everywhere, but I know another place that got a great fiber deal from them; consider it a rumor though cause I can't remember the numbers. Post your solution please. Ra?l From pete at nomadlogic.org Fri Mar 14 14:08:24 2014 From: pete at nomadlogic.org (Pete Wright) Date: Fri, 14 Mar 2014 11:08:24 -0700 Subject: mid-town connectivity In-Reply-To: <532317CB.2080507@ceetonetechnology.com> References: <532317CB.2080507@ceetonetechnology.com> Message-ID: <53234598.8080206@nomadlogic.org> On 03/14/14 07:52, George Rosamond wrote: > Here's my problem: > > mid-town crappy connectivity. We have full control of building, roof, etc. > > TWC "business" class is supposed to be around 20m down, which only > reflects in testing on the TWC test site. So add the usual lies and > latency that their network bundles (for free :) and it's 5m with any > other test site, wget, etc. > > The uptime is horrific for a 24/7 business. > > FIOS not available. The copper is crappy in the area. Verizon ethernet > has a long lead time for installs, and is expensive. > > Towerstream WiMax is very expensive for comparable bandwidth. > > What are others using in midtown, between 34th and 42nd? Custom ATT > fiber, etc? > > Why is this not a dead thread in 2014 NYC? If someone says ISDN I'll > realize my clock is broken. > have you checked out zayo (aka aboveNet)? not sure what your budget is but they may already have dark fiber in your building - then you can try to partner up with a ISP (nyi maybe?) for connectivity or see what other IP services they offer: http://www.zayo.com/interactive-map/dark-fiber-map?zoom=12&mapCenter=-74.007118,40.714550&gclid=CLvljITOkr0CFSJo7AodJ0UAiQ i reckon this is on the way more expensive, not to mention labor intensive, way to go... also i remember back in the day con edison pretty much had every building outfitted with dark fiber, but i think they got out of the comms business a while ago. -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From jhellenthal at dataix.net Fri Mar 14 15:52:44 2014 From: jhellenthal at dataix.net (Jason Hellenthal) Date: Fri, 14 Mar 2014 15:52:44 -0400 Subject: I declare not so stupid unix tricks thread open In-Reply-To: References: Message-ID: These should all be posix compliant sh(1) FreeBSD compatible. Not sure about other systems. # Kill or list some stopped processes alias stopkill='kill `ps a -o stat,pid |awk '\''/^T /{print $2}'\''`' alias stoplist='ps -p `ps a -o stat,pid |awk '\''/^T /{print $2}'\''` 2>/dev/null' # Kill or list some zombies alias zomkill='kill `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''`' alias zomlist='ps -p `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''` 2>/dev/null' # Need a new 10 char[] password ? alias newpass='LANG=C tr -ucd [:alnum:][:punct:] wrote: > We need to share our not so stupid unix tricks. Here is one: > > ssh key access audit script: > for i in $(seq 254) ; do x="192.168.49.$i"; echo -n "$x: " ; ssh -o > PasswordAuthentication=no -o NumberOfPasswordPrompts=0 $x hostname ; > > this will give you one of 3 things: > 1: hostname > 2: failed login > 3: unreachable error > > Easy way to find out where your keys work, who's next? > > 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 > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > -------------- next part -------------- An HTML attachment was scrubbed... URL: From spork at bway.net Fri Mar 14 15:56:47 2014 From: spork at bway.net (Charles Sprickman) Date: Fri, 14 Mar 2014 15:56:47 -0400 Subject: mid-town connectivity In-Reply-To: <53234598.8080206@nomadlogic.org> References: <532317CB.2080507@ceetonetechnology.com> <53234598.8080206@nomadlogic.org> Message-ID: On Mar 14, 2014, at 2:08 PM, Pete Wright wrote: > also i remember back in the day con edison pretty much had every > building outfitted with dark fiber, but i think they got out of the > comms business a while ago. The network still exists, and other than a major issue with a spanning tree meltdown a few months back, it's been rock-solid - we're a layer 2 customer and we sell tail circuits through them. Detailed and well in advance notices of service-impacting and non-service-impacting maintenance, a very responsive NOC, and in most places a point of entry to the building that's on the opposite end of the telco stuff which is nice since backhoes avoid power lines better than telco lines. Sandy had no impact except for one unique location where we had a VZ tail circuit and something got wet that shouldn't have. Sidera was really big into the financial firms, and had some very high-priced offerings where they'd shave a few milliseconds or microseconds off the route. We obviously don't bother with those services, but we do share in the benefits when they're making hardware decisions. I see no evidence they grossly oversell any of the layer 2 services we buy. If I remember correctly, the chain of ownership goes like this: ConEd -> RCN -> Sidera -> Lightower (merger?) I can't speak about their IP transit services, we don't use them. We are legacy Level3 and Hurricane, with Level3 being preferred. We also get IPv6 transit from both upstreams. We are Cogent-free (with good reason). Charles > -pete > > > > -- > Pete Wright > pete at nomadlogic.org > twitter => @nomadlogicLA > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk From mikel.king at gmail.com Fri Mar 14 16:55:25 2014 From: mikel.king at gmail.com (Mikel King) Date: Fri, 14 Mar 2014 16:55:25 -0400 Subject: venue recommendations Message-ID: <098EC38C-1043-4DA9-8C61-5F387F06E8C7@gmail.com> If anyone knows of a decent venue(s) for an Open Source event that can support 6 class room like sessions of 75 - 125 attendees each (approximately 800 ppl total). This will be full day Saturday & full day Sunday in NYC either in Aug or Sept. Please ping me off list. Regards, mikel -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 801 bytes Desc: Message signed with OpenPGP using GPGMail URL: From mark.saad at ymail.com Fri Mar 14 17:41:49 2014 From: mark.saad at ymail.com (Mark Saad) Date: Fri, 14 Mar 2014 17:41:49 -0400 Subject: [talk] mid-town connectivity In-Reply-To: References: <532317CB.2080507@ceetonetechnology.com> <53234598.8080206@nomadlogic.org> Message-ID: <76B3F846-7B1B-4C9C-AF99-A07053988AB9@ymail.com> > On Mar 14, 2014, at 3:56 PM, Charles Sprickman wrote: > >> On Mar 14, 2014, at 2:08 PM, Pete Wright wrote: >> >> also i remember back in the day con edison pretty much had every >> building outfitted with dark fiber, but i think they got out of the >> comms business a while ago. > > The network still exists, and other than a major issue with a > spanning tree meltdown a few months back, it's been rock-solid - > we're a layer 2 customer and we sell tail circuits through them. > Detailed and well in advance notices of service-impacting and > non-service-impacting maintenance, a very responsive NOC, and in > most places a point of entry to the building that's on the opposite > end of the telco stuff which is nice since backhoes avoid power > lines better than telco lines. Sandy had no impact except for one > unique location where we had a VZ tail circuit and something got wet > that shouldn't have. > > Sidera was really big into the financial firms, and had some very > high-priced offerings where they'd shave a few milliseconds or > microseconds off the route. We obviously don't bother with those > services, but we do share in the benefits when they're making > hardware decisions. I see no evidence they grossly oversell any of > the layer 2 services we buy. > > If I remember correctly, the chain of ownership goes like this: > > ConEd -> RCN -> Sidera -> Lightower (merger?) > Coned CEC merged with RCN , RCN metro networks split from RCN and merged with another transit provider and became Sidera . Sidera merged with Lightower . I work in a lightower facility , as a customer . > I can't speak about their IP transit services, we don't use them. > We are legacy Level3 and Hurricane, with Level3 being preferred. We > also get IPv6 transit from both upstreams. We are Cogent-free (with > good reason). > > Charles > >> -pete >> >> At $WORK we use two sets of mux'd ds3 and t1 links . The sets are from cogent and Verizon iirc it's 20 t1 mux'd to one 100m ethernet handoff and for the ds3's it's 4 mux'd into a gig ethernet handoff . Verizon took a while To get it up and running . Cogent was faster but had to use verizon for in building handoffs . We also have a private point to point fiber set from lightower and they have been good at making things happen , however we do not use them for ip services . So what's it like : It's good enough that the c levels down to everyone but network operations likes the speed and reliability . I have no idea what the cost is but we have fairly deep pockets . --- Mark From george at ceetonetechnology.com Fri Mar 14 19:00:03 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Fri, 14 Mar 2014 19:00:03 -0400 Subject: [talk] mid-town connectivity In-Reply-To: <76B3F846-7B1B-4C9C-AF99-A07053988AB9@ymail.com> References: <532317CB.2080507@ceetonetechnology.com> <53234598.8080206@nomadlogic.org> <76B3F846-7B1B-4C9C-AF99-A07053988AB9@ymail.com> Message-ID: <532389F3.5030201@ceetonetechnology.com> Mark Saad: > > >> On Mar 14, 2014, at 3:56 PM, Charles Sprickman wrote: >> >>> On Mar 14, 2014, at 2:08 PM, Pete Wright wrote: >>> >>> also i remember back in the day con edison pretty much had every >>> building outfitted with dark fiber, but i think they got out of the >>> comms business a while ago. >> >> The network still exists, and other than a major issue with a >> spanning tree meltdown a few months back, it's been rock-solid - >> we're a layer 2 customer and we sell tail circuits through them. >> Detailed and well in advance notices of service-impacting and >> non-service-impacting maintenance, a very responsive NOC, and in >> most places a point of entry to the building that's on the opposite >> end of the telco stuff which is nice since backhoes avoid power >> lines better than telco lines. Sandy had no impact except for one >> unique location where we had a VZ tail circuit and something got wet >> that shouldn't have. >> >> Sidera was really big into the financial firms, and had some very >> high-priced offerings where they'd shave a few milliseconds or >> microseconds off the route. We obviously don't bother with those >> services, but we do share in the benefits when they're making >> hardware decisions. I see no evidence they grossly oversell any of >> the layer 2 services we buy. >> >> If I remember correctly, the chain of ownership goes like this: >> >> ConEd -> RCN -> Sidera -> Lightower (merger?) >> > Coned CEC merged with RCN , RCN metro networks split from RCN and merged with another transit provider and became Sidera . Sidera merged with Lightower . I work in a lightower facility , as a customer . > > >> I can't speak about their IP transit services, we don't use them. >> We are legacy Level3 and Hurricane, with Level3 being preferred. We >> also get IPv6 transit from both upstreams. We are Cogent-free (with >> good reason). >> >> Charles >> >>> -pete >>> >>> > > At $WORK we use two sets of mux'd ds3 and t1 links . The sets are from cogent and Verizon iirc it's 20 t1 mux'd to one 100m ethernet handoff and for the ds3's it's 4 mux'd into a gig ethernet handoff . Verizon took a while To get it up and running . Cogent was faster but had to use verizon for in building handoffs . We also have a private point to point fiber set from lightower and they have been good at making things happen , however we do not use them for ip services . > > So what's it like : It's good enough that the c levels down to everyone but network operations likes the speed and reliability . I have no idea what the cost is but we have fairly deep pockets . > Thanks for all the replies. Good stuff. I browsed through them as they came in, but will work through more thoroughly tonight. It just kills me that if we were in SK, JP and a bunch of other places, such a post would be pointless. I'd have ipv6 with 30m synchronous for $19.95 a month myself. It's just a reminder of how infrastructure is sooo broken in this city. I could have made the same post ten or 15 years ago. Very appreciated for the input though... g From matt at atopia.net Fri Mar 14 19:05:49 2014 From: matt at atopia.net (Matt Juszczak) Date: Fri, 14 Mar 2014 23:05:49 +0000 (UTC) Subject: [talk] BSDJobs Message-ID: Hi all, As many of you know, I used to run BSDJobs.net, and it used to be used realtively often. I let the site slack a bit, unfortunately, and it's pretty much non-existant at this point. However, it seems there still might be a place for it as a community resource, and I wanted to gauge some opinions. 1) If anyone would like to take over the project, and is serious about bringing the site back, I'll happily transfer the domain to them. 2) Alternatively, I can bring the site back on top of Wordpress and a Job Board theme provided folks feel there's still a use for it. Would love some feedback. Have a good weekend! Best, Matt From george at ceetonetechnology.com Fri Mar 14 19:26:23 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Fri, 14 Mar 2014 19:26:23 -0400 Subject: [talk] I declare not so stupid unix tricks thread open In-Reply-To: References: Message-ID: <5323901F.8020508@ceetonetechnology.com> Jason Hellenthal: > These should all be posix compliant sh(1) FreeBSD compatible. Not sure > about other systems. > > # Kill or list some stopped processes > alias stopkill='kill `ps a -o stat,pid |awk '\''/^T /{print $2}'\''`' > alias stoplist='ps -p `ps a -o stat,pid |awk '\''/^T /{print $2}'\''` > 2>/dev/null' > > # Kill or list some zombies > alias zomkill='kill `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''`' > alias zomlist='ps -p `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''` > 2>/dev/null' Some great stuff here. We need to have a marathon meeting in which 10 or more people do a one-liner each.. and others can critique, expand, etc. Say, June? Who's in? g From md at mailq.de Sat Mar 15 05:38:04 2014 From: md at mailq.de (Mischa Diehm) Date: Sat, 15 Mar 2014 10:38:04 +0100 Subject: [talk] I declare not so stupid unix tricks thread open In-Reply-To: <5323901F.8020508@ceetonetechnology.com> References: <5323901F.8020508@ceetonetechnology.com> Message-ID: On systems that have lsof (could probably be changed for fstat): # fast changedir to directories already open by other shells fcd () { COUNT=0 lsof -d cwd | grep '^bash' | sed -e 's/^.* \///' | \ sort -u | grep -v '^home/mischa$' | while read i; do COUNT=$((COUNT+1)); echo "[$COUNT] /$i"; done read j; case $j in [1-9]) DIR="/$(lsof -d cwd | grep '^bash' | sed -e 's/^.* \///' | sort -u | \ grep -v '^home/mischa$' | sed -n ${j}p)" if [ "$DIR" != "/" ]; then cd "$DIR" fi ;; *) ;; esac } [10:23:03] mischa at lebenske:include$ fcd [1] /Users/mischa [2] /Users/mischa/bin [3] /Users/mischa/x40 [4] /usr/local/Cellar/yubikey-personalization [5] /usr/local/include 4 [10:23:14] mischa at lebenske:yubikey-personalization$ == #local manpages lman () { nroff -e -man "$1" | less -s } [10:34:50] mischa at lebenske:tek$ lman plan9/sys/man/8/wol == #jump to search with vim vipat () { vim -R +/$1 `egrep -Rl $1 .` } cheers, Mischa On 15.03.2014, at 00:26, George Rosamond wrote: > Jason Hellenthal: >> These should all be posix compliant sh(1) FreeBSD compatible. Not sure >> about other systems. >> >> # Kill or list some stopped processes >> alias stopkill='kill `ps a -o stat,pid |awk '\''/^T /{print $2}'\''`' >> alias stoplist='ps -p `ps a -o stat,pid |awk '\''/^T /{print $2}'\''` >> 2>/dev/null' >> >> # Kill or list some zombies >> alias zomkill='kill `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''`' >> alias zomlist='ps -p `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''` >> 2>/dev/null' > > > Some great stuff here. We need to have a marathon meeting in which 10 > or more people do a one-liner each.. and others can critique, expand, etc. > > Say, June? Who's in? > > g > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > From freebsd-listen at fabiankeil.de Sat Mar 15 09:38:23 2014 From: freebsd-listen at fabiankeil.de (Fabian Keil) Date: Sat, 15 Mar 2014 14:38:23 +0100 Subject: [talk] I declare not so stupid unix tricks thread open In-Reply-To: References: Message-ID: <20140315143823.1ff4a188@fabiankeil.de> Marc Spitzer wrote: > We need to share our not so stupid unix tricks. Here is one: Here's another one (and I demand extra points because even the name is stupid and misleading): explain_type() { local type="$1" if [ -z "${type}" ]; then echo "No type given, dude" return 1 fi sudo dtrace -qn 'BEGIN {p=('"${type}"'*)alloca(sizeof('"${type}"')); print(*p);exit(0)}' } fk at r500 ~ $explain_type "struct g_geom" struct g_geom { char *name = 0 struct g_class *class = 0 struct geom = { struct g_geom *le_next = 0 struct g_geom **le_prev = 0 } struct consumer = { struct g_consumer *lh_first = 0 } struct provider = { struct g_provider *lh_first = 0 } struct geoms = { struct g_geom *tqe_next = 0 struct g_geom **tqe_prev = 0 } int rank = 0 g_start_t *start = 0 g_spoiled_t *spoiled = 0 g_attrchanged_t *attrchanged = 0 g_dumpconf_t *dumpconf = 0 g_access_t *access = 0 g_orphan_t *orphan = 0 g_ioctl_t *ioctl = 0 g_provgone_t *providergone = 0 g_resize_t *resize = 0 void *spare0 = 0 void *spare1 = 0 void *softc = 0 unsigned int flags = 0 } BTW, is NYCBUG interested in supporting a protest weekend against mass surveillance on April 12th in Cologne, Germany? http://cologne.stopwatchingus.info/demo-12-april/en.html There may even be a DTrace (and thus BSD) connection: http://www.listbox.com/member/archive/184261/2014/03/sort/time_rev/page/1/entry/0:16/20140314025218:46D0A7E8-AB45-11E3-AE09-FF17A2C259A7/ Please don't book your flights prematurely, though ... Fabian -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 196 bytes Desc: not available URL: From crossd at gmail.com Sat Mar 15 10:27:25 2014 From: crossd at gmail.com (Dan Cross) Date: Sat, 15 Mar 2014 10:27:25 -0400 Subject: [talk] I declare not so stupid unix tricks thread open In-Reply-To: References: Message-ID: On Fri, Mar 14, 2014 at 3:52 PM, Jason Hellenthal wrote: > These should all be posix compliant sh(1) FreeBSD compatible. Not sure > about other systems. > > # Kill or list some stopped processes > alias stopkill='kill `ps a -o stat,pid |awk '\''/^T /{print $2}'\''`' > alias stoplist='ps -p `ps a -o stat,pid |awk '\''/^T /{print $2}'\''` > 2>/dev/null' > > # Kill or list some zombies > alias zomkill='kill `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''`' > This one doesn't make a lot of sense to me. You can't kill a zombie; zombies are by definition already dead, and just hanging around until their parent wait()'s for their exit status (or the parent dies, in which case they're inherited by init, which will wait for them unless your system is totally messed up). The only resource a zombie occupies is a slot in your process table. > alias zomlist='ps -p `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''` > 2>/dev/null' > > # Need a new 10 char[] password ? > alias newpass='LANG=C tr -ucd [:alnum:][:punct:] OZkl |tr -us [:alnum:][:punct:] |head -c 10 && echo' > > # cisco Systems like ping(1) > alias cping='sudo /sbin/ping -n -c 5 -t 2 -i 0.1 -s 72' > > # Sweep through sizes when ping'ing a host > alias sweep='ping -D -G 56 -h 7' > > # Need some specific permissions of files devices or directories ? > alias permof='stat -f "%p %Sp %Su %Sg %N"' > > # List major minor and type of a device ? > alias lsdev='stat -f > "%nName:%t%N%nType:%t%HT%nMajor:%t%Hr%nMinor:%t%Lr%n%n"' > > > On Fri, Mar 14, 2014 at 11:29 AM, Marc Spitzer wrote: > >> We need to share our not so stupid unix tricks. Here is one: >> >> ssh key access audit script: >> for i in $(seq 254) ; do x="192.168.49.$i"; echo -n "$x: " ; ssh -o >> PasswordAuthentication=no -o NumberOfPasswordPrompts=0 $x hostname ; >> >> this will give you one of 3 things: >> 1: hostname >> 2: failed login >> 3: unreachable error >> >> Easy way to find out where your keys work, who's next? >> >> 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 >> >> _______________________________________________ >> talk mailing list >> talk at lists.nycbug.org >> http://www.nycbug.org/mailman/listinfo/talk >> > > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > -------------- next part -------------- An HTML attachment was scrubbed... URL: From jhellenthal at dataix.net Sat Mar 15 11:25:05 2014 From: jhellenthal at dataix.net (Jason Hellenthal) Date: Sat, 15 Mar 2014 11:25:05 -0400 Subject: [talk] I declare not so stupid unix tricks thread open In-Reply-To: References: Message-ID: Lol I was waiting for that. But a attempted head(1) shot is worthwhile anyway ! I had some process's that actually took the headshot and exited. Always worth a shot. -- Jason Hellenthal Voice: 95.30.17.6/616 JJH48-ARIN > On Mar 15, 2014, at 10:27, Dan Cross wrote: > >> On Fri, Mar 14, 2014 at 3:52 PM, Jason Hellenthal wrote: >> These should all be posix compliant sh(1) FreeBSD compatible. Not sure about other systems. >> >> # Kill or list some stopped processes >> alias stopkill='kill `ps a -o stat,pid |awk '\''/^T /{print $2}'\''`' >> alias stoplist='ps -p `ps a -o stat,pid |awk '\''/^T /{print $2}'\''` 2>/dev/null' >> >> # Kill or list some zombies >> alias zomkill='kill `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''`' > > This one doesn't make a lot of sense to me. You can't kill a zombie; zombies are by definition already dead, and just hanging around until their parent wait()'s for their exit status (or the parent dies, in which case they're inherited by init, which will wait for them unless your system is totally messed up). The only resource a zombie occupies is a slot in your process table. > >> alias zomlist='ps -p `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''` 2>/dev/null' >> >> # Need a new 10 char[] password ? >> alias newpass='LANG=C tr -ucd [:alnum:][:punct:] > >> # cisco Systems like ping(1) >> alias cping='sudo /sbin/ping -n -c 5 -t 2 -i 0.1 -s 72' >> >> # Sweep through sizes when ping'ing a host >> alias sweep='ping -D -G 56 -h 7' >> >> # Need some specific permissions of files devices or directories ? >> alias permof='stat -f "%p %Sp %Su %Sg %N"' >> >> # List major minor and type of a device ? >> alias lsdev='stat -f "%nName:%t%N%nType:%t%HT%nMajor:%t%Hr%nMinor:%t%Lr%n%n"' >> >> >>> On Fri, Mar 14, 2014 at 11:29 AM, Marc Spitzer wrote: >>> We need to share our not so stupid unix tricks. Here is one: >>> >>> ssh key access audit script: >>> for i in $(seq 254) ; do x="192.168.49.$i"; echo -n "$x: " ; ssh -o PasswordAuthentication=no -o NumberOfPasswordPrompts=0 $x hostname ; >>> >>> this will give you one of 3 things: >>> 1: hostname >>> 2: failed login >>> 3: unreachable error >>> >>> Easy way to find out where your keys work, who's next? >>> >>> 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 >>> >>> _______________________________________________ >>> talk mailing list >>> talk at lists.nycbug.org >>> http://www.nycbug.org/mailman/listinfo/talk >> >> >> _______________________________________________ >> talk mailing list >> talk at lists.nycbug.org >> http://www.nycbug.org/mailman/listinfo/talk > -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 6118 bytes Desc: not available URL: From justin at shiningsilence.com Sat Mar 15 12:11:23 2014 From: justin at shiningsilence.com (Justin Sherrill) Date: Sat, 15 Mar 2014 12:11:23 -0400 Subject: [talk] BSDJobs In-Reply-To: References: Message-ID: I don't have enough time to work on it, but if you do bring it back, please let me know - especially if there's an RSS feed. I'd like to note any job postings on the DragonFly Digest. On Fri, Mar 14, 2014 at 7:05 PM, Matt Juszczak wrote: > Hi all, > > As many of you know, I used to run BSDJobs.net, and it used to be used > realtively often. I let the site slack a bit, unfortunately, and it's > pretty much non-existant at this point. However, it seems there still > might be a place for it as a community resource, and I wanted to gauge some > opinions. > > 1) If anyone would like to take over the project, and is serious about > bringing the site back, I'll happily transfer the domain to them. > > 2) Alternatively, I can bring the site back on top of Wordpress and a Job > Board theme provided folks feel there's still a use for it. > > Would love some feedback. Have a good weekend! > > Best, > > Matt > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk > -------------- next part -------------- An HTML attachment was scrubbed... URL: From jhellenthal at dataix.net Sat Mar 15 18:14:28 2014 From: jhellenthal at dataix.net (Jason Hellenthal) Date: Sat, 15 Mar 2014 18:14:28 -0400 Subject: [talk] I declare not so stupid unix tricks thread open In-Reply-To: <5323901F.8020508@ceetonetechnology.com> References: <5323901F.8020508@ceetonetechnology.com> Message-ID: <9D3DF34A-F790-4FF9-A346-4204B8ECA041@dataix.net> I'm in if I'm in that area around that time. I have fun dropping this stuff on a dime and would be a fun meeting IMHO -- Jason Hellenthal Voice: 95.30.17.6/616 JJH48-ARIN > On Mar 14, 2014, at 19:26, George Rosamond wrote: > > Jason Hellenthal: >> These should all be posix compliant sh(1) FreeBSD compatible. Not sure >> about other systems. >> >> # Kill or list some stopped processes >> alias stopkill='kill `ps a -o stat,pid |awk '\''/^T /{print $2}'\''`' >> alias stoplist='ps -p `ps a -o stat,pid |awk '\''/^T /{print $2}'\''` >> 2>/dev/null' >> >> # Kill or list some zombies >> alias zomkill='kill `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''`' >> alias zomlist='ps -p `ps a -o stat,pid |awk '\''/^Z /{print $2}'\''` >> 2>/dev/null' > > > Some great stuff here. We need to have a marathon meeting in which 10 > or more people do a one-liner each.. and others can critique, expand, etc. > > Say, June? Who's in? > > g > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 6118 bytes Desc: not available URL: From zippy1981 at gmail.com Sun Mar 16 14:27:00 2014 From: zippy1981 at gmail.com (Justin Dearing) Date: Sun, 16 Mar 2014 14:27:00 -0400 Subject: [talk] Thoughts on TinySSH? Message-ID: http://tinyssh.org/index.html Someone is making a tiny ssh server without using malloc (pure static memory analysis). Its not supporting ssh1, sftp or scp and not supporting AES or DES.They're expecting an alpha in 2015 and a beta in 2016. Some of my thoughts: - Two years seems a little long to reimplement ssh. However, I don't know enough about ssh internals to comment - I don't see the source code on his site, just directions to download a deb. - Even if all the memory is statically allocated, isn't it still potentially vulnerable to pointer math errors? I'll defer to those who actively write C to tell me otherwise. - Its a server, so saying it only supports newer encryption protocols is ok. As something for embedded devices, this is an ok design decision. - No SFTP or SCP support is questionable. SCP as a payload delivery mechanism would be useful, but perhaps that can be added later. - If this code can compile on windows without cygwin, that would be an awesome win. However, its limited scope means there's little chance its going to support Kerberos authentication. Anyone else have any thoughts? -------------- next part -------------- An HTML attachment was scrubbed... URL: From mark.saad at ymail.com Tue Mar 18 19:51:10 2014 From: mark.saad at ymail.com (Mark Saad) Date: Tue, 18 Mar 2014 19:51:10 -0400 Subject: [talk] mid-town connectivity In-Reply-To: <532389F3.5030201@ceetonetechnology.com> References: <532317CB.2080507@ceetonetechnology.com> <53234598.8080206@nomadlogic.org> <76B3F846-7B1B-4C9C-AF99-A07053988AB9@ymail.com> <532389F3.5030201@ceetonetechnology.com> Message-ID: <0CC6D932-66CC-479E-8041-937BBD3E5468@ymail.com> --- > On Mar 14, 2014, at 7:00 PM, George Rosamond wrote: > > Mark Saad: >> >> >>>> On Mar 14, 2014, at 3:56 PM, Charles Sprickman wrote: >>>> >>>> On Mar 14, 2014, at 2:08 PM, Pete Wright wrote: >>>> >>>> also i remember back in the day con edison pretty much had every >>>> building outfitted with dark fiber, but i think they got out of the >>>> comms business a while ago. >>> >>> The network still exists, and other than a major issue with a >>> spanning tree meltdown a few months back, it's been rock-solid - >>> we're a layer 2 customer and we sell tail circuits through them. >>> Detailed and well in advance notices of service-impacting and >>> non-service-impacting maintenance, a very responsive NOC, and in >>> most places a point of entry to the building that's on the opposite >>> end of the telco stuff which is nice since backhoes avoid power >>> lines better than telco lines. Sandy had no impact except for one >>> unique location where we had a VZ tail circuit and something got wet >>> that shouldn't have. >>> >>> Sidera was really big into the financial firms, and had some very >>> high-priced offerings where they'd shave a few milliseconds or >>> microseconds off the route. We obviously don't bother with those >>> services, but we do share in the benefits when they're making >>> hardware decisions. I see no evidence they grossly oversell any of >>> the layer 2 services we buy. >>> >>> If I remember correctly, the chain of ownership goes like this: >>> >>> ConEd -> RCN -> Sidera -> Lightower (merger?) >> Coned CEC merged with RCN , RCN metro networks split from RCN and merged with another transit provider and became Sidera . Sidera merged with Lightower . I work in a lightower facility , as a customer . >> >> >>> I can't speak about their IP transit services, we don't use them. >>> We are legacy Level3 and Hurricane, with Level3 being preferred. We >>> also get IPv6 transit from both upstreams. We are Cogent-free (with >>> good reason). >>> >>> Charles >>> >>>> -pete >> >> At $WORK we use two sets of mux'd ds3 and t1 links . The sets are from cogent and Verizon iirc it's 20 t1 mux'd to one 100m ethernet handoff and for the ds3's it's 4 mux'd into a gig ethernet handoff . Verizon took a while To get it up and running . Cogent was faster but had to use verizon for in building handoffs . We also have a private point to point fiber set from lightower and they have been good at making things happen , however we do not use them for ip services . >> >> So what's it like : It's good enough that the c levels down to everyone but network operations likes the speed and reliability . I have no idea what the cost is but we have fairly deep pockets . > > Thanks for all the replies. Good stuff. I browsed through them as they > came in, but will work through more thoroughly tonight. > > It just kills me that if we were in SK, JP and a bunch of other places, > such a post would be pointless. I'd have ipv6 with 30m synchronous for > $19.95 a month myself. It's just a reminder of how infrastructure is > sooo broken in this city. I could have made the same post ten or 15 > years ago. > > Very appreciated for the input though... > > g So I don't like the comparison of Asia to North America . Japan and South Korea ( a/k/a the republic of samsung ) have been in a tech arms race for so long it's not even interesting any more . They can do things there we will probably not see for various reasons . * What I mean is their offerings are market , and infrastructure driven . We have monopolies who control vast parts of the infrastructure and fight to keep it with in their bounds for various reasons . ConEd Laid down so much fiber it's unreal and they are a public utility . Verizon owns so much of the remaining infrastructure most "providers" have to piggy back on verizon gear or use Verizon's right aways to get you service . What it boils down to is if you want to provide network services in NYC you have to do it with out invading the market space of the providers you have to use to get customers service . * slight rant but keep this in mind . 4k tv / video is the new big deal . Popular is Asia and Europe the requirements to deliver 4k content is almost unbelievable . Scott long made reference to this at the con; i can't remember if he said it was an order of magnitude greater the 1080p or not but it was huge none the less . In any case Scott stated , most people in North America can't see netflix's 4k content , not due to lacy of tech but due to network providers infighting on costs of delivery ; and the subtext is also do to the lack of customer complaints . So while we complain that we can't get fast network service in NYC the reality is our market doesn't need or want it yet . No provider really wants to take on Verizon's fios or comcast / Time Warner , not until enough customers complain that they can't get 4k tv at home . Then they will upgrade their networks . Also do you really need to see that much detail , with 1080p I can tell the fox 5 news guy has nasty razor bumps ,so at 4k can I see into his soul ? Mark > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://www.nycbug.org/mailman/listinfo/talk From gnn at neville-neil.com Tue Mar 18 20:58:56 2014 From: gnn at neville-neil.com (George Neville-Neil) Date: Tue, 18 Mar 2014 20:58:56 -0400 Subject: [talk] mid-town connectivity In-Reply-To: <0CC6D932-66CC-479E-8041-937BBD3E5468@ymail.com> References: <532317CB.2080507@ceetonetechnology.com> <53234598.8080206@nomadlogic.org> <76B3F846-7B1B-4C9C-AF99-A07053988AB9@ymail.com> <532389F3.5030201@ceetonetechnology.com> <0CC6D932-66CC-479E-8041-937BBD3E5468@ymail.com> Message-ID: <190CD31B-E735-403D-97FF-2A67B484A5DC@neville-neil.com> Mar 18, 2014 19:51?Mark Saad ??????: > > > --- > > >> On Mar 14, 2014, at 7:00 PM, George Rosamond wrote: >> >> Mark Saad: >>> >>> >>>>> On Mar 14, 2014, at 3:56 PM, Charles Sprickman wrote: >>>>> >>>>> On Mar 14, 2014, at 2:08 PM, Pete Wright wrote: >>>>> >>>>> also i remember back in the day con edison pretty much had every >>>>> building outfitted with dark fiber, but i think they got out of the >>>>> comms business a while ago. >>>> >>>> The network still exists, and other than a major issue with a >>>> spanning tree meltdown a few months back, it's been rock-solid - >>>> we're a layer 2 customer and we sell tail circuits through them. >>>> Detailed and well in advance notices of service-impacting and >>>> non-service-impacting maintenance, a very responsive NOC, and in >>>> most places a point of entry to the building that's on the opposite >>>> end of the telco stuff which is nice since backhoes avoid power >>>> lines better than telco lines. Sandy had no impact except for one >>>> unique location where we had a VZ tail circuit and something got wet >>>> that shouldn't have. >>>> >>>> Sidera was really big into the financial firms, and had some very >>>> high-priced offerings where they'd shave a few milliseconds or >>>> microseconds off the route. We obviously don't bother with those >>>> services, but we do share in the benefits when they're making >>>> hardware decisions. I see no evidence they grossly oversell any of >>>> the layer 2 services we buy. >>>> >>>> If I remember correctly, the chain of ownership goes like this: >>>> >>>> ConEd -> RCN -> Sidera -> Lightower (merger?) >>> Coned CEC merged with RCN , RCN metro networks split from RCN and merged with another transit provider and became Sidera . Sidera merged with Lightower . I work in a lightower facility , as a customer . >>> >>> >>>> I can't speak about their IP transit services, we don't use them. >>>> We are legacy Level3 and Hurricane, with Level3 being preferred. We >>>> also get IPv6 transit from both upstreams. We are Cogent-free (with >>>> good reason). >>>> >>>> Charles >>>> >>>>> -pete >>> >>> At $WORK we use two sets of mux'd ds3 and t1 links . The sets are from cogent and Verizon iirc it's 20 t1 mux'd to one 100m ethernet handoff and for the ds3's it's 4 mux'd into a gig ethernet handoff . Verizon took a while To get it up and running . Cogent was faster but had to use verizon for in building handoffs . We also have a private point to point fiber set from lightower and they have been good at making things happen , however we do not use them for ip services . >>> >>> So what's it like : It's good enough that the c levels down to everyone but network operations likes the speed and reliability . I have no idea what the cost is but we have fairly deep pockets . >> >> Thanks for all the replies. Good stuff. I browsed through them as they >> came in, but will work through more thoroughly tonight. >> >> It just kills me that if we were in SK, JP and a bunch of other places, >> such a post would be pointless. I'd have ipv6 with 30m synchronous for >> $19.95 a month myself. It's just a reminder of how infrastructure is >> sooo broken in this city. I could have made the same post ten or 15 >> years ago. >> >> Very appreciated for the input though... >> >> g > > > So I don't like the comparison of Asia to North America . Japan and South Korea ( a/k/a the republic of samsung ) have been in a tech arms race for so long it's not even interesting any more . They can do things there we will probably not see for various reasons . * What I mean is their offerings are market , and infrastructure driven . We have monopolies who control vast parts of the infrastructure and fight to keep it with in their bounds for various reasons . ConEd Laid down so much fiber it's unreal and they are a public utility . Verizon owns so much of the remaining infrastructure most "providers" have to piggy back on verizon gear or use Verizon's right aways to get you service . What it boils down to is if you want to provide network services in NYC you have to do it with out invading the market space of the providers you have to use to get customers service . NTT is a monopoly. In the 80s they laid fiber all over Japan. Why? Because the Japanese believe in infrastructure. They are now reaping the benefits of that investment. I can get more bandwidth, cheaper, up a mountain in Japan than I can in much of the US including NYC. Americans do not believe in sharing in this way. Note cars vs trains for example. The ways in which Verizon and the like have twisted the market scum but they can do it because Americans don't travel and so don't know anything better and they do not believe in infrastructure as they once did. Sorry, I just came back from Japan and there are plenty of criticisms I can make of the Japanese ( xenophobia, racism, sexism and homophobia to start) but a belief in sharing for the common good of the Japanese and a belief in building your way to the future are common cultural traits and that's why my apartment building had a 1g fiber (mentioned in the ad BTW) in 2006. > > * slight rant but keep this in mind . 4k tv / video is the new big deal . Popular is Asia and Europe the requirements to deliver 4k content is almost unbelievable . Scott long made reference to this at the con; i can't remember if he said it was an order of magnitude greater the 1080p or not but it was huge none the less . In any case Scott stated , most people in North America can't see netflix's 4k content , not due to lacy of tech but due to network providers infighting on costs of delivery ; and the subtext is also do to the lack of customer complaints . So while we complain that we can't get fast network service in NYC the reality is our market doesn't need or want it yet . No provider really wants to take on Verizon's fios or comcast / Time Warner , not until enough customers complain that they can't get 4k tv at home . Then they will upgrade their networks . > > > Also do you really need to see that much detail , with 1080p I can tell the fox 5 news guy has nasty razor bumps ,so at 4k can I see into his soul ? > If he had one.... I don't have or watch TV but I want 4K TV because I want to live in the future and in the future we get more stuff faster. End of my rant :-) Best George From pete at nomadlogic.org Wed Mar 19 13:39:01 2014 From: pete at nomadlogic.org (Pete Wright) Date: Wed, 19 Mar 2014 10:39:01 -0700 Subject: [talk] mid-town connectivity In-Reply-To: <0CC6D932-66CC-479E-8041-937BBD3E5468@ymail.com> References: <532317CB.2080507@ceetonetechnology.com> <53234598.8080206@nomadlogic.org> <76B3F846-7B1B-4C9C-AF99-A07053988AB9@ymail.com> <532389F3.5030201@ceetonetechnology.com> <0CC6D932-66CC-479E-8041-937BBD3E5468@ymail.com> Message-ID: <5329D635.6060208@nomadlogic.org> On 03/18/14 16:51, Mark Saad wrote: > * slight rant but keep this in mind . 4k tv / video is the new big deal . Popular is Asia and Europe the requirements to deliver 4k content is almost unbelievable . Scott long made reference to this at the con; i can't remember if he said it was an order of magnitude greater the 1080p or not but it was huge none the less . In any case Scott stated , most people in North America can't see netflix's 4k content , not due to lacy of tech but due to network providers infighting on costs of delivery ; and the subtext is also do to the lack of customer complaints . So while we complain that we can't get fast network service in NYC the reality is our market doesn't need or want it yet . No provider really wants to take on Verizon's fios or comcast / Time Warner , not until enough customers complain that they can't get 4k tv at home . Then they will upgrade their networks . > > > Also do you really need to see that much detail , with 1080p I can tell the fox 5 news guy has nasty razor bumps ,so at 4k can I see into his soul ? > > (putting on my postproduction/vfx hat for a sec) just to put this in context - here are the data stream sizes for various film formats "HD" film works out like so: Aspect Ratio: 1920 x 1080 Frame Rate : 24fps (common for film) Uncompressed Frame filesize: ~3MB Uncompressed bitrate: ~150MB/s MPEG2 bitrate: ~10MB/s interestingly enough if you get your HD TV feed over the air via "rabbit ears" more likely than not you are actually receiving an uncompressed stream, I am not sure what codec Netflix uses - but let's assume it is close to MPEG2 (actually it's probably MPEG4 but close enough for our example ;) Now lets look at 4k: Aspect Ratio: 4096 x 2160 (hence "4k") Frame Rate: 24fps (again common for film) Uncompressed Frame filesize: ~13MB Uncompressed bitrate: ~637MB MGPEG2 bitrate: ~38MB/s At least in the film world that is quite a big jump is bit rates from "HD" to 4k. Also remember, this is streaming data so an ISP needs to be able to sustain ~40mb/s to deliver 4k video to an end user. I think this is what scares most consumer ISP's in the states - they have invested in IP over cable which has issues with shared bandwidth among users in a neighborhood. regardless if netflix cache's data at comcast's POP (to save on transit) - they now need to stop horribly oversubscribing their end users to be able to deliver this bitrates. anywho thanks for the VFX memories, glad i don't live in that world anymore :) -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From ike at blackskyresearch.net Wed Mar 19 16:23:57 2014 From: ike at blackskyresearch.net (Isaac (.ike) Levy) Date: Wed, 19 Mar 2014 20:23:57 +0000 Subject: [talk] JVM Questions Message-ID: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Hi All, A Java question, for my own sanity. Some folks I know are convinced that Oracle JDK and OpenJDK are identical. (OpenJDK is our Production standard for many reasons, yet follks keep using to Oracle JDK for dev...) QUESTIONS: Do folks here know anywhere I can find some practical code examples which expose the nasty bugs which can be encountered? Do folks here have any such examples, perhaps from your fine Java colleagues? (unsigned/signed int mania differences in particular)? In the vast and disparate world of Java, is there another good place where I should ask to find examples of this stuff? I'm dying to get my hands on some code to demonstrate the JVM implementation differences, so we can quit focusing on high level language spec discussions. Rocket- .ike From bonsaime at gmail.com Wed Mar 19 17:36:34 2014 From: bonsaime at gmail.com (Jesse Callaway) Date: Wed, 19 Mar 2014 17:36:34 -0400 Subject: [talk] JVM Questions In-Reply-To: <201403192023.s2JKNvsi020570@rs102.luxsci.com> References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: On Wed, Mar 19, 2014 at 4:23 PM, Isaac (.ike) Levy wrote: > > Hi All, > > A Java question, for my own sanity. > > Some folks I know are convinced that Oracle JDK and OpenJDK are identical. > (OpenJDK is our Production standard for many reasons, yet follks keep > using to Oracle JDK for dev...) > > QUESTIONS: > > Do folks here know anywhere I can find some practical code examples which > expose the nasty bugs which can be encountered? > > Do folks here have any such examples, perhaps from your fine Java > colleagues? (unsigned/signed int mania differences in particular)? > > In the vast and disparate world of Java, is there another good place where > I should ask to find examples of this stuff? > > I'm dying to get my hands on some code to demonstrate the JVM > implementation differences, so we can quit focusing on high level language > spec discussions. > > Rocket- > .ike > > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > One place where you can find documentation of differences is when looking at their next generation garbage collectors. Specifically not what you asked for: Regardless of which is better for the purposes at hand, when there are breakages due to the runtime stack they will be different between the stacks since they are in fact different and have different underlying code. I'm guessing you're thinking the same thing in re benchmarking and error testing in development, you'd want to compare apples to apples. In this case if OpenJDK is necessary for prod, I'd say that it's necessary for dev as well.... at least from a testing and benchmarking perspective. -- -jesse -------------- next part -------------- An HTML attachment was scrubbed... URL: From john at netpurgatory.com Wed Mar 19 16:55:44 2014 From: john at netpurgatory.com (John C. Vernaleo) Date: Wed, 19 Mar 2014 16:55:44 -0400 (EDT) Subject: [talk] JVM Questions In-Reply-To: <201403192023.s2JKNvsi020570@rs102.luxsci.com> References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: Not exactly a small or contained example, but the last time I worked with it, Android absoultely would not build with anything other than the Oracle (I still feel funny saying that and not Sun) JVM. Haven't touched it in over a year though. ------------------------------------------------------- John C. Vernaleo, Ph.D. www.netpurgatory.com john at netpurgatory.com ------------------------------------------------------- On Wed, 19 Mar 2014, Isaac (.ike) Levy wrote: > > Hi All, > > A Java question, for my own sanity. > > Some folks I know are convinced that Oracle JDK and OpenJDK are identical. > (OpenJDK is our Production standard for many reasons, yet follks keep using > to Oracle JDK for dev...) > > QUESTIONS: > > Do folks here know anywhere I can find some practical code examples which > expose the nasty bugs which can be encountered? > > Do folks here have any such examples, perhaps from your fine Java colleagues? > (unsigned/signed int mania differences in particular)? > > In the vast and disparate world of Java, is there another good place where I > should ask to find examples of this stuff? > > I'm dying to get my hands on some code to demonstrate the JVM implementation > differences, so we can quit focusing on high level language spec discussions. > > Rocket- > .ike > > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk > From edlinuxguru at gmail.com Wed Mar 19 18:26:18 2014 From: edlinuxguru at gmail.com (Edward Capriolo) Date: Wed, 19 Mar 2014 18:26:18 -0400 Subject: [talk] JVM Questions In-Reply-To: References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: I have run into these. This is really very very rare. One time it can happen is if an application does some high end native stuff. An example would be cassandra uses off heap memory to avoid gc. Sometmes the semantics of such features are not clear in the jvm spec i would guess. In any case you might hit a case where the software might get an error loading its data file in one jvm but not the other. I have jad similar issues with commercial vms as well. I always find out exactly what the developers use and i use that. Oracle jvm is more " enterprise " so it is typically paired with enterprise java products and thus havr better coverage. I do a lot of java work and i am rarely aware of if a given system is using one or the other, but i find oracle jvm more compatible. Never found anything that did work right on open jdk and not oracle. On Wednesday, March 19, 2014, John C. Vernaleo wrote: > Not exactly a small or contained example, but the last time I worked with it, Android absoultely would not build with anything other than the Oracle (I still feel funny saying that and not Sun) JVM. Haven't touched it in over a year though. > > ------------------------------------------------------- > John C. Vernaleo, Ph.D. > www.netpurgatory.com > john at netpurgatory.com > ------------------------------------------------------- > > On Wed, 19 Mar 2014, Isaac (.ike) Levy wrote: > >> >> Hi All, >> >> A Java question, for my own sanity. >> >> Some folks I know are convinced that Oracle JDK and OpenJDK are identical. >> (OpenJDK is our Production standard for many reasons, yet follks keep using to Oracle JDK for dev...) >> >> QUESTIONS: >> >> Do folks here know anywhere I can find some practical code examples which expose the nasty bugs which can be encountered? >> >> Do folks here have any such examples, perhaps from your fine Java colleagues? (unsigned/signed int mania differences in particular)? >> >> In the vast and disparate world of Java, is there another good place where I should ask to find examples of this stuff? >> >> I'm dying to get my hands on some code to demonstrate the JVM implementation differences, so we can quit focusing on high level language spec discussions. >> >> Rocket- >> .ike >> >> >> _______________________________________________ >> 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 > -- Sorry this was sent from mobile. Will do less grammar and spell check than usual. -------------- next part -------------- An HTML attachment was scrubbed... URL: From edlinuxguru at gmail.com Wed Mar 19 18:29:25 2014 From: edlinuxguru at gmail.com (Edward Capriolo) Date: Wed, 19 Mar 2014 18:29:25 -0400 Subject: [talk] JVM Questions In-Reply-To: References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: Because the jvm has a specification they should function exactly the same at all times. Because most of java is written in java that should be the same. However certain libraries especially the compiled c parts could be implemented differently. On Wednesday, March 19, 2014, Edward Capriolo wrote: > I have run into these. This is really very very rare. One time it can happen is if an application does some high end native stuff. An example would be cassandra uses off heap memory to avoid gc. Sometmes the semantics of such features are not clear in the jvm spec i would guess. In any case you might hit a case where the software might get an error loading its data file in one jvm but not the other. I have jad similar issues with commercial vms as well. > > I always find out exactly what the developers use and i use that. Oracle jvm is more " enterprise " so it is typically paired with enterprise java products and thus havr better coverage. > > I do a lot of java work and i am rarely aware of if a given system is using one or the other, but i find oracle jvm more compatible. Never found anything that did work right on open jdk and not oracle. > > On Wednesday, March 19, 2014, John C. Vernaleo wrote: >> Not exactly a small or contained example, but the last time I worked with it, Android absoultely would not build with anything other than the Oracle (I still feel funny saying that and not Sun) JVM. Haven't touched it in over a year though. >> >> ------------------------------------------------------- >> John C. Vernaleo, Ph.D. >> www.netpurgatory.com >> john at netpurgatory.com >> ------------------------------------------------------- >> >> On Wed, 19 Mar 2014, Isaac (.ike) Levy wrote: >> >>> >>> Hi All, >>> >>> A Java question, for my own sanity. >>> >>> Some folks I know are convinced that Oracle JDK and OpenJDK are identical. >>> (OpenJDK is our Production standard for many reasons, yet follks keep using to Oracle JDK for dev...) >>> >>> QUESTIONS: >>> >>> Do folks here know anywhere I can find some practical code examples which expose the nasty bugs which can be encountered? >>> >>> Do folks here have any such examples, perhaps from your fine Java colleagues? (unsigned/signed int mania differences in particular)? >>> >>> In the vast and disparate world of Java, is there another good place where I should ask to find examples of this stuff? >>> >>> I'm dying to get my hands on some code to demonstrate the JVM implementation differences, so we can quit focusing on high level language spec discussions. >>> >>> Rocket- >>> .ike >>> >>> >>> _______________________________________________ >>> 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 >> > > -- > Sorry this was sent from mobile. Will do less grammar and spell check than usual. > -- Sorry this was sent from mobile. Will do less grammar and spell check than usual. -------------- next part -------------- An HTML attachment was scrubbed... URL: From mark.saad at ymail.com Wed Mar 19 18:45:09 2014 From: mark.saad at ymail.com (Mark Saad) Date: Wed, 19 Mar 2014 18:45:09 -0400 Subject: [talk] JVM Questions In-Reply-To: <201403192023.s2JKNvsi020570@rs102.luxsci.com> References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: > On Mar 19, 2014, at 4:23 PM, "Isaac (.ike) Levy" wrote: > > > Hi All, > > A Java question, for my own sanity. > > Some folks I know are convinced that Oracle JDK and OpenJDK are identical. > (OpenJDK is our Production standard for many reasons, yet follks keep using to Oracle JDK for dev...) > > QUESTIONS: > > Do folks here know anywhere I can find some practical code examples which expose the nasty bugs which can be encountered? > So openjdk 7 is the reference implementation of the version 7 of the java spec . Iirc there is no single place for this only bug reports on how open behaves poorly compared to the reference . > Do folks here have any such examples, perhaps from your fine Java colleagues? (unsigned/signed int mania differences in particular)? > There are lots if bullshit reports that mistakenly point to openjdk6 issues . Openjdk6 is not the same as openjdk7 . 6 was made by taking 7 and hacking in the old spec . It breaks in weird ways . Forget it exists . > In the vast and disparate world of Java, is there another good place where I should ask to find examples of this stuff? > > I'm dying to get my hands on some code to demonstrate the JVM implementation differences, so we can quit focusing on high level language spec discussions. > The java test kit that the FreeBSD foundation used to certify the diablo jkd as compliant , would be a good staring point . > Rocket- > .ike > > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk --- Mark saad | mark.saad at longcount.org From pete at nomadlogic.org Wed Mar 19 18:49:33 2014 From: pete at nomadlogic.org (Pete Wright) Date: Wed, 19 Mar 2014 15:49:33 -0700 Subject: [talk] JVM Questions In-Reply-To: <201403192023.s2JKNvsi020570@rs102.luxsci.com> References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: <532A1EFD.6030606@nomadlogic.org> On 03/19/14 13:23, Isaac (.ike) Levy wrote: > > Hi All, > > A Java question, for my own sanity. > > Some folks I know are convinced that Oracle JDK and OpenJDK are identical. > (OpenJDK is our Production standard for many reasons, yet follks keep > using to Oracle JDK for dev...) > it depends on the JRE you are targeting. Java8 development is happening on the OpenJDK side of things first. I believe the intent is to make it as performant as the Oracle JRE. I believe this is also true, to a lesser extent, with Java 7. thus spoke the wikipedia: "The OpenJDK 8 project, which will be the basis for JDK 8, has a planned release date of 3/18/2014." > QUESTIONS: > > Do folks here know anywhere I can find some practical code examples > which expose the nasty bugs which can be encountered? > > Do folks here have any such examples, perhaps from your fine Java > colleagues? (unsigned/signed int mania differences in particular)? > > In the vast and disparate world of Java, is there another good place > where I should ask to find examples of this stuff? > In my personal experience I found that there rarely are issues running the OpenJDK as a client system, or even for development. for example, I use jmxtrans to collect tomcat metrics via JMX at work. I develop against, and run jmxtrans on freebsd/openjdk6 with out issues. Although when it comes to production releases we standardize on specific Oracle JRE's. This is due to mostly to reduce headaches when working with external vendors or code-bases that may "require" - kinda how if you want commercial support from Oracle 11g you need to run either RHEL or orable linux, not CentOS. It also removes a variable when trying to squash hard to pin down bugs. > I'm dying to get my hands on some code to demonstrate the JVM > implementation differences, so we can quit focusing on high level > language spec discussions. hoping this will be a moot point now that JDK 8 has been released by Oracle as of Tuesday (March 18th). -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From pete at nomadlogic.org Wed Mar 19 18:51:10 2014 From: pete at nomadlogic.org (Pete Wright) Date: Wed, 19 Mar 2014 15:51:10 -0700 Subject: [talk] JVM Questions In-Reply-To: References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: <532A1F5E.8070703@nomadlogic.org> On 03/19/14 14:36, Jesse Callaway wrote: > > > > On Wed, Mar 19, 2014 at 4:23 PM, Isaac (.ike) Levy > > wrote: > > > Hi All, > > A Java question, for my own sanity. > > Some folks I know are convinced that Oracle JDK and OpenJDK are > identical. > (OpenJDK is our Production standard for many reasons, yet follks > keep using to Oracle JDK for dev...) > > QUESTIONS: > > Do folks here know anywhere I can find some practical code examples > which expose the nasty bugs which can be encountered? > > Do folks here have any such examples, perhaps from your fine Java > colleagues? (unsigned/signed int mania differences in particular)? > > In the vast and disparate world of Java, is there another good place > where I should ask to find examples of this stuff? > > I'm dying to get my hands on some code to demonstrate the JVM > implementation differences, so we can quit focusing on high level > language spec discussions. > > Rocket- > .ike > > > _________________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/__mailman/listinfo/talk > > > > > > One place where you can find documentation of differences is when > looking at their next generation garbage collectors. > > Specifically not what you asked for: > Regardless of which is better for the purposes at hand, when there are > breakages due to the runtime stack they will be different between the > stacks since they are in fact different and have different underlying > code. I'm guessing you're thinking the same thing in re benchmarking and > error testing in development, you'd want to compare apples to apples. In > this case if OpenJDK is necessary for prod, I'd say that it's necessary > for dev as well.... at least from a testing and benchmarking perspective. > +1 Keep Is conSistant Stupid? ;^) -pete -- Pete Wright pete at nomadlogic.org twitter => @nomadlogicLA From pvarga at pvrg.net Wed Mar 19 20:48:45 2014 From: pvarga at pvrg.net (pvarga at pvrg.net) Date: Wed, 19 Mar 2014 20:48:45 -0400 Subject: [talk] JVM Questions In-Reply-To: <201403192023.s2JKNvsi020570@rs102.luxsci.com> References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: <20140320004845.GA14992@pvrg.net> On Wed, Mar 19, 2014 at 08:23:57PM +0000, Isaac (.ike) Levy wrote: > > Hi All, > > A Java question, for my own sanity. > > Some folks I know are convinced that Oracle JDK and OpenJDK are identical. > (OpenJDK is our Production standard for many reasons, yet follks keep > using to Oracle JDK for dev...) About 5 years ago, when glassfish was around along other java vms, jdks, the difference was runtime performance and memory usage. I would expect that be still an differentiating factor. I think glassfish went on to become Oracle's, so this may be irrelevant. > > QUESTIONS: > > Do folks here know anywhere I can find some practical code examples > which expose the nasty bugs which can be encountered? > > Do folks here have any such examples, perhaps from your fine Java > colleagues? (unsigned/signed int mania differences in particular)? > > In the vast and disparate world of Java, is there another good place > where I should ask to find examples of this stuff? > > I'm dying to get my hands on some code to demonstrate the JVM > implementation differences, so we can quit focusing on high level > language spec discussions. > > Rocket- > .ike > when i used to write java code, the prod platform was the goal to run on and as a good test i routinely ran with different java implementations. i did not really have a lot of interesting problems that could not be solved with the simple principle of writing code to lang-specs. From pvarga at pvrg.net Wed Mar 19 20:50:06 2014 From: pvarga at pvrg.net (pvarga at pvrg.net) Date: Wed, 19 Mar 2014 20:50:06 -0400 Subject: [talk] JVM Questions In-Reply-To: References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: <20140320005006.GB14992@pvrg.net> On Wed, Mar 19, 2014 at 04:55:44PM -0400, John C. Vernaleo wrote: > Not exactly a small or contained example, but the last time I worked with > it, Android absoultely would not build with anything other than the Oracle > (I still feel funny saying that and not Sun) JVM. Haven't touched it in > over a year though. Yes, i second that. this is what drove me away from android dev. even though it is pretty much a lot of fun. hello, beagleboard. :-) > > ------------------------------------------------------- > John C. Vernaleo, Ph.D. > www.netpurgatory.com > john at netpurgatory.com > ------------------------------------------------------- > > On Wed, 19 Mar 2014, Isaac (.ike) Levy wrote: > > > > >Hi All, > > > >A Java question, for my own sanity. > > > >Some folks I know are convinced that Oracle JDK and OpenJDK are identical. > >(OpenJDK is our Production standard for many reasons, yet follks keep > >using to Oracle JDK for dev...) > > > >QUESTIONS: > > > >Do folks here know anywhere I can find some practical code examples which > >expose the nasty bugs which can be encountered? > > > >Do folks here have any such examples, perhaps from your fine Java > >colleagues? (unsigned/signed int mania differences in particular)? > > > >In the vast and disparate world of Java, is there another good place where > >I should ask to find examples of this stuff? > > > >I'm dying to get my hands on some code to demonstrate the JVM > >implementation differences, so we can quit focusing on high level language > >spec discussions. > > > >Rocket- > >.ike > > > > > >_______________________________________________ > >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 sjt.kar at gmail.com Fri Mar 21 09:02:18 2014 From: sjt.kar at gmail.com (Sujit K M) Date: Fri, 21 Mar 2014 18:32:18 +0530 Subject: [talk] JVM Questions In-Reply-To: <201403192023.s2JKNvsi020570@rs102.luxsci.com> References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: On Thu, Mar 20, 2014 at 1:53 AM, Isaac (.ike) Levy wrote: > > Hi All, > > A Java question, for my own sanity. > > Some folks I know are convinced that Oracle JDK and OpenJDK are identical. > (OpenJDK is our Production standard for many reasons, yet follks keep using > to Oracle JDK for dev...) Might be the compilers they use might sun studio for Oracle JDK and GCC for OpenJDK. > QUESTIONS: > > Do folks here know anywhere I can find some practical code examples which > expose the nasty bugs which can be encountered? The ODBC/JDBC Drivers are particular to these. One bug I find and keep finding is the one below resultSet.getInt(getParameter()); // will most probably lead to a very bad bug resultSet.get(i);// is not going to get this. > Do folks here have any such examples, perhaps from your fine Java > colleagues? (unsigned/signed int mania differences in particular)? > > In the vast and disparate world of Java, is there another good place where I > should ask to find examples of this stuff? One of the major bugs I found in the Collections Framework was the HashMap and ConcurrenAccessException in JDK5 long time ago. I Think Best way to Look at this is in Exceptions that a method or class instance might throw and try writing a unit test case for that. > > I'm dying to get my hands on some code to demonstrate the JVM implementation > differences, so we can quit focusing on high level language spec > discussions. Best way would be to compile OpenJDK using Sun Studio and seeing how it goes from there. From sjt.kar at gmail.com Sat Mar 22 04:50:59 2014 From: sjt.kar at gmail.com (Sujit K M) Date: Sat, 22 Mar 2014 14:20:59 +0530 Subject: [talk] Looking for Prospective Clients Message-ID: All, As I was Moving on to start up(owner), which basically Looks at a new way of generating interest in Open source software, A new Model at contributing and earning the revenue to keep a place afloat, which is much better than an Only TCO Objective. This model will never be disclosed but attempts to support FreeBSD and Linux Operating Systems at cost at ease space for all involved. As the subject says I am looking for clients. Hope some of you promote this as well as I could get some clients to hit me off the list. Hoping for a good response. Regards, Sujit K M From sjt.kar at gmail.com Mon Mar 24 09:40:03 2014 From: sjt.kar at gmail.com (Sujit K M) Date: Mon, 24 Mar 2014 19:10:03 +0530 Subject: [talk] Looking for Prospective Clients In-Reply-To: References: Message-ID: As an FYI, The site is put up at http://trusted-site.net/. On Sat, Mar 22, 2014 at 2:20 PM, Sujit K M wrote: > All, > As I was Moving on to start up(owner), which basically > Looks at a new way of > generating interest in Open source software, A new Model at > contributing and earning the > revenue to keep a place afloat, which is much better than an Only TCO > Objective. This > model will never be disclosed but attempts to support FreeBSD and > Linux Operating > Systems at cost at ease space for all involved. > > As the subject says I am looking for clients. Hope some of you promote > this as well as I > could get some clients to hit me off the list. > > Hoping for a good response. > > Regards, > > Sujit K M -- -- Sujit K M blog(http://kmsujit.blogspot.com/) From george at ceetonetechnology.com Mon Mar 24 09:42:51 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Mon, 24 Mar 2014 09:42:51 -0400 Subject: [talk] Looking for Prospective Clients In-Reply-To: References: Message-ID: <5330365B.8080006@ceetonetechnology.com> Sujit K M: > As an FYI, > > The site is put up at http://trusted-site.net/. Hey Sujit: If you notice, there is never commercial queries on this list. It is not appropriate to use it for such purposes. This a volunteer organization, and using our free labor for commercial purposes is not acceptable. Thanks g From sjt.kar at gmail.com Mon Mar 24 09:51:42 2014 From: sjt.kar at gmail.com (Sujit K M) Date: Mon, 24 Mar 2014 19:21:42 +0530 Subject: [talk] Looking for Prospective Clients In-Reply-To: <5330365B.8080006@ceetonetechnology.com> References: <5330365B.8080006@ceetonetechnology.com> Message-ID: On Mon, Mar 24, 2014 at 7:12 PM, George Rosamond wrote: > Sujit K M: >> As an FYI, >> >> The site is put up at http://trusted-site.net/. > > Hey Sujit: > > If you notice, there is never commercial queries on this list. It is > not appropriate to use it for such purposes. Sorry about that, But as previous mail indicates, it is a way help Open Source movement and Quality we achieve and improve on it. > This a volunteer organization, and using our free labor for commercial > purposes is not acceptable. Yeah I had thought of a .org domain but then I had to stick with a .net domain as we strictly are not an non profit organization. > > Thanks > > g > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk -- -- Sujit K M blog(http://kmsujit.blogspot.com/) From mark.saad at ymail.com Fri Mar 28 14:34:58 2014 From: mark.saad at ymail.com (Mark Saad) Date: Fri, 28 Mar 2014 11:34:58 -0700 (PDT) Subject: [talk] Arm Board Message-ID: <1396031698.40429.YahooMailBasic@web140104.mail.bf1.yahoo.com> Hey Fellow ARM users Has anyone tried this board / CPU combo out http://www.newegg.com/Product/Product.aspx?Item=N82E16813190005 Its has a 4 core NVIDIA Tegra K1 , 2G RAM, SATA ports , Mini PCI-Express and the list goes on . -- Mark Saad mark.saad at ymail.com From gnn at neville-neil.com Fri Mar 28 15:42:42 2014 From: gnn at neville-neil.com (George Neville-Neil) Date: Fri, 28 Mar 2014 15:42:42 -0400 Subject: [talk] Arm Board In-Reply-To: <1396031698.40429.YahooMailBasic@web140104.mail.bf1.yahoo.com> References: <1396031698.40429.YahooMailBasic@web140104.mail.bf1.yahoo.com> Message-ID: On Mar 28, 2014, at 14:34 , Mark Saad wrote: > Hey Fellow ARM users > > Has anyone tried this board / CPU combo out > > http://www.newegg.com/Product/Product.aspx?Item=N82E16813190005 > > Its has a 4 core NVIDIA Tegra K1 , 2G RAM, SATA ports , Mini PCI-Express and the list goes on . > Have not tried it, it?s on the high end of those types of boards at $200. Best, George From george at ceetonetechnology.com Fri Mar 28 17:19:22 2014 From: george at ceetonetechnology.com (George Rosamond) Date: Fri, 28 Mar 2014 17:19:22 -0400 Subject: [talk] Arm Board In-Reply-To: References: <1396031698.40429.YahooMailBasic@web140104.mail.bf1.yahoo.com> Message-ID: <5335E75A.2000308@ceetonetechnology.com> George Neville-Neil: > > On Mar 28, 2014, at 14:34 , Mark Saad wrote: > >> Hey Fellow ARM users >> >> Has anyone tried this board / CPU combo out >> >> http://www.newegg.com/Product/Product.aspx?Item=N82E16813190005 >> >> Its has a 4 core NVIDIA Tegra K1 , 2G RAM, SATA ports , Mini PCI-Express and the list goes on . >> > Have not tried it, it?s on the high end of those types of boards at $200. I think this board came up on the FBSD arm list the other day. The boards will keep proliferating, and from what it seems from my view, the divergence in porting will grow with it. Meaning that there has to be some prioritization in what devs focus on. These firms are using onsite devs to build Linux ports.. .and none of the BSD projects can keep pace with that. The fact that NVIDIA is joining the crowd is significant, but I think we're looking at a lot more boards from the big boys as they catch-up. If there is a question of sorting out the 'short list' of ports to be focused on, I'd assume there are other multi-core boards people are focused on... besides the lower-end RPi, BeagleBone, Wandaboards and Cubieboards. I know Pete is playing with something... g From mark.saad at ymail.com Fri Mar 28 21:58:01 2014 From: mark.saad at ymail.com (Mark Saad) Date: Fri, 28 Mar 2014 21:58:01 -0400 Subject: [talk] Arm Board In-Reply-To: <5335E75A.2000308@ceetonetechnology.com> References: <1396031698.40429.YahooMailBasic@web140104.mail.bf1.yahoo.com> <5335E75A.2000308@ceetonetechnology.com> Message-ID: <9E2932D8-6243-4C28-8AA4-1AB09710A67A@ymail.com> > On Mar 28, 2014, at 5:19 PM, George Rosamond wrote: > > George Neville-Neil: >> >>> On Mar 28, 2014, at 14:34 , Mark Saad wrote: >>> >>> Hey Fellow ARM users >>> >>> Has anyone tried this board / CPU combo out >>> >>> http://www.newegg.com/Product/Product.aspx?Item=N82E16813190005 >>> >>> Its has a 4 core NVIDIA Tegra K1 , 2G RAM, SATA ports , Mini PCI-Express and the list goes on . >> Have not tried it, it?s on the high end of those types of boards at $200. > > I think this board came up on the FBSD arm list the other day. > > The boards will keep proliferating, and from what it seems from my view, > the divergence in porting will grow with it. Meaning that there has to > be some prioritization in what devs focus on. These firms are using > onsite devs to build Linux ports.. .and none of the BSD projects can > keep pace with that. > > The fact that NVIDIA is joining the crowd is significant, but I think > we're looking at a lot more boards from the big boys as they catch-up. > So weirdly nvidia is not targeting a known distro but a nvidia developed thing . Whilst Linux is Linux per say that aspect of the board lends me to think its to exotic , to waste time with . > If there is a question of sorting out the 'short list' of ports to be > focused on, I'd assume there are other multi-core boards people are > focused on... besides the lower-end RPi, BeagleBone, Wandaboards and > Cubieboards. > The sexy aspect of this board was the ram, expansion, and dedicated db9 . I don't care about the video aspects just yet they seam hard to use . Ie nvidia specific decoders for some apps and Libs . That being said it feels like today's arm setups are a headless mess of random bits . Vendor x has on way of laying out the board and vendor y has a completely different way to do their setup of the same parts . While I get that each wants to protect their IP ; Most of the variation boils down to "how can I do this cheaper and faster " with little care for how it will work in a year . Am I wrong ? > I know Pete is playing with something... > > g > > _______________________________________________ > talk mailing list > talk at lists.nycbug.org > http://lists.nycbug.org/mailman/listinfo/talk From sjt.kar at gmail.com Sun Mar 30 07:33:33 2014 From: sjt.kar at gmail.com (Sujit K M) Date: Sun, 30 Mar 2014 17:03:33 +0530 Subject: [talk] JVM Questions In-Reply-To: References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: On Tue, Mar 25, 2014 at 2:11 AM, George Magiros wrote: > The specification doesn't specify a garbage collection algorithm fyi, " I don't understand why you top posted to this particular mail. > Implementation details that are not part of the Java Virtual Machine's > specification would unnecessarily constrain the creativity of implementors. > .For example, the memory layout of run-time data areas, the > garbage-collection algorithm used, and any internal optimization of the Java > Virtual Machine instructions (for example, translating them into machine > code) are left to the discretion of the implementor." > http://docs.oracle.com/javase/specs/jvms/se7/html/jvms-2.html This as per me is commonsense, I don't understand why you are bringing this as an issue. This a common characteristic of any Language to leave the implementation to the implementer, example C++. > > > On Fri, Mar 21, 2014 at 9:02 AM, Sujit K M wrote: >> >> On Thu, Mar 20, 2014 at 1:53 AM, Isaac (.ike) Levy >> wrote: >> > >> > Hi All, >> > >> > A Java question, for my own sanity. >> > >> > Some folks I know are convinced that Oracle JDK and OpenJDK are >> > identical. >> > (OpenJDK is our Production standard for many reasons, yet follks keep >> > using >> > to Oracle JDK for dev...) >> >> Might be the compilers they use might sun studio for Oracle JDK and GCC >> for >> OpenJDK. >> >> > QUESTIONS: >> > >> > Do folks here know anywhere I can find some practical code examples >> > which >> > expose the nasty bugs which can be encountered? >> >> The ODBC/JDBC Drivers are particular to these. One bug I find and keep >> finding >> is the one below >> resultSet.getInt(getParameter()); // will most probably lead >> to a very bad bug >> resultSet.get(i);// is not going to get this. >> >> > Do folks here have any such examples, perhaps from your fine Java >> > colleagues? (unsigned/signed int mania differences in particular)? >> > >> > In the vast and disparate world of Java, is there another good place >> > where I >> > should ask to find examples of this stuff? >> >> One of the major bugs I found in the Collections Framework was the HashMap >> and ConcurrenAccessException in JDK5 long time ago. I Think Best way >> to Look at this is in Exceptions that a method or class instance might >> throw >> and try writing a unit test case for that. >> >> > >> > I'm dying to get my hands on some code to demonstrate the JVM >> > implementation >> > differences, so we can quit focusing on high level language spec >> > discussions. >> >> Best way would be to compile OpenJDK using Sun Studio and seeing >> how it goes from there. >> _______________________________________________ >> talk mailing list >> talk at lists.nycbug.org >> http://lists.nycbug.org/mailman/listinfo/talk > > -- -- Sujit K M blog(http://kmsujit.blogspot.com/) From sjt.kar at gmail.com Sun Mar 30 10:12:51 2014 From: sjt.kar at gmail.com (Sujit K M) Date: Sun, 30 Mar 2014 19:42:51 +0530 Subject: [talk] JVM Questions In-Reply-To: References: <201403192023.s2JKNvsi020570@rs102.luxsci.com> Message-ID: On Tue, Mar 25, 2014 at 2:11 AM, George Magiros wrote: > The specification doesn't specify a garbage collection algorithm fyi, " > Implementation details that are not part of the Java Virtual Machine's > specification would unnecessarily constrain the creativity of implementors. > .For example, the memory layout of run-time data areas, the > garbage-collection algorithm used, and any internal optimization of the Java > Virtual Machine instructions (for example, translating them into machine > code) are left to the discretion of the implementor." > http://docs.oracle.com/javase/specs/jvms/se7/html/jvms-2.html Not sure whether you wanted to mention (http://www.1001javatips.com/) -- -- Sujit K M blog(http://kmsujit.blogspot.com/)