Contributed by paul on from the openbsd-hpc-wtf? dept.
Chris Kuethe (ckuethe@) wrote in about how he and his team (University of Alberta) used OpenBSD in their win in the Cluster Challenge at the Supercomputing '07 Conference in Reno, Nevada. The competition included six teams from different universities and their partners:
Indiana University & Apple,
National Tsing Hua University (Taiwan) & ASUSTek,
Purdue University & HP,
Stony Brook University & Dell,
University of Alberta (Canada) & SGI and
University of Colorado & Aspen Systems.
Below is Chris's account of the competition (and OpenBSD's role in it).
In November 2007 I participated in the Supercomputing '07 Cluster Challenge, along with Gordon Klok (gwk@), Paul Greidanus, Stephan Portillo, Antoine Fillion and Andrew Nisbet. Coaching the team was a professor at the University of Alberta, Dr. Paul Lu, as well as Bob Beck (beck@). The idea of the competition was to build a high-performance cluster in a small (some would say unrealistically low) power budget: 26A at 120V. Once the cluster was built, commonly used scientific applications were run on the clusters. Points were awarded for the successful completion of data sets and penalties were assessed for power overages. I also worked on the DHCP server and handled the security concerns; showing up at a conference with a cluster of linux machines optimized for performance rather than security is not a recipe for high uptime.
Because conference networks are not to be trusted, and to simplify management and because the competition rules forbade incoming connections to the cluster, we hid our SGI Altix XE310 cluster behind a Nexcom EBS1563 running OpenBSD. This machine was also our DNS and DHCP server. Because the cluster itself was built with SystemImager (which uses special DHCP options for configuration), we used the isc-dhcpd port which supports a few more dhcp options than the system dhcpd. Installing the dhcpd port meant that we could have one single dhcp server for our laptops, the compute nodes, and their onboard baseboard management controllers (BMC). We installed the ipmitool port to interact with the BMC cards. IPMI is quite powerful it can query sensors, peek at the operating system and control the hardware but, we just used it to power cycle the nodes.
A part of the competition was visualization. We were given workloads for POP (Parallel Ocean Program), GAMESS (General Atomic and Molecular Electronic Structure System) and POV-Ray(port); it was strongly suggested that we be prepared to display the results of our computation in an aesthetically appealing visual form. The POV-Ray jobs were obvious, each job was a movie. Molden could be used to quickly render molecular geometry as calculated by GAMESS. We also used openbabel to convert GAMESS output to PDB format. Povchem can take PDB files and convert them into POV-ray files, which we then used to produce shiny anti-aliased molecule(s).
It was decided that we'd use an AppleTV to simplify the display of the generated imagery. Single frames were a non-issue, but the AppleTV could be quite picky about what video formats it would accept. For whatever reason, the various linux laptops we'd brought along didn't generate videos that would play correctly, at least not easily. Since we were operating under a deadline, we didn't have time to figure out why 2 or 3 distributions couldn't get a working video transcoding pipeline together. Ffmpeg from the ports tree to save the day! Just one command and a few minutes of transcoding later, we had movies that would play. I quickly realized that I could test whether a movie would play by loading it on to my ipod with gtkpod, which saved me the effort of interrupting a presentation on the AppleTV to see if a movie would play. Unfortunately the AppleTV has a glaring defect - the inability to loop videos. We initially worked around this by using mencoder to generate a long video by copying the shorter videos end to end a few times.
A big thank you goes to everyone involved in keeping ports working well - the ports tree allowed us to install third party software with reasonable assurance that it "does what it says on the tin."
Thank you to Chris Kuethe for his account of the event.
(Comments are closed)
By Anonymous Coward (85.178.86.126) on
Comments
By clvrmnky (69.28.228.76) on
>
>
Well, you either ant vanilla DHCP, or you want DHCP with the bells and whistles.
Until I needed DDNS I wouldn't have considered the non-default dhcpd.
By Anonymous Coward (198.175.14.193) on
Comments
By Simon Bertrang (213.128.132.194) simon@ on
[SNIP]
What's your name? Where's your contribution? I don't see you doing
anything that could give you the right to complain... so either do
something to prove your claimed standards, come up with actual
criticism or... well, shut up.
Thanks to Chris for presenting another case where OpenBSD did its job :-)
Comments
By eax (82.233.114.52) on
> [SNIP]
>
> What's your name? Where's your contribution? I don't see you doing
> anything that could give you the right to complain... so either do
> something to prove your claimed standards, come up with actual
> criticism or... well, shut up.
>
> Thanks to Chris for presenting another case where OpenBSD did its job :-)
it will be cool to have more details on the cluster configuration ;)
By chris cappuccio (198.175.14.193) on
> [SNIP]
>
> What's your name? Where's your contribution? I don't see you doing
> anything that could give you the right to complain... so either do
> something to prove your claimed standards, come up with actual
> criticism or... well, shut up.
>
I was chris@ for a while, but I haven't had much to commit for some time. So, take it with a grain of salt. But I did contribute the initial altq port, vlan interface port, various other driver ports, mostly stuff that would add to openbsd's reputation as a "router platform" -- but the truth is that it is really a great unix platform and I think it's silly that people aren't talking about THAT.
This article opens with summary text about "OpenBSD's role" in a "competition" in "High Performance Computing" only to actually explain how it was Chris' Linux cluster firewall...Please, what a joke. What the fuck are they using all these "Linux laptops" for? Ones that they apparently don't even know how to use? Why not OpenBSD? Where's the OpenBSD here? Why isn't the cluster running OpenBSD? What are the limitations in OpenBSD that need to be fixed for X cluster application that force you to run Linux? Why isn't there anything even remotely interesting in the article when the material is clearly there? Maybe it's group dynamics where the whole project was developed on Linux from the beginning? If so, why? What level of scalability is lacking from OpenBSD? Or from Linux? You can imagine the range of discussion that's missing from this article.
Comments
By Anonymous Coward (62.103.255.199) on
> > [SNIP]
> >
> > What's your name? Where's your contribution? I don't see you doing
> > anything that could give you the right to complain... so either do
> > something to prove your claimed standards, come up with actual
> > criticism or... well, shut up.
> >
>
> I was chris@ for a while, but I haven't had much to commit for some time. So, take it with a grain of salt. But I did contribute the initial altq port, vlan interface port, various other driver ports, mostly stuff that would add to openbsd's reputation as a "router platform" -- but the truth is that it is really a great unix platform and I think it's silly that people aren't talking about THAT.
>
> This article opens with summary text about "OpenBSD's role" in a "competition" in "High Performance Computing" only to actually explain how it was Chris' Linux cluster firewall...Please, what a joke. What the fuck are they using all these "Linux laptops" for? Ones that they apparently don't even know how to use? Why not OpenBSD? Where's the OpenBSD here? Why isn't the cluster running OpenBSD? What are the limitations in OpenBSD that need to be fixed for X cluster application that force you to run Linux? Why isn't there anything even remotely interesting in the article when the material is clearly there? Maybe it's group dynamics where the whole project was developed on Linux from the beginning? If so, why? What level of scalability is lacking from OpenBSD? Or from Linux? You can imagine the range of discussion that's missing from this article.
Yup.It's kind of sad reading about how a giant such as OpenBSD gets used as a, ahmm firewall, for a linux cluster..
I wouldnt like to think that obsd's "job" there could only be restricted to building ports and filtering packets..
By Simon Bertrang (213.128.132.194) simon@ on
> > [SNIP]
> >
> > What's your name? Where's your contribution? I don't see you doing
> > anything that could give you the right to complain... so either do
> > something to prove your claimed standards, come up with actual
> > criticism or... well, shut up.
> >
>
> I was chris@ for a while, but I haven't had much to commit for some
> time. So, take it with a grain of salt. But I did contribute the
> initial altq port, vlan interface port, various other driver ports,
> mostly stuff that would add to openbsd's reputation as a "router
> platform" -- but the truth is that it is really a great unix platform
> and I think it's silly that people aren't talking about THAT.
>
People actually are... maybe not using the same words but more implying
that the used unix platform is really great of course.
> This article opens with summary text about "OpenBSD's role" in a
> "competition" in "High Performance Computing" only to actually explain
> how it was Chris' Linux cluster firewall...Please, what a joke. What
> the fuck are they using all these "Linux laptops" for? Ones that they
> apparently don't even know how to use? Why not OpenBSD? Where's the
> OpenBSD here? Why isn't the cluster running OpenBSD? What are the
> limitations in OpenBSD that need to be fixed for X cluster application
> that force you to run Linux? Why isn't there anything even remotely
> interesting in the article when the material is clearly there? Maybe
> it's group dynamics where the whole project was developed on Linux
> from the beginning? If so, why? What level of scalability is lacking
> from OpenBSD? Or from Linux? You can imagine the range of discussion
> that's missing from this article.
Oh, i think you missed some parts of the article then. From what i read
the whole data visualization was done with the help of OpenBSD ports.
I even remember when ckuethe@ was asking for OKs on the molden port
before it got imported. Also ipmitool, povray, ffmpeg and gtkpod from
the ports were used.
And there was a comparison part too, where some tools didn't work on
other systems - if that doesn't speak for OpenBSD... :-)
Your point of view seems to be more from the base system ignoring the
whole application part which is at least as important as the underlying
base empowering the whole.
You still have some good questions...
Comments
By Anonymous Coward (85.178.86.126) on
> > > [SNIP]
> > >
> > > What's your name? Where's your contribution? I don't see you doing
> > > anything that could give you the right to complain... so either do
> > > something to prove your claimed standards, come up with actual
> > > criticism or... well, shut up.
> > >
> >
> > I was chris@ for a while, but I haven't had much to commit for some
> > time. So, take it with a grain of salt. But I did contribute the
> > initial altq port, vlan interface port, various other driver ports,
> > mostly stuff that would add to openbsd's reputation as a "router
> > platform" -- but the truth is that it is really a great unix platform
> > and I think it's silly that people aren't talking about THAT.
> >
>
> People actually are... maybe not using the same words but more implying
> that the used unix platform is really great of course.
>
> > This article opens with summary text about "OpenBSD's role" in a
> > "competition" in "High Performance Computing" only to actually explain
> > how it was Chris' Linux cluster firewall...Please, what a joke. What
> > the fuck are they using all these "Linux laptops" for? Ones that they
> > apparently don't even know how to use? Why not OpenBSD? Where's the
> > OpenBSD here? Why isn't the cluster running OpenBSD? What are the
> > limitations in OpenBSD that need to be fixed for X cluster application
> > that force you to run Linux? Why isn't there anything even remotely
> > interesting in the article when the material is clearly there? Maybe
> > it's group dynamics where the whole project was developed on Linux
> > from the beginning? If so, why? What level of scalability is lacking
> > from OpenBSD? Or from Linux? You can imagine the range of discussion
> > that's missing from this article.
>
> Oh, i think you missed some parts of the article then. From what i read
> the whole data visualization was done with the help of OpenBSD ports.
> I even remember when ckuethe@ was asking for OKs on the molden port
> before it got imported. Also ipmitool, povray, ffmpeg and gtkpod from
> the ports were used.
> And there was a comparison part too, where some tools didn't work on
> other systems - if that doesn't speak for OpenBSD... :-)
>
> Your point of view seems to be more from the base system ignoring the
> whole application part which is at least as important as the underlying
> base empowering the whole.
>
> You still have some good questions...
Bullshit...
OpenBSD and "speed" is like Linux and security.. it just does not match...
And OpenBSD and "virtualisation"? Are you nuts.. with qhat? qemu? Seriously....
Comments
By Anonymous Coward (70.141.212.164) on
> OpenBSD and "speed" is like Linux and security.. it just does not match...
This has been said to be true for a long while, but why does it have to be so? I wish that could be addressed. It would be especially cool if the performance could be improved with multi-threading and multi-core CPU's. It would be awesome to see OpenBSD become synonymous with performance as it is with security.
Comments
By Anonymous Coward (85.178.107.138) on
> OpenBSD and "speed" is like Linux and security.. it just does not match...
>
>
> This has been said to be true for a long while, but why does it have to be so? I wish that could be addressed. It would be especially cool if the performance could be improved with multi-threading and multi-core CPU's. It would be awesome to see OpenBSD become synonymous with performance as it is with security.
Absolutly but you need a lot knowledge and of course GOOD coders and seriously: I'm neither.... like the other 98% of the OpenBSD userbase?!
Developers tell me to send in reports and tell 'em what nerves me and if I do so I offen get told "Well where is oyur oatch?".
I would love to use OpenBSD a VM_Server but how should I do it? With kqemue every report of me becomes useless because I don't use a "default" Kernel and with qemu each VM is as slow as a P2 233Mhz CPU...
Something wich is importent for me: Security comes first (OpenBSD is not as secure as it could be! Other OSs have better solutions partly f.e. for disk encryption!) but after that it's realy about the speed.
The security went down lately (users get forced to run current :( ) and the speed didn't improved either that much (the IP stack comes into my mind or the SMP iimplementation..). OpenBSD looses some grounds where it could simply just "rock" (f.e. most other OSs provide virtualisation! NetBSD/FreeBSD with Xen, Solaris, Linux... even MS Server 2008...). But hey.. we'll get a snmp-Daemon but have to install 3rd party tools to query it... maybe it's just me who thinks that doesn't make much sense at all always. :-/
I can't change anything and I don't complain. I'll just use something else but it doesn't make it easier to raise some bucks from my boss...
Comments
By Matthew Dempsky (69.232.192.105) on
I know! It's almost as outrageous as how Apache is included in base, but I have to install Firefox as a 3rd party package! Where's my OpenWebBrowser!? Why don't the developers care about MY needs, huh?
Seriously though, grow up. It's not the end of the world to use the package system. Do you think they would spend so much time maintaining it if you were supposed to avoid it? Is "pkg_add net-snmp" really so incredibly painful for you that it's easier for you to bitch about it?
Comments
By Anonymous Coward (85.178.86.78) on
>
> I know! It's almost as outrageous as how Apache is included in base, but I have to install Firefox as a 3rd party package! Where's my OpenWebBrowser!? Why don't the developers care about MY needs, huh?
>
> Seriously though, grow up. It's not the end of the world to use the package system. Do you think they would spend so much time maintaining it if you were supposed to avoid it? Is "pkg_add net-snmp" really so incredibly painful for you that it's easier for you to bitch about it?
You miss the topic!
to test your "basic" webserver you've LYNX
to test your retarded DNS you've dig/host
to test your fucking spam relay you've MAIL
but you don't have anything for snmpd.. at least I don't know the command to request snmp-stuff (even from other servers...)
Comments
By Anonymous Coward (2001:388:f000::8bb) on
> >
> > I know! It's almost as outrageous as how Apache is included in base, but I have to install Firefox as a 3rd party package! Where's my OpenWebBrowser!? Why don't the developers care about MY needs, huh?
> >
> > Seriously though, grow up. It's not the end of the world to use the package system. Do you think they would spend so much time maintaining it if you were supposed to avoid it? Is "pkg_add net-snmp" really so incredibly painful for you that it's easier for you to bitch about it?
>
> You miss the topic!
>
> to test your "basic" webserver you've LYNX
> to test your retarded DNS you've dig/host
> to test your fucking spam relay you've MAIL
>
> but you don't have anything for snmpd.. at least I don't know the command to request snmp-stuff (even from other servers...)
Whine, whine whine....
It's such a shit OS that it causes you grief?
Then piss off and use something really great and please don't come back here to let me know what it is. You will be whining at whatever your other choice is sooner or later.
Comments
By anonymous pedro (201.53.181.220) on
By Anonymous Coward (91.21.64.45) on
> > >
> > > I know! It's almost as outrageous as how Apache is included in base, but I have to install Firefox as a 3rd party package! Where's my OpenWebBrowser!? Why don't the developers care about MY needs, huh?
> > >
> > > Seriously though, grow up. It's not the end of the world to use the package system. Do you think they would spend so much time maintaining it if you were supposed to avoid it? Is "pkg_add net-snmp" really so incredibly painful for you that it's easier for you to bitch about it?
> >
> > You miss the topic!
> >
> > to test your "basic" webserver you've LYNX
> > to test your retarded DNS you've dig/host
> > to test your fucking spam relay you've MAIL
> >
> > but you don't have anything for snmpd.. at least I don't know the command to request snmp-stuff (even from other servers...)
>
> Whine, whine whine....
>
> It's such a shit OS that it causes you grief?
> Then piss off and use something really great and please don't come back here to let me know what it is. You will be whining at whatever your other choice is sooner or later.
>
Sometimes it seems Theo is the guru of some sect. It's disgustful to see such anti-social people in the web. Go take your shitty OS and hide with Theo on the highest mountain. Well at least he is able to show you the path.
By Anonymous Coward (85.158.45.32) on
> >
> > I know! It's almost as outrageous as how Apache is included in base, but I have to install Firefox as a 3rd party package! Where's my OpenWebBrowser!? Why don't the developers care about MY needs, huh?
> >
> > Seriously though, grow up. It's not the end of the world to use the package system. Do you think they would spend so much time maintaining it if you were supposed to avoid it? Is "pkg_add net-snmp" really so incredibly painful for you that it's easier for you to bitch about it?
>
> You miss the topic!
>
> to test your "basic" webserver you've LYNX
> to test your retarded DNS you've dig/host
> to test your fucking spam relay you've MAIL
>
> but you don't have anything for snmpd.. at least I don't know the command to request snmp-stuff (even from other servers...)
Oh, c'mon. There's 101 ways to query SNMP. Before snmpd there was approximately 1 way to respond to those queries on OpenBSD, and that sucked in many situations. Also it's another piece of software that must be updated in lockstep with certain kernel changes, which is somewhat annoying to do with ports/packages.
Given that developer resources are limited, which would you rather they worked on first:
1. The part which needs to run on every machine that needs to be monitored, where the current implementation has various problems,
-or-
2. The part where there are a number of alternative implementations which don't work too badly, only need to run on a limited number of management stations, and don't have any special kernel dependency.
Comments
By Anonymous Coward (67.19.172.58) on
> > >
> > > I know! It's almost as outrageous as how Apache is included in base, but I have to install Firefox as a 3rd party package! Where's my OpenWebBrowser!? Why don't the developers care about MY needs, huh?
> > >
> > > Seriously though, grow up. It's not the end of the world to use the package system. Do you think they would spend so much time maintaining it if you were supposed to avoid it? Is "pkg_add net-snmp" really so incredibly painful for you that it's easier for you to bitch about it?
> >
> > You miss the topic!
> >
> > to test your "basic" webserver you've LYNX
> > to test your retarded DNS you've dig/host
> > to test your fucking spam relay you've MAIL
> >
> > but you don't have anything for snmpd.. at least I don't know the command to request snmp-stuff (even from other servers...)
>
> Oh, c'mon. There's 101 ways to query SNMP. Before snmpd there was approximately 1 way to respond to those queries on OpenBSD, and that sucked in many situations. Also it's another piece of software that must be updated in lockstep with certain kernel changes, which is somewhat annoying to do with ports/packages.
>
> Given that developer resources are limited, which would you rather they worked on first:
>
> 1. The part which needs to run on every machine that needs to be monitored, where the current implementation has various problems,
> -or-
> 2. The part where there are a number of alternative implementations which don't work too badly, only need to run on a limited number of management stations, and don't have any special kernel dependency.
Sure there plenty of ways to do so. I just say none is provided by OpenBSD (nc doesn't count) for now.
I can check every server they deliver (of for the routing it is a littlebit harder, i angree) with tools from the base-system.
And snmp is "more" importent. SO where it the benefit to have a OpenBSD with a snmp and using a Windows snmp-Browser or being forced to install 3rd party software on OpenBSD.
I can provie you more examples:
sshd -> ssh
cvs -> cvs
sendmail -> mail
httpd -> lynx
dhcpd -> dhcpclient
ntpd -> ntp
ftpd -> ftp
tftpd -> tftp
fingerd -> finger
*cut*
You see for kinda any service there's a programm to query it.
Just for snmp there should be no application to do so?
There's just one client application wich has no matching server: telnet (but rshd is still provided.. I mean "omg"...rly. Wasn't tlenetd kicked off because of security reasons? anyway that's not the topic here).
So why is snmp more complex then any of the above services wich do all have a client application as well (and not all where "imported").
Comments
By Anonymous Coward (68.227.223.107) on
> I can provie you more examples:
>
> sshd -> ssh
> cvs -> cvs
> sendmail -> mail
> httpd -> lynx
> dhcpd -> dhcpclient
> ntpd -> ntp
> ftpd -> ftp
> tftpd -> tftp
> fingerd -> finger
> *cut*
>
> You see for kinda any service there's a programm to query it.
There are many reasons something isn't in base even though the forces of logic would suggest its placement.
Perhaps the plan is to write a nice snmp program after they get the host daemon working? Remember, it is a very resource-limited project - things get done in an order, not all at once. Plus, the order is their choice, not ours. The more nuanced response is : ask nicely and maybe someone will tell you; maybe you're the bleemin' genius that thought it up and they'll get right on it.
Comments
By Anonymous Coward (85.178.107.251) on
>
> > I can provie you more examples:
> >
> > sshd -> ssh
> > cvs -> cvs
> > sendmail -> mail
> > httpd -> lynx
> > dhcpd -> dhcpclient
> > ntpd -> ntp
> > ftpd -> ftp
> > tftpd -> tftp
> > fingerd -> finger
> > *cut*
> >
> > You see for kinda any service there's a programm to query it.
>
> There are many reasons something isn't in base even though the forces of logic would suggest its placement.
>
> Perhaps the plan is to write a nice snmp program after they get the host daemon working? Remember, it is a very resource-limited project - things get done in an order, not all at once. Plus, the order is their choice, not ours. The more nuanced response is : ask nicely and maybe someone will tell you; maybe you're the bleemin' genius that thought it up and they'll get right on it.
Well WHY should I or anybody help 'em?
OpenBSD becomes in fact more and more useless and unimportent.
OpenBSDs res. are limited? Yes that's damn true but WHY?
From my point of view the developers are pretty responseable to this shortcome too!
Even NetBSD has MORE manpower/money then OpenBSD.. good job!
But I wont troll but tell you you my oppinion:
OpenBSD developers lost the contact to their userbase.
They love to kick off fancy uberprojects wich later get forgetten, not finished or whatever because they "loose" interrest.
Sure they do coding for fun and in their FREE TIME. Yes.. but hey if I walk along with my GF I don't make evey decission nor do I seriously love any decission being made but I accapt the decission 'course I love her...
Is it so hard to do the same with OpenBSD? Most developers claim they like OpenBSD a lot but I see just a few showing this seriously. All others should go an fork a retarded Linux kernel for their ubergreat patches...
1. OpenBSD lost the contact to it's userbase and main-sponsor: US, the users...
No manpower? Well Port-Updates don't even get included even if a "user" made the job.
This leads us to:
2. Not everybody can have CVS access
Well bullshit.. wasn't OpenCVS planed to make this possible?
Wich leads us to:
3. Projects take a shitload of time because they get kicked off and never finished
OpenCVS is "comming soon"...
rthreads may NEVER come...
And wow.. OpenBSD has now chroot for sftp and foo? Wasn't this possible with patches years ago? I mean WTF?!
What does me piss off and stops me from sending HW/Money or buy CD Sets is:
Even if I do a portupgrade it wont get included...
There NO security patches for the STABLE Ports-Tree no matter if a dev. with CVS access just would have to submit it....
OpenBSD offen starts porjects, tools or campaigns and then fails to fullfill what they promised/said
GNU tools in base system? YOu said you _wont_ include 'em but HELL WHERE IS THIS RETARDED CD BURNING comming from?
rthreads wont get finished 'course nobody cares...
It was the same with OpenCVS until recently...
Kernel pppoe versus userland pppoe? Wasn't the kernel-foo supposed to replace the userland? Now I#ve the choice if I either: connect a small office via the userland pppoed where my FW gets reloaded automaticaly but the whole connection is fucking slow or the kernel pppoe where I get the speed but have to reload the FW rules each time the ISP kicks us out (once every 24h)...
So is OpenBSD useable for me as FW? fuck no... not anymore...
I'm not living in the USA where cable-connections do rule. I'm from europe and we use DSL...
It's because the IP and co changes.. OpenBSD pppoe can't even get the DNS-Servers from the ISP.... it's all like half done...
It feels like most developers do think like this: "Hey it works for what _I_ do need so lets stop it and start off something else".
Of course that's their right and I wont critisize them directly for this but the phillosophy of OpenBSD was always a different one: Do it right, do it secure.
Linux developers do develop like this.. and I see the same problem rising here. "Oh it does what I need so hey lets stop". No matter if the code has a better quality or not the overall "project" (like kernel pppoe, rthreads?!) gets done halfway....
I'm pissed off to read things at undeadly WICH NEVER COME.
Seriously: How many things you've read here became true lately?
I remember the intel WLAN driver as one of the latest things.
rthreads? How great was the storry and the blabla about ti and what happened? kernel pppoe? Better SMP implementations? A better IP Stack (OpenBSDs is just ugly.. seriously :/ ).
So how can I convience my boss to spend some money to the project?
I can hardly convience myself to buy CD sets further.. 'course I've the same retarded problem at home!
I mean... long ago (from ~2.6) I choosed OpenBSD because it was ahead..
It was FUN to use it.... FUN to support it.... FUN in every way because mostly OpenBSD was ahead in point of security. Now not even this is true anymore..
Why do I need a ubersafe OS if my CCs get stolen via a Firefox hole?
Or why do I need a ubersavfe Mailserver if it crashs everytime a modified packets gets received by clamav?! Oh.. does anybody filters out the most common crap with clamav? A personal question: Do you also use personal "stable" Ports because none was provided?
Well I don't use CCs anyway but you get my comparsion hopefully.
I love OpenBSD.. and I'm loyal.. but it doesn't need a wizzard to see that something is gonna get done in the wrong way.
Sure the developers do all the stuiff in their free time and they give it away for free and for this I thank _all_ of them very much!
But if you see the overall situation more strategicly you'll see OpenBSD looses grounds and other Systems take over places OpenBSD holds.
I seriously wont bitch and I'll explain why I think so:
What do you think are the most computers used for today?
Routers? Firewalls? "Sendmail servers"?
I would guess more users sit in front of workstations or home PCs.
I wont kick off a "OpenBSD needs to rule the desktop"-campaign (even OpenBSD has BETTER chances then Linux (if you forget 3D and ISDN...))
What do I need personal?
- Better pppoe implementation (maybe one where I don't have to reload my FW rules because the IP changed?! Just as a hint..)
And what would I need at work?
A OS to count on....
Not a OS where I#ve to update every shit by hand. Of course I cans cript it but I can't spread OpenBSD nowhere if my boss asks me "And how's about the updates" "Oh sure.. updates get made by myself in my work/productiv time so please gimme 10hrs a month just for this...".
I'm working for a rly big big big MS partner. We'Ve JUST MS-retarded Servers here and most of them do suck after a while...
But hey it's partly my job to keep the crap running.
We've now ONE OpenBSD server... ONE.....
And it has NO internetconnection and works just as SMB Server in a seperated LAN.
Why no connection? Because there wont be updates for STABLE
Why didn't I choosed current? Because I'm not nuts and I don't have the time to spend 10hrs a month just because the server might crashed or something like this.
Why samba? It's used as deployment server.... for the PCs the company ships. And it's doing a good job...
But I am not able to recomment it for anything else....
Proxy? Well what should I use??? The default Apache-Proxy?!
SMTP-Filtering? And risking that it hangs because of outdatd packages? Or even it works fine: How do I get clamav updates regulary?! No way...
FW? Sure.. but there Sonicwalls right now. Of course I can convience my boss to try it but this gets HARD. There's just ONE server.. ONE server... and trust needs to get build.
Most of the dudes working there HATE VIsta and partly 2008-Server and they're pissed by MS.. so OpenBSD has SERIOUSLY A CHANCE to cover work wich was done by MS-Servers....
I can't name the company and it's just my impression fromt he talks I had.. and of course it's my oppinion.
It's your project..your free time....
But I'm loyal to OpenBSD, like a lot other users. YOu just decided to kick us out because you "do it for yourself" except for "OpenBSD"...
Comments
By Wouter (2001:888:10:b6b::2) on
>
> Well bullshit.. wasn't OpenCVS planed to make this possible?
You have CVS read access. See http://www.openbsd.org/anoncvs.html. You just can't commit to CVS. And looking at the rest of your post, I think that's a good thing.
> [blah blah blah f*cking blah]
> GNU tools in base system? YOu said you _wont_ include 'em but HELL WHERE IS THIS RETARDED CD BURNING comming from?
Why don't you write a BSD-licensed burning app then? Oh wait, you're probably too stupid to do that, all you can do is whine and blabber on about your love-hate relationship with OpenBSD.
You're not entitled to anything you know. No OpenBSD developer is under any obligation to help you. If they do, it's from the kindness of their hearts. And from your writing style I can easily see why they don't.
> So is OpenBSD useable for me as FW? fuck no... not anymore...
> I'm not living in the USA where cable-connections do rule. I'm from europe and we use DSL...
So? I live in Europe too, The Netherlands to be exact. And I also use DSL. I just use a non-retarded provider and a non-retarded modem. And DHCP really does fine wrt. DNS servers.
> It's because the IP and co changes.. OpenBSD pppoe can't even get the DNS-Servers from the ISP.... it's all like half done...
Why not write a patch then? Oh no, wait, you're too lazy and stupid to do so.
> I remember the intel WLAN driver as one of the latest things.
If Intel won't provide documentation with which to _write_ a driver it ends there and then.
> A OS to count on....
> Not a OS where I#ve to update every shit by hand.
> [more blah blah cut]
> I'm working for a rly big big big MS partner. We'Ve JUST MS-retarded Servers here and most of them do suck after a while...
> But hey it's partly my job to keep the crap running.
>
> We've now ONE OpenBSD server... ONE.....
> And it has NO internetconnection and works just as SMB Server in a seperated LAN.
>
> Why no connection? Because there wont be updates for STABLE
So? Go away! Go use Debian or something! They spend so much time updating their current system they can only muster a major release every aeon.
At least OpenBSD releases are regular as clockwork. So once every 6 months you get the latest and greatest, no delays. And I think that's great.
Comments
By Anonymous Coward (85.179.38.66) on
I've no time if I keep Servers and Networks running.
But it's great to see you've plain facts you can talk off and you don't start trolling... *takes a photo*.
> You're not entitled to anything you know. No OpenBSD developer is under any obligation to help you. If they do, it's from the kindness of their hearts. And from your writing style I can easily see why they don't.
You didn't read my post carefully. You're just full of anger or whatever..
> > So is OpenBSD useable for me as FW? fuck no... not anymore...
> > I'm not living in the USA where cable-connections do rule. I'm from europe and we use DSL...
>
> So? I live in Europe too, The Netherlands to be exact. And I also use DSL. I just use a non-retarded provider and a non-retarded modem. And DHCP really does fine wrt. DNS servers.
You must be just a lot smarter then me... that's it! :/
> > It's because the IP and co changes.. OpenBSD pppoe can't even get the DNS-Servers from the ISP.... it's all like half done...
>
> Why not write a patch then? Oh no, wait, you're too lazy and stupid to do so.
Because nobody cares. A developer would have to check it and nobody has either pppoe or is interested so I simply gave up. I tried submitting something 2-3 times (well since 2.6 I admit).
> > I remember the intel WLAN driver as one of the latest things.
> If Intel won't provide documentation with which to _write_ a driver it ends there and then.
You missunderstood my post completly...
> > A OS to count on....
> > Not a OS where I#ve to update every shit by hand.
> > [more blah blah cut]
> > I'm working for a rly big big big MS partner. We'Ve JUST MS-retarded Servers here and most of them do suck after a while...
> > But hey it's partly my job to keep the crap running.
> >
> > We've now ONE OpenBSD server... ONE.....
> > And it has NO internetconnection and works just as SMB Server in a seperated LAN.
> >
> > Why no connection? Because there wont be updates for STABLE
>
> So? Go away! Go use Debian or something! They spend so much time updating their current system they can only muster a major release every aeon.
You don't get it or? Why is Linux "bigger" then OpenBSD? Is it better?
No.. Does it has the better license? No... but companies use it.
And companies tend to donate sometimes MONEY wich is needed for HACKATRONS and where people like YOU do develop.
You just see black'n'white but not the shades of gray out there nor the colors.
> At least OpenBSD releases are regular as clockwork. So once every 6 months you get the latest and greatest, no delays. And I think that's great.
That's not realy true.
In fact OpenBSD 4.2 was "tagged" and finished nearly 1.5 months before the official release date.
But I wont critisize this.
A "stable" OS with no regular software updates is as usefull as a unpatched WIndows 2k3 Server connected to the Internet via fiber optic...
If you're about "you get it asap + if it's done" then tell me why it takes ages to get Xorg patches or such stuff for stable.
If you're realy so full of anger and pissed about me that you don't see the disadvantages or negatives then just focus on the base-system. Where are the latest PCRE-Patches?! And I don't talk about the hting they fixed like 2 months ago or so.
So if you like flaming me do so. YOU do not have the money to tell Theo "Hey dude here are 30k USD go and make a hackatron".
Neither do I#ve so much money but companies do have it.
Open your eyes....
Nothing i _just_ good nor _just_ bad so be open for critical oppinions otherwise this project dies. If pople would stuck into their own oppinion there wont be any progress.
And it's your attitude wich pisses me off and makes me NOT submitting anything (not even bugreports).
You just take every critic, every critical word as whinning or flaming or whatever.
So:
First off you didn't read carefully. The story about the intel WLAN driver was the last thing wich was announced on undeadly wich came true (got into stable.. i read a lot about rthreads but where are they?! Just check the archives and compare... undeadly became a "if we could/would then we might get..."-Website except a Website for serious news).
And the second point: You missunderstood the post completly...
So feel free to repost.
Btw:
If all people would go away your OS would just disappear (no dev meetings, no nice CVS servers.. theo may wouldn't be able to even pay the energy bills..). But you don't get it...
A user belongs to you as you (as project) belongs to the user.
If you disappear the user has no OS anymore and if the user disappears you wont have the power to develop it further... (not like you do right now and that's something even you should have noticed already or do you life in a country where money is meaningless?! If so: I had no clue.. let me join...)
Comments
By Anonymous Coward (76.250.126.209) on
If you patch gets rejected, whose fault is that? You blame blame blame and whine whine whine and I have not seen a single constructive piece of criticism. You know, PPPoE works fine for me; including changing IP addresses. I have no idea what you are doing wrong (I do but I won't indulge).
I'd like for you to read your postings (which are painful due to all the spelling errors but what the hell) and realize what you really are saying.
Comments
By Anonymous Coward (85.178.65.219) on
>
> If you patch gets rejected, whose fault is that? You blame blame blame and whine whine whine and I have not seen a single constructive piece of criticism.
That's not true
If the developer/guy has no time (I was told this at leats once) wich fault is it? Are you nuts? It's not my fault and not his.. just that doesn't change the fact that something wasn't updated.
> You know, PPPoE works fine for me; including changing IP addresses. I have no idea what you are doing wrong (I do but I won't indulge).
Well I would be happy if you would tell me the thing I did wrong. I accapt to make misstakes I just need to identify 'em to prevent that I'll do 'em further.
> I'd like for you to read your postings (which are painful due to all the spelling errors but what the hell) and realize what you really are saying.
Well learn my language.. to exspect that everybody speaks english perfectly is rly "Omg"...
But that's not the point here so please don't take it as "flaming" or so.
Comments
By Anonymous Coward (76.250.126.209) on
I am a Portugese born Dutch citizen. My native tongue is not English yet I am somehow able to run a spelling checker. I also try to improve my English skills over time.
I hate to tell you this but German is not the language of the internet; English is. If you want to participate you should try to be a good citizen.
Comments
By Anonymous Coward (76.7.248.160) on
>
> I am a Portugese born Dutch citizen. My native tongue is not English yet I am somehow able to run a spelling checker. I also try to improve my English skills over time.
>
> I hate to tell you this but German is not the language of the internet; English is. If you want to participate you should try to be a good citizen.
I LIKE TURTLES!!!
By Anonymous Coward (85.158.45.32) on
Haven't got time? No, you just choose to use your time on other things. Like posting here.
There are quite a few developers who do the same sort of work, and have similar time pressures. Yet somehow they manage to find the time to make their code suitable for other people to use. Perhaps the time you spend complaining about things, they spend on actually doing something about it.
> > > It's because the IP and co changes.. OpenBSD pppoe can't even get the DNS-Servers from the ISP.... it's all like half done...
Well, you're the one who has been most vocal about this particular thing. But strange, I don't see any reports from you about Peter Philipp's diffs to pppoe(4) and ifconfig(8) which, ermm, report the DNS servers returned from the peer back to userland so you can do something with them...
> Because nobody cares. A developer would have to check it and nobody has either pppoe or is interested so I simply gave up. I tried submitting something 2-3 times (well since 2.6 I admit).
Developers with commit access have to check things all the time. See where it says "ok" in many of the commit messages? There's a whole bunch of careful reading and a whole bunch of testing behind those two letters. Of course there are developers using pppoe(4). Maybe they just think their ISP's caching resolvers suck.
By Anonymous Coward (213.41.185.88) on
> What does me piss off and stops me from sending HW/Money or buy CD Sets is:
>
> Even if I do a portupgrade it wont get included...
Huh ? we updated hundreds of ports over the last six months.
Or you mean a portupgrade tool ? what for ?
pkg_add -u works just fine, and does exactly what you want.
We have a working ports tree with update capabilities, and things.
It's getting better all the time. The next step is probably to convince
libtool to stop fucking up and easy updates.
> OpenBSD offen starts porjects, tools or campaigns and then fails to fullfill what they promised/said
Well, I can speak only for myself, but so far I've done what I said I would.
- GNU autoconf works with our base m4, has for years.
- the current batch of pkg tools just work, *including* update capabilities.
- parallel make is usable. It can build src and X11 and quite a few ports in parallel... there are two issues left that will be fixed after 4.3. And it works like it should: namely you don't have to patch makefiles to cater for bugs in it.
Comments
By Anonymous Coward (213.41.185.88) on
> Well, I can speak only for myself, but so far I've done what I said I would.
And of course, I forgot to sign in, but hey, you can probably figure out who I am...
Comments
By Anonymous Coward (85.178.65.219) on
> > Well, I can speak only for myself, but so far I've done what I said I would.
>
>
> And of course, I forgot to sign in, but hey, you can probably figure out who I am...
Sure I can :-)
Well it was no critic about the ports (tools)!
It more about your "politic". F.e. there where (none?) just realy few updates for 4.2-STABLE. Xvid f.e. was never updated even it's exploitable.
So of course there ARE updates. That#s not what I do miss. But for 4.2 there where kinda none real updates so far. :-/
It's just I can't wait 6 months until clamav gets update because in the emantime somebody (a worm propably9 exploits it wich leads at least to a crash of my mailgateways. Sure I can do update it by myself but hopefully you get what I wanted to point out. It can't be the ubergoal to download each tgz yourself and compile it at your own (let the ports do it! And those tools so far are great! Also the packages tools... :) ). :)
Comments
By Marc Espie (213.41.185.88) espie@openbsd.org on
> It more about your "politic". F.e. there where (none?) just realy few updates for 4.2-STABLE. Xvid f.e. was never updated even it's exploitable.
In short, not enough people we trust, no reliable process to get new developers quickly enough, plus the fact we want to keep having fun.
As it is, I could spend all my free time doing administrative stuff for OpenBSD, I already lag behind in code review seriously.
Instead, I choose to do some fun things, and some not so fun things.
This is a chronic problem. Quite simply, there are areas I don't have any time for, like dealing with -stable branches. Apparently, no-one does.
If people do show up, and do things correctly, with the usual try at high quality standards we enjoy in the project, they can help.
They need to have a thick hide, because of the specific `socialization' of the project, but it is a neat environment to work in. As long as you don't refuse criticism, and realize we lack time, things will go forward.
By Can Acar (66.75.248.152) canacar@ on
>
> Well bullshit.. wasn't OpenCVS planed to make this possible?
This comment shows you have no clue. OpenCVS planned to be a complete replacement of GNU CVS. Not to provide extra access to anything. In fact, OpenBSD CVS tree has been open (read only) since the beginning. It is not going to change after OpenCVS becomes the standard CVS program.
[snip more rant]
> Even if I do a portupgrade it wont get included...
Looking at the quality of your writings, it is no surprise some people are not taking you seriously enough to consider importing your work.
> There NO security patches for the STABLE Ports-Tree no matter if a dev. with CVS access just would have to submit it....
Maintaining a stable ports tree requires the responsible developer to maintain old versions of OpenBSD on EVERY supported architecture. This is lots of hardware/space/power/effort. Otherwise it is simply not possible to test -stable ports and generate the packages.
It is not just a matter of having CVS access. It is a lot more.
> OpenBSD offen starts porjects, tools or campaigns and then fails to fullfill what they promised/said
>
> GNU tools in base system? YOu said you _wont_ include 'em but HELL WHERE IS THIS RETARDED CD BURNING comming from?
Try cdio(1) in base some time. It *can* burn CDs you know.
> rthreads wont get finished 'course nobody cares...
> It was the same with OpenCVS until recently...
Threads work in OpenBSD just fine. If you want more/better functionality fix it, or get someone else to fix it for you. Hint: trolling/ranting in lists/forums is not the way to do it.
> Kernel pppoe versus userland pppoe? Wasn't the kernel-foo supposed to replace the userland? Now I#ve the choice if I either: connect a small office via the userland pppoed where my FW gets reloaded automaticaly but the whole connection is fucking slow or the kernel pppoe where I get the speed but have to reload the FW rules each time the ISP kicks us out (once every 24h)...
No, as long as you have the correct ruleset, kernel pppoe changing IP addresses works perfectly with pf. I know because I am the maintainer of kernel pppoe, and have been using it that way for years. Hint: use (pppoe0) in your pf rules (the braces '()' are the key).
Again it seems you are throwing accusations without knowing what you are talking about.
> So is OpenBSD useable for me as FW? fuck no... not anymore...
> I'm not living in the USA where cable-connections do rule. I'm from europe and we use DSL...
> It's because the IP and co changes.. OpenBSD pppoe can't even get the DNS-Servers from the ISP.... it's all like half done...
Quick solution: Either use a local DNS cache, or hard code the DNS server to your resolv.conf.
None of the patches I received for getting DNS from the ISP was clean/correct enough. They were always some dirty hacks, so none of them were put into the tree.
Hint: only very few OpenBSD developers live in US. Most are concentrated in in Europe and Canada I suppose.
[snip more clueless rant]
> It was FUN to use it.... FUN to support it.... FUN in every way because mostly OpenBSD was ahead in point of security. Now not even this is true anymore..
>
> Why do I need a ubersafe OS if my CCs get stolen via a Firefox hole?
> Or why do I need a ubersavfe Mailserver if it crashs everytime a modified packets gets received by clamav?! Oh.. does anybody filters out the most common crap with clamav? A personal question: Do you also use personal "stable" Ports because none was provided?
The fact that the focus on security shifted from 'increasingly getting harder to exploit' server stuff back in 90's to new more fertile fields like exploiting clients takes away all the fun right? But occasional stuff, like the latest and greatest series of Linux kernel local privilege elevation exploits reminds us that there are other things than XSS and SQL injection still out there.
With all the buggy 3rd party software around it is a really hard job to keep the ports secure. I personally use -current, and update my ports to keep up to date. This way -current gets more testing, and this is also one of the reasons (in addition to lack of resources) that stable ports are not done any more. If you want to use the latest in ports, use -current. period.
> Well I don't use CCs anyway but you get my comparsion hopefully.
>
> I love OpenBSD.. and I'm loyal.. but it doesn't need a wizzard to see that something is gonna get done in the wrong way.
>
> Sure the developers do all the stuiff in their free time and they give it away for free and for this I thank _all_ of them very much!
>
> But if you see the overall situation more strategicly you'll see OpenBSD looses grounds and other Systems take over places OpenBSD holds.
>
> I seriously wont bitch and I'll explain why I think so:
>
> What do you think are the most computers used for today?
> Routers? Firewalls? "Sendmail servers"?
>
> I would guess more users sit in front of workstations or home PCs.
> I wont kick off a "OpenBSD needs to rule the desktop"-campaign (even OpenBSD has BETTER chances then Linux (if you forget 3D and ISDN...))
>
> What do I need personal?
>
> - Better pppoe implementation (maybe one where I don't have to reload my FW rules because the IP changed?! Just as a hint..)
It already works that way. Even if it did not, I do not see a bug report from you statin it either. The correct way of reporting bugs is not ranting in mailing lists/forums.
OpenBSD is not a shrink-wrap one size fits all platform. It is a framework where you roll out your solutions, mixing and matching parts, setting up systems to support your infrastructure. If you are too lazy/clueless to research/learn/implement it do not expect others do your work for you.
> And what would I need at work?
>
> A OS to count on....
> Not a OS where I#ve to update every shit by hand. Of course I cans cript it but I can't spread OpenBSD nowhere if my boss asks me "And how's about the updates" "Oh sure.. updates get made by myself in my work/productiv time so please gimme 10hrs a month just for this...".
>
> I'm working for a rly big big big MS partner. We'Ve JUST MS-retarded Servers here and most of them do suck after a while...
> But hey it's partly my job to keep the crap running.
>
> We've now ONE OpenBSD server... ONE.....
> And it has NO internetconnection and works just as SMB Server in a seperated LAN.
>
> Why no connection? Because there wont be updates for STABLE
> Why didn't I choosed current? Because I'm not nuts and I don't have the time to spend 10hrs a month just because the server might crashed or something like this.
OpenBSD -current is more stable then most other software out there claiming to be the 'stable'
> Why samba? It's used as deployment server.... for the PCs the company ships. And it's doing a good job...
>
> But I am not able to recomment it for anything else....
> Proxy? Well what should I use??? The default Apache-Proxy?!
> SMTP-Filtering? And risking that it hangs because of outdatd packages? Or even it works fine: How do I get clamav updates regulary?! No way...
>
> FW? Sure.. but there Sonicwalls right now. Of course I can convience my boss to try it but this gets HARD. There's just ONE server.. ONE server... and trust needs to get build.
>
> Most of the dudes working there HATE VIsta and partly 2008-Server and they're pissed by MS.. so OpenBSD has SERIOUSLY A CHANCE to cover work wich was done by MS-Servers....
>
> I can't name the company and it's just my impression fromt he talks I had.. and of course it's my oppinion.
> It's your project..your free time....
> But I'm loyal to OpenBSD, like a lot other users. YOu just decided to kick us out because you "do it for yourself" except for "OpenBSD"...
I use OpenBSD on des
Comments
By Anonymous Coward (66.75.248.152) on
> > I can't name the company and it's just my impression fromt he talks I had.. and of course it's my oppinion.
> > It's your project..your free time....
> > But I'm loyal to OpenBSD, like a lot other users. YOu just decided to kick us out because you "do it for yourself" except for "OpenBSD"...
>
> I use OpenBSD on des
I use OpenBSD on desktop for more than 7 years (and almost nothing else on my personal computers). I used it during my PhD as my main development workstation, (I did simulations, OpenGL and parallel processing stuff). I used it to implement firewalls and servers. Even when I am developing software for other platforms, I try to do tests on OpenBSD, as it tends to catch some class of bugs more easily.
I recently had to use a FreeBSD 6.3 -stable desktop. I found the ports management to be a real pain compared with OpenBSD. Yes I know about portupgrade and various other ways to mess with ports in FreeBSD, and I can probably learn to manage it better in time. However, with OpenBSD it is soo simple and it just works.
I used Debian for a couple of years on desktop at work. It is also quite usable, but you really do not get updates if you are running -stable. You need to run -testing to get decent updates. Believe me, OpenBSD -current is more stable than Debian -testing.
Right now I am slacking as an OpenBSD developer. Too much things to do, too little time...
PS: waiting for your pppoe bug report
Can
By Anonymous Coward (206.248.190.11) on
Oh noes! OpenBSD is causing landslides!
By Anonymous Coward (71.255.109.177) on
> OpenBSD and "speed" is like Linux and security.. it just does not match...
>
>
> This has been said to be true for a long while, but why does it have to be so? I wish that could be addressed. It would be especially cool if the performance could be improved with multi-threading and multi-core CPU's. It would be awesome to see OpenBSD become synonymous with performance as it is with security.
>
There basic problem is that doing virtualization, SMP, and high-performance networking stuff is that it adds a tremendous amount of operational complexity, which can be difficult to secure.
For example, FreeBSD responds to a massive volume of socket() calls by preallocating file descriptors; this means the kernel has to do less work for each individual socket() call, drastically improving performance. However, from a security standpoint, what happens if you try to read or write on a preallocated socket? What is the "correct" behavior in this situation, and how do you make sure that happens?
Similarly, without SMP a computer is essentially a state machine, so you can make assumptions about when certain interrupts are possible. OTOH, the further you drift into perfect SMP, the more state machine illusion breaks down, and the more you have to re-check things that were previously secure. If security is not your #1 priority, you can just hack something together like A Certain Other Free Operating System, and then fix bugs as they show up.
And lastly, with virtualization, not only does the host system need to be fully secure, but the virtualization layer does too. That's ANOTHER thing that has to be checked for security, ANOTHER component which can fail.
In summary, these things aren't impossible for OpenBSD, but any second-year Comp Sci student will tell you something must be correct before it can be optimized.
Comments
By anonymous pedro (201.53.181.220) on
That's not the problem. That's the solution. The problem is not having enough people willing to do it, or even trying to.
Comments
By Anonymous Coward (85.178.107.251) on
>
> That's not the problem. That's the solution. The problem is not having enough people willing to do it, or even trying to.
That's partly the fault of the OpenBSD developers who screw most people off...
So take a look at the mirror sometimes...
Comments
By Anonymous Coward (71.237.209.121) on
> >
> > That's not the problem. That's the solution. The problem is not having enough people willing to do it, or even trying to.
>
> That's partly the fault of the OpenBSD developers who screw most people off...
I got hella flamed for being a dumb newbie when I first came
to OpenBSD. just look in the archives. now I have an account on cvs.
sorry, but most people who complain that OpenBSD devs are mean are just cry-babies.
By Anonymous Coward (24.84.108.103) on
> > the whole data visualization was done with the help of OpenBSD ports.
> And OpenBSD and "virtualisation"? Are you nuts.. with qhat? qemu? Seriously....
Visualization != virtualization.
By Gordon Willem Klok (68.148.16.154) gwk@gwk.ca on
Comments
By chris cappuccio (198.175.14.193) on
You are missing my point. I don't care to read a negative article that damns openbsd!?!? That has nothing to do with the discussion that we SHOULD be having.
> Why would anyone want to read such a thing? If you have been around OpenBSD for any period of time you should already be aware that SMP performance is pretty much non existent , IO performance in general is not great, there are no good compilers nor support for high speed interconnects or RDMA in short its not even remotely the right tool for the job. Instead he chose to focus on where openbsd was helpful in this project or that it was at all, which should have been the surprising thing.
Dude I've been using OpenBSD since the first sparc kernel that Theo put up on ftp.theos.com in 1995. It had his sw (scsi wierd) driver that got 1MB/sec instead of a pathetic 100KB/sec from NetBSD on my Sun 4/110. Combine that with the paranoid security outlook which perfectly matches my own and I've been a user and (small) contributor ever since.
SMP performance just got a kick in 4.2-current on i386 by the way. Your example sucks. I/O performance is not great? How do we improve that? What needs to be done??
What in god's name is "interesting" about Chris' focus on how OpenBSD was "helpful" as a firewall? Hey, I tied my shoes this morning too.
The discussion that everyone avoids because for some stupid reason it got "forbidden," or people got shouted down too much for it, is why don't we support RDMA, or what other scalability issues can be solved here?
What are the I/O limitations that we see which Linux doesn't? Why? I don't want to have an "insanely negative story" about how "OpenBSD cannot be used for clusters" or anything even remotely close to it! It wouldn't kill people to occasionally talk about this positively, as in, where to go from here?
Perhaps the rest of us that didn't graduate in computer science from U Berkeley (or U Alberta :) aren't really going to have a top notch discussion about this sort of topic. We're all idiots who aren't qualified to talk because we haven't already written the code. Well fuck all, I don't care, it's STILL INTERESTING TO TALK ABOUT THE LIMITATIONS AND SEE IF WE CAN PUSH THEM, EVEN ONE AT A TIME!!!!
What is so horribly negative about this ???? It's not. Stop being a pussy and conforming to every last bit of bullshit that people keep buying into here.
Comments
By Anonymous Coward (85.178.86.78) on
>
> You are missing my point. I don't care to read a negative article that damns openbsd!?!? That has nothing to do with the discussion that we SHOULD be having.
>
> > Why would anyone want to read such a thing? If you have been around OpenBSD for any period of time you should already be aware that SMP performance is pretty much non existent , IO performance in general is not great, there are no good compilers nor support for high speed interconnects or RDMA in short its not even remotely the right tool for the job. Instead he chose to focus on where openbsd was helpful in this project or that it was at all, which should have been the surprising thing.
>
> Dude I've been using OpenBSD since the first sparc kernel that Theo put up on ftp.theos.com in 1995. It had his sw (scsi wierd) driver that got 1MB/sec instead of a pathetic 100KB/sec from NetBSD on my Sun 4/110. Combine that with the paranoid security outlook which perfectly matches my own and I've been a user and (small) contributor ever since.
>
> SMP performance just got a kick in 4.2-current on i386 by the way. Your example sucks. I/O performance is not great? How do we improve that? What needs to be done??
>
> What in god's name is "interesting" about Chris' focus on how OpenBSD was "helpful" as a firewall? Hey, I tied my shoes this morning too.
>
> The discussion that everyone avoids because for some stupid reason it got "forbidden," or people got shouted down too much for it, is why don't we support RDMA, or what other scalability issues can be solved here?
>
> What are the I/O limitations that we see which Linux doesn't? Why? I don't want to have an "insanely negative story" about how "OpenBSD cannot be used for clusters" or anything even remotely close to it! It wouldn't kill people to occasionally talk about this positively, as in, where to go from here?
>
> Perhaps the rest of us that didn't graduate in computer science from U Berkeley (or U Alberta :) aren't really going to have a top notch discussion about this sort of topic. We're all idiots who aren't qualified to talk because we haven't already written the code. Well fuck all, I don't care, it's STILL INTERESTING TO TALK ABOUT THE LIMITATIONS AND SEE IF WE CAN PUSH THEM, EVEN ONE AT A TIME!!!!
>
> What is so horribly negative about this ???? It's not. Stop being a pussy and conforming to every last bit of bullshit that people keep buying into here.
Well I can give you my "non student" point of view.
OpenBSD is not useable for "many" tasks because it plain "sucks". that sounds rude I know but let me explain: OpenBSD mostly starts kicking project wich then gets "lost" somehow
"Lets replace all GNU tool" and with what do we burn CDs?
"Lets make a good CVS tool" yeah.. "comming soon" became a running joke
better SMP handling? well you mentioned it.. in 4.2-current (how long did it took?)
Network speed on 10Gbit?!
"Dri"?
*threads?
not even cd9660 is fullly supported I mean... it feels somehow "halfdone" even the half is the most secure half of anything I can get...
why do I've to use ports to get a snmp-query tool if a snmp gets developed? why do I#ve to use GNU-tools to burn a CD even OpenBSD claims that it dislikes GNU tools in the base-system?
What else drives me nuts: "Patches for Ports should get into mainstream.."
Some ports stay at "stable" (like john the ripper! wich is stuck at 1.7.0.2 and the maintainer refused to update it to 1.7.2 even there OPENBSD-related updates.. SSE2-Codebase was developed on my OpenBSD box by Solar Designer personally...). Nobody wanna check in a sec. update for xvid-codec.... hail 1.0.3 realy.
1.7.2 is "unstable"..bullshit.. if that's true OpenBSD should have carried jtr 1.6 (not 1.6.34 or so.. 1.6 was "stable"..).
that are the things wich make me not wanna send updates to ports@ because nobody cares anyway so I keep my ports for myself. at least the 4-5 I need where the oBSD versiosn are just outdated...
Some may say "not everybody can get CVS access" but wasn't OpenCVS ment to handle the limitations of GNU CVS?
Why can't a normal user be a maintainer for a port without the need to be a uberprogrammer? Or even that's disliked: Why doesn't somebody gets name who cares for the packets wich do not have a official maintianer? like 5 guys who can get queried in case a update was done f.e. by me (a normal user).
Other projects have far more MONEY (yeah! MONEY... :/), manpower and co... OpenBSD does not have this but the things it owns get splitted in like 1k projects or different codes.
snmp will rock, I've no doubt about it.
but isn't updating the base http (or replacing it, henning still owns the domain..w00w00... openhttpd.org :/) better for the "most" people (don't know abour you but it doesn't look like all devs are billionairs who can give a fuck about support from their userbase...)?
sure you develop openbsd for yourself! yeah... but without the normal dudes you wont be able to play with it like you do.
what if nobody spends HW or money? realy "nobody"... I don't think hackatrons would be possible or 10Gbit support or so...
well but I'm just a retard... but oBSD is rly not useable for anything "high performence" and that's SAD.... but it's just a piece of the "problem". And hey.. " OpenCVS is to be released soon." (for about how many years now....? you ever noticed that linus wrote git in what? 2 months..)
Ok lately there's some progress too wich is great..
But... it's to be "released" soon for some years now...
http://www.openbsd.org/cgi-bin/cvsweb/www/opencvs/index.html
Take a look at Revision 1.1...
To be released "soon"...
Mon Dec 6 02:01:33 2004 UTC (3 years, 2 months ago)
So hopefully some dudes get the critic I'm talking about... if not.. hey I'm just a user...... or another retard.
Comments
By Anonymous Coward (219.90.169.23) on
By nobody (193.136.56.161) on
>
well, yes, you're retarded.
Are you using opencvs? are you submitting patches? bug reports?
Besides, openbsd developers are volunteers, they aren't payed. If you really need opencvs you can start paying for it. Otherwise it will take as much time as needed.
And this is not about "stop complaining and do something", is about being reasonable.
By Jonathan Thornburg (152.78.41.173) on http://www.soton.ac.uk/~jt1c06
> already be aware that SMP performance is pretty much non existent,
> IO performance in general is not great, there are no good compilers
> nor support for high speed interconnects or RDMA
The SMP is certainly a bit weak, but it's not "pretty much non existent".
It's BigLock, i.e. basically only one process can be in the kernel at any
given time. For CPU-bound applications that's often sufficient to get
excellent performance.
As to the compilers, this comment is simply wrong. Ports offers fairly
modern gcc (current 4.2-stable is gcc-4.2.20070307p2), including C,
C++, and Fortran 90, which together cover the main languages used for
high-performance scientific computing.
As to IO and high-speed interconnects... well, they're surely not
perfect, but they're quite adequate for CPU-bound applications doing
moderate interprocessor communication. Notably, I've had no problems
using MPICH in OpenBSD with the Cactus computational toolkit
(http://www.cactuscode.org), a toolkit for solving 3-D PDEs using
domain-decomposition parallelism.
I find OpenBSD to be a fine platform for my own work in scientific
computing (simulations of what happens when a small black hole
spirals into a large black hole). My current code is around 20,000
lines of C++. I've also used OpenBSD in the past to develop larger
parallel codes (using MPI for parallelism) in a mixed C/C++/F77/F90
environment.
And oh yes, since "visualization" was mentioned, I should say that
I find OpenBSD a fine platform for that, too. I generate all my
scientific movies on OpenBSD, usually using gnuplot (ports) to
render individual frames, netpbm (ports) to do things like swap
black/white backgrounds, and ppmtompeg (part of netpbm) to encode
frames into an mpeg. As usual in OpenBSD, it "just works".
Comments
By glthornberry (glthornberry) on
> perfect, but they're quite adequate for CPU-bound applications doing
> moderate interprocessor communication. Notably, I've had no problems
> using MPICH in OpenBSD with the Cactus computational toolkit
> (http://www.cactuscode.org), a toolkit for solving 3-D PDEs using
> domain-decomposition parallelism.
>
> I find OpenBSD to be a fine platform for my own work in scientific
> computing (simulations of what happens when a small black hole
> spirals into a large black hole). My current code is around 20,000
> lines of C++. I've also used OpenBSD in the past to develop larger
> parallel codes (using MPI for parallelism) in a mixed C/C++/F77/F90
> environment.
>
> And oh yes, since "visualization" was mentioned, I should say that
> I find OpenBSD a fine platform for that, too. I generate all my
> scientific movies on OpenBSD, usually using gnuplot (ports) to
> render individual frames, netpbm (ports) to do things like swap
> black/white backgrounds, and ppmtompeg (part of netpbm) to encode
> frames into an mpeg. As usual in OpenBSD, it "just works".
>
What you've described sounds more interesting than this HPC article. Maybe you could submit a more detailed article to Undeadly about your work and why you chose "our favorite OS" for the job? Seriously...
Comments
By Paul Greidanus (paul) on I won't bore you with a website now.
> > perfect, but they're quite adequate for CPU-bound applications doing
> > moderate interprocessor communication. Notably, I've had no problems
> > using MPICH in OpenBSD with the Cactus computational toolkit
> > (http://www.cactuscode.org), a toolkit for solving 3-D PDEs using
> > domain-decomposition parallelism.
> >
> > I find OpenBSD to be a fine platform for my own work in scientific
> > computing (simulations of what happens when a small black hole
> > spirals into a large black hole). My current code is around 20,000
> > lines of C++. I've also used OpenBSD in the past to develop larger
> > parallel codes (using MPI for parallelism) in a mixed C/C++/F77/F90
> > environment.
> >
> > And oh yes, since "visualization" was mentioned, I should say that
> > I find OpenBSD a fine platform for that, too. I generate all my
> > scientific movies on OpenBSD, usually using gnuplot (ports) to
> > render individual frames, netpbm (ports) to do things like swap
> > black/white backgrounds, and ppmtompeg (part of netpbm) to encode
> > frames into an mpeg. As usual in OpenBSD, it "just works".
> >
>
> What you've described sounds more interesting than this HPC article. Maybe you could submit a more detailed article to Undeadly about your work and why you chose "our favorite OS" for the job? Seriously...
I'll second that idea, I'd love to hear some more about this, including maybe some pretty pictures! :)
By iru (189.25.160.250) on
> > already be aware that SMP performance is pretty much non existent,
> > IO performance in general is not great, there are no good compilers
> > nor support for high speed interconnects or RDMA
>
> The SMP is certainly a bit weak, but it's not "pretty much non existent".
> It's BigLock, i.e. basically only one process can be in the kernel at any
> given time. For CPU-bound applications that's often sufficient to get
> excellent performance.
>
> As to the compilers, this comment is simply wrong. Ports offers fairly
> modern gcc (current 4.2-stable is gcc-4.2.20070307p2), including C,
> C++, and Fortran 90, which together cover the main languages used for
> high-performance scientific computing.
>
> As to IO and high-speed interconnects... well, they're surely not
> perfect, but they're quite adequate for CPU-bound applications doing
> moderate interprocessor communication. Notably, I've had no problems
> using MPICH in OpenBSD with the Cactus computational toolkit
> (http://www.cactuscode.org), a toolkit for solving 3-D PDEs using
> domain-decomposition parallelism.
>
> I find OpenBSD to be a fine platform for my own work in scientific
> computing (simulations of what happens when a small black hole
> spirals into a large black hole). My current code is around 20,000
> lines of C++. I've also used OpenBSD in the past to develop larger
> parallel codes (using MPI for parallelism) in a mixed C/C++/F77/F90
> environment.
>
> And oh yes, since "visualization" was mentioned, I should say that
> I find OpenBSD a fine platform for that, too. I generate all my
> scientific movies on OpenBSD, usually using gnuplot (ports) to
> render individual frames, netpbm (ports) to do things like swap
> black/white backgrounds, and ppmtompeg (part of netpbm) to encode
> frames into an mpeg. As usual in OpenBSD, it "just works".
>
It's very goood to know that.
I recently ported GROMACS (http://www.gromacs.org) for doing Molecular Dynamics stuff and found OpenBSD to 'just work' too.
Comments
By Paul Greidanus (paul) on I won't bore you with a website now.
> It's very goood to know that.
> I recently ported GROMACS (http://www.gromacs.org) for doing Molecular Dynamics stuff and found OpenBSD to 'just work' too.
>
Have you built it as a port? Can you submit it so it can me it into ports? I'd love to take a look at it.
Comments
By iru (189.25.207.172) on
> > It's very goood to know that.
> > I recently ported GROMACS (http://www.gromacs.org) for doing Molecular Dynamics stuff and found OpenBSD to 'just work' too.
> >
>
> Have you built it as a port? Can you submit it so it can me it into ports? I'd love to take a look at it.
>
already sent it to ports@ http://marc.info/?l=openbsd-ports&m=120183501308528&w=2
By Gordon Willem Klok (68.148.16.154) gwk@gwk.ca on
I replied in haste to the OP and I should have pointed out some facts about the hardware that were overlooked: durring the competition 6 of our nodes had 16GB of RAM, 2 had 24GB currently OpenBSD does not support more than 4GB of ram on amd64 yes there are diffs in the works for this they are not ready now never mind 6 months ago when we got our hardware. Without that kind of ram we would have lost, GAMESS (one of the applications we were required to run is very memory hungry (well it can trade runtime for memory but in a competition where you have 44 hours to complete as many data sets as possible...) we would have had to scale GAMESS out to get the memory it needed like some of our competitors GAMESS-UK using DDI DOES NOT SCALE WELL AT ALL ACROSS THE NETWORK further reducing performance.
Re: SMP
I am acutely aware that the kernel lives under a giant lock, and yes an entirely CPU bound load would perform OK on open, however back in the real world such embarrassingly parallel workloads are few and far between the apps we were required to run only POVRAY would have fit the bill (and due to the lack of affinity (xeons have a lot of cache after all) and the fact that povray would do a fair bit of IO it wouldnt have performed any where near as well as on linux. GAMESS which doesn't scale well off the node but does do well in parallel with shared memory wouldn't have performed well. Additionally giant locked kernel exhibit terrible speed ups particularly with our apps that required a tremendous amount of communication (MPI or the vanilla unix kind net result absent RDMA lots of system calls), contention for the one great lock is bad on a 2 way machine ours were 8 way it would have been *brutal*. Finally POP would have been a complete and utter disaster requiring more memory than any single node could provide. I made it hit 120GB here in Edmonton before we left for the conference on all 10 nodes (we could only use 8 durring the competiton because of the power limit). Pop was also insanely hungry for bandwidth: running on gigabit with all 80 cores (again we only had 64 for the competition) I had to kill a run after it chewed for an entire weekend: we only had 44 hours for the competition.
Re: Compilers
I said *GOOD* compilers, we benched GCC versus Intel: on our xeon processors the code GCC generated ran a minimum of 20% slower for C code, Fortran was worse.
Re: Interconnects:
When we were running POP with 4x DDR Infinibad (20Gbit and much lower latency than gigabit ethernet) inter-node communication still dominated to such extent that we could not keep the CPUs consistently busy, gigabit just wasn't going to cut it . Now this could very well be due to inefficiencies in the POP code however none of us was a domain expert and could have been counted on to correct these inefficiencies if they existed while still ensuring the scientific accuracy of the results... Which is all mute because OpenBSD does not even support Infiniband and the performance problems with the slower alternative 10Gbe is well documented:
http://www.openbsd.org/papers/cuug2007/mgp00008.html
(note none of the issues they discuss in the next slide related to 10GBe have been addressed since that talk and this inst a criticism of the networking guys)
Finally:
I agree whole heartily: openbsd can be used for a lot of visualization , basically its what the article should have been about and that the ports guys do a very good job of making sure software runs as advertised.
By Marc Espie (213.41.185.88) espie@openbsd.org on
> As to IO and high-speed interconnects... well, they're surely not
> perfect, but they're quite adequate for CPU-bound applications doing
> moderate interprocessor communication. Notably, I've had no problems
> using MPICH in OpenBSD with the Cactus computational toolkit
> (http://www.cactuscode.org), a toolkit for solving 3-D PDEs using
> domain-decomposition parallelism.
Hey, we need people like you.
Since it just works, how about writing ports, or helping to update mpich to a more recent version ?
Like I said in a sibling article, what we miss is people who spend time immersed in this stuff, and who actually *use* those tools.
By Bob Beck (129.128.11.43) beck@ualberta.ca on
> > [SNIP]
> >
>
> This article opens with summary text about "OpenBSD's role" in a "competition" in "High Performance Computing" only to actually explain how it was Chris' Linux cluster firewall...Please, what a joke. What the fuck are they using all these "Linux laptops" for? Ones that they apparently don't even know how to use? Why not OpenBSD? Where's the OpenBSD here? Why isn't the cluster running OpenBSD? What are the limitations in OpenBSD that need to be fixed for X cluster application that force you to run Linux? Why isn't there anything even remotely interesting in the article when the material is clearly there? Maybe it's group dynamics where the whole project was developed on Linux from the beginning? If so, why? What level of scalability is lacking from OpenBSD? Or from Linux? You can imagine the range of discussion that's missing from this article.
The real problem with this article, is actually the content. Chris didn't write anything about what they did but resorted to name dropping and a bunch of hyperlinks to other sites. In fact, the interesting part of what they did was the acutal visualization of output of a lot of th output from the cluster was done on OpenBSD machines - simply because the software to do so (molden, ffmpeg, etc.) worked reliably there.
Unfortunately chris spends all his time talking about a firewall - which while yes was there, is not news to anyone. Ports being much more stable and nice than rpm's, and heavy visualization stuff working well in the real world, that's "news" - but what is written in the article gives no
details on that.
Now, having said that, why wasn't OpenBSD run on the cluster? get real you guys. Clusters run scientific code written by academics. if you think you like code quality in general applications, try code quality in
applications written to solve specific problems by professors and grad qstudents running on one platform, and at the moment, guess what that is....... Secondly the name of the game is performance, not stability
and security. I *do not* run OpenBSD ANYWHERE for best performance. I run it for stability and security. Nobody gives a fuck if your linux cluster box chokes once in a while, or that the filesystem is asynchronous so when you take a reboot you spin the cylinders and put the gun to your head. Nobody cares as long as it's fast - most of the time. Could OpenBSD be faster if we made those sorts of design decisions, yes. Do we really want that? Well, I don't.
Could changes be made to OpenBSD to make it *more attractive* as a cluster platform, absolutely - starting with better threading. But
if you think for a minute that you're going to win a cluster race
this way you're fucking delusional. Scientific computing like that is
a different mindset from production computing with different goals.
I don't care if my individual boxes are 65% the speed of something else
if they are stable and secure - scientific computing cares all about that other number and frankly (althought they'll pay lip service to it, not a whit about the latter)
Comments
By anonymous pedro (201.53.181.220) on
Your whole argument is predicated on the assumption that there is an antagonism between security and performance.
I see no reason why clean, well structured code can't be secure *and* efficient.
Naturally, operating systems seeking one or the other will never have both. They will simply consider the job done once the code is safe or fast enough, and move on.
But why can't performance and security coexist as a goal?
Comments
By tedu (38.99.3.113) on
>
> Your whole argument is predicated on the assumption that there is an antagonism between security and performance.
no, his argument is predicated on the assumption that you want to build your cluster today and that you're going to use an operating system that's available today to do it, not some hypothetical awesome-os of the future.
Comments
By anonymous pedro (201.53.181.220) on
I believe Bob is not only talking about "clustered solutions today" when he makes such broad, timeless and rhetorical statements as:
"I *do not* run OpenBSD ANYWHERE for best performance."
and
"Could OpenBSD be faster if we made those sorts of design decisions, yes. Do we really want that? Well, I don't."
"ANYWHERE" to me, means, well, anywhere. And the second question does not seem to be restricted to the "clustered solutions today" domain at all.
Besides, what's the point of discussing anything if you can't raise a hypothesis?
By Amir S Mesry (66.23.227.241) on
Comments
By Mark in Dublin (62.231.57.115) on
Thank god u spelt that 2nd last word correctly.
By Marc Espie (213.41.185.88) espie@ on
> I was chris@ for a while, but I haven't had much to commit for some time. So, take it with a grain of salt. But I did contribute the initial altq port, vlan interface port, various other driver ports, mostly stuff that would add to openbsd's reputation as a "router platform" -- but the truth is that it is really a great unix platform and I think it's silly that people aren't talking about THAT.
We don't have anyone interested in cluster mpi stuff currently.
The only stuff we ever had was ported by yannick cote, and after a lot of prodding, yannick@ just vanished from the project.
This is stuff that actually needs people to port and to take care of it. Preferably people who grok parallel computations, and are able to really take charge of the code.
In my opinion, this is an area where OpenBSD is notoriously poor. It's not a big surprise, the few talented people who work on math and simulation stuff are chronically overworked, and they cannot port everything on the planet.
The OpenBSD part of porting that stuff is often fairly easy. But you have to keep up-to-date on the maths and CS stuff, and to actually know the application domain.
Also, a BIG proportion of that stuff is idiosyncratic. That is: it doesn't look at all like your typical unix software, and does often package and install in weird ways (csh install scripts, for instance...)
By Rich (195.212.199.56) on
I'm not knocking in any way at all what these guys have done, and if they want to use Linux then that's no skin off my nose - I'm certainly not going to beat them up about it. But yes, yet another article explaining how OBSD has been used as a firewall is ...well ...a bit dull? I think there's more than enough info out there on how to put together an OBSD firewall setup, and at the end of the day, it's not actually very interesting any more. I know I can't honestly say I find my own pf rules very exciting. And DHCP server's aren't really ...well ...sexy. I DO like a good ssh configuration though... phroaahhhhh... look at the certificates on THAT!
And just for the record, no, I have not made any significant public contribution to OBSD. I'm sorry. I've fixed some bugs in that truly atrocious code called 'qmail', though :-) Ok - it's not the same.... :-(
By nicram (78.8.51.126) webmaster@gumzamet.pl on http://nicram.sytes.net/
This is the reason that ppl don;t talk about problems that can be solved, because noone still believe that telling about it will change anything.
How many times OBSD devs say "it;s our project, we don't care what ppl want, we care what we want".
So this is the reaction. The second thing is that OBSD do not support such things like WPA, that every other OS got from years now. Where is the security that they say about in OBSD when there is no stupid WPA. There is no suppport for modern hardware. There are still some functions that do not work correctly, so some software is unstable (XMail). & yes, i were trying to help by testing it for many many times & talking with XMail author about any errors, debugging etc.
So the new developers are just, not interested in project that don't got some fundamentals for today world. I try to make OBSD more popular, so maybe some new pathes will arrive. But well...
Normal user asks me "why i may install mysql in linux & it works just fine, but when i install it on obsd i have to tweak many system things so it will be stable? how could i know what to change? do i really have to be geek?".
The only moment, when OpenBSD try to speak with userbase, it;s when it need some funds. & ppl don;t like it.
The situation will be worst, until devs will start to LISTEN what users needs. Because without users, there are no funds. Without funds, there is no OpenBSD :(
Comments
By Anonymous Coward (88.192.76.90) on
Comments
By Leonardo Rodrigues (201.88.85.48) on
Comments
By nicram (62.87.244.55) nicram@nicram.sytes.net on http://nicram.sytes.net/
>
>
When i read the mailing list & all flame wars. starting from Theo i think there is no ANY culture at all... Sad but true.
By Thordur Ivar Bjornsson (83.246.89.10) on http://www.secnorth.net
Of course! Developers should spend there unpaid free time hacking on
features they don't want, don't need and dislike just because "users"
want them ?
If you really, really want a feature, contact a developer and pay him
in hard cash to develop said feature. It might not make it into CVS but
hey, you'll have it.
And are you really saying that your donations are wasted because there
are never any new features, never any new drivers, never any new arch's
supported, never any new performance increases, never any new anything ?
And do you really think that developers just sit down, one day and write
SMP support, for what, 3 different architectures and be done with it in
half an hour ? There are hundreds of unpaid man hours that go into this...
comon...
Comments
By anonymous pedro (201.53.181.220) on
>
> Of course! Developers should spend there unpaid free time hacking on
> features they don't want, don't need and dislike just because "users"
> want them ?
Or because the fuehrer said so!
Now seriously, my guess is that in the above sentence, "to listen to" does not mean "to be dictated by", but instead "take into account the concerns expressed by".
Comments
By Anonymous Coward (85.19.140.12) on
> >
> > Of course! Developers should spend there unpaid free time hacking on
> > features they don't want, don't need and dislike just because "users"
> > want them ?
>
> Or because the fuehrer said so!
>
> Now seriously, my guess is that in the above sentence, "to listen to" does not mean "to be dictated by", but instead "take into account the concerns expressed by".
I think they do. The fact that they ask for your dmesg not only in the first mail you get (on your newly installed system), but also request it here and in the mailing lists shows that they /do/ listen to their users. They want OpenBSD to work on *your* hardware, not just their own.
As for the software, I guess given enough developers they'd pay more attention to the needs of their users. However, having fancy software is no good if OpenBSD doesn't work with your hardware.
By Nicram (62.87.244.55) nicram@nicram.sytes.net on http://nicram.sytes.net/
>
> Of course! Developers should spend there unpaid free time hacking on
> features they don't want, don't need and dislike just because "users"
> want them ?
I say abut listening, not hacking. Please read again & then reply.
I'm wondering, why users should send hardware then. Just because devs want it?
If You take something from users. Then give back something too. & shut up. This is how world works.
Comments
By Ruan (82.69.179.14) shearwater+undeadly@gmail.com on
> >
> > Of course! Developers should spend there unpaid free time hacking on
> > features they don't want, don't need and dislike just because "users"
> > want them ?
>
>
> I say abut listening, not hacking. Please read again & then reply.
>
> I'm wondering, why users should send hardware then. Just because devs want it?
>
> If You take something from users. Then give back something too. & shut up. This is how world works.
I think you might have this backwards. Dev requests hardware to work on something. You provide the hardware, and in return the feature gets completed. That's what they give back to the donors. Don't want the feature? Don't donate anything. There's no 'forcing' or 'taking'. They don't want to write a feature? Then they aren't going to.
You might also consider that OpenBSD is made by the devlopers for their own benefits. This makes them the end-users, not anyone who chooses to download any of the fruits of their work.
By Anonymous Coward (216.68.198.1) on
So what if the title was not perfect? This isn't big dollar cryptography here.
We take what we can get from busy developers.
Why the criticisms are ok from a perfectionist, give me all you got, hard driving attitude; which has helped drive OpenBSD to where it is; you can't always expect that here on undeadly.
Would be nice if OpenBSD was better than linux in high computing. Might have to dump the whole security approach then, whoops, sure isn't the aim of OpenBSD, although some might be spoiled into expecting that. Sure not a realtime OS either, but what do you expect? There is no silver spoon.
Peace all.
By Anonymous Coward (89.77.162.243) on
Anyway to change track a bit, anyone know how DRI is progressing? Or seeing as there seem to be a lot of commits to our audio infrastructure recently, well when I look, anyone working on porting jack/ardour? I would love to have OpenBSD's stability for pro-audio recording.
Comments
By Anonymous Coward (85.158.44.148) on
look at ports@ for a jack port for testing.
Comments
By sloppy updater (71.237.209.121) jakemsr@sdf.lonestar.org on
>
> look at ports@ for a jack port for testing.
ardour is in the works. needs some C99 math not yet in OpenBSD: INFINITY, fmin(), fmax() ... maybe more. anyone got a pointer to the FreeBSD code for fmin() and fmax()? I looked in src/lib/libm on the FreeBSD cvsweb, but found no code.
Comments
By Marc Espie (213.41.185.88) espie@ on
> >
> > look at ports@ for a jack port for testing.
>
> ardour is in the works. needs some C99 math not yet in OpenBSD: INFINITY, fmin(), fmax() ... maybe more. anyone got a pointer to the FreeBSD code for fmin() and fmax()? I looked in src/lib/libm on the FreeBSD cvsweb, but found no code.
>
look in msun.
Not that simple to port, since they have some endian-dependent macros to extract mantissa and exponents that we don't have.
Try me and otto@ if you make any real progress on it, probably not for 4.3 though...
Comments
By sloppy updater (71.237.209.121) jakemsr@sdf.lonestar.org on
> > >
> > > look at ports@ for a jack port for testing.
> >
> > ardour is in the works. needs some C99 math not yet in OpenBSD: INFINITY, fmin(), fmax() ... maybe more. anyone got a pointer to the FreeBSD code for fmin() and fmax()? I looked in src/lib/libm on the FreeBSD cvsweb, but found no code.
> >
>
> look in msun.
>
> Not that simple to port, since they have some endian-dependent macros to extract mantissa and exponents that we don't have.
>
> Try me and otto@ if you make any real progress on it, probably not for 4.3 though...
thanks. no hurry here. jack is going to wait until after release for sure.
By Jacob Meuser (71.237.209.121) jakemsr@sdf.lonestar.org on
> Or seeing as there seem to be a lot of commits to our audio infrastructure recently, well when I look, anyone working on porting jack/ardour? I would love to have OpenBSD's stability for pro-audio recording.
see here: http://jakemsr.trancell.org/