OpenBSD Journal

OpenBSD server compatibility list

Contributed by Srebrenko Sehic on from the help-them-to-help-you-to-help-us dept.

Here at Armorlogic, we regularly test new servers from major hardware vendors. Since our product Profense is based on OpenBSD, our testing results also show OpenBSD level of support.

OSCL (OpenBSD server compatibility list) currently shows the status of 13 servers from Dell, IBM, HP, Sun and Fujitsu-Siemens.

We stress-test the servers with regards to CPU(s), network and disk using tools like iogen, stress, dd, netperf, etc. The list will be updated as we test new servers or re-test old ones with new OpenBSD releases. We hope you find the information useful.

If you are running OpenBSD on a server not currently listed, please send a dmesg together with the relevant information about the server to ssehic _at_ armorlogic.com and we will include it in the list.

(Comments are closed)

Comments

  1. By m0rf (68.104.17.51) on

    is this stuff fed back to the project?

    Comments

    1. By Brad (216.138.195.228) brad at comstyle dot com on

      They're providing a list of compatible hardware with specific servers, what is there to feedback?

    2. By Srebrenko Sehic (130.226.138.37) on

      We send (like everybody should) our dmesgs to dmesg@. Also, we add support for some new devices we find or simply submit pcidevs info to the project. Eg. we helped with some various SmartArray controllers and helped fix some bugs in the new ciss driver.

      Comments

      1. By Kevin (24.148.72.216) on http://nycbug.org/index.php?NAV=dmesgd

        Srebrenko Sehic wrote: We send (like everybody should) our dmesgs to dmesg@.

        Sending your dmesg to dmesg@ is all well and good, except that this information is kept secret, not visible to the public, per Theo.

        If you want your dmesg to be immediately useful to other users and not just a few developers, if you want to make your dmesgs public, submit them to NYCBUG's dmesgd.

  2. By Anonymous Coward (82.182.25.187) on

    This is just what i have been looking for. From time to time (long time ago though) we buy HP servers for our customers. Always the same googing around for hours to check RAID-cars, chipsets etc. Tiring. As soon as we get the chance we will contribute to this site. Great work!

    Comments

    1. By Marco Peereboom (67.64.89.177) marco@peereboom.us on http://www.peereboom.us/

      Just about every Dell server works BTW. I tested them myself. I am not aware of anything that doesn't work with -current.

      Comments

      1. By Alan Watson (132.248.81.29) alan@alan-watson.org on http://www.alan-watson.org/

        Hmm. I spent a couple of days earlier this week failing to get OpenBSD/i386 and OpenBSD/amd64 3.8-release to recognize hyperthreading on a Dell SC1425. Fedora Core 4 sees them. Is there any special trick I need to convince OpenBSD to see them? The SC1425 has a pair of 3.0 GHz Xeons with 2 MB L2 cache, if that makes any difference.

        I don't want to get into an argument about whether HTT is a good idea in general. I have a science code that runs 1.6 times faster with HTT on Linux. I was hoping to get the same improvement from OpenBSD.

        Comments

        1. By Marco Peereboom (67.64.89.177) marco@peereboom.us on http://www.peereboom.us

          HT is hype and a waste of time. The Dell systems don't advertise it in the BIOS so OpenBSD doesn't use them. Don't worry, it won't buy you anything anyway.

          Comments

          1. By Alan Watson (132.248.81.29) alan@alan-watson.org on http://www.alan-watson.org/

            As I said, I have a science code that works very nicely with HT. I've just repeated this test on a dual Xeon box running Linux. With two threads, I can run 1 unit of work in 6.5 seconds. With four threads, I can run 1 unit of work in 4.2 seconds. This an improvement of 1.5.

            Unless I can get OpenBSD to recognize HTT on this Dell box, my options seem to be taking a hit of a factor of 1.5 or running Linux. Neither seem especially paletable.

            Comments

            1. By Anonymous Coward (151.188.247.78) on

              You may wish to consider Trustix, then. Among the GNU/Linuxes, this one goes through more scrutiny for security than the other mainstream distributions. Not an OpenBSD-style code audit, for sure, but certainly better than, say, Fedora or SuSE, for example.

              If you're looking for something that reminds you of *BSD (as I do), then you might also consider Slackware. Patrick Volkerding, its chief maintainer, seems to be rather anally retentive about keeping a nice, clean distribution, thus making it pretty easy to maintain. Think of it as GNU/Linux for FreeBSD users.

              Both of these options support HyperThreading and multiprocessor boxes very well. I use both here on all of my SMP boxes, and OpenBSD on most of the uniprocessor boxes. Yes, I am a fan of OpenBSD, but my primary consideration, especially at work where my paycheck is provided, is to Get The Job Done (TM), and to do so with Free Software in spite of the Windoze preference here. If you're getting that kind of performance boost from running your application under HyperThreading environments, then by all means, do what you need to do.

              Comments

              1. By Alan Watson (132.248.81.29) alan@alan-watson.org on http://www.alan-watson.org/

                Thanks for the advice on Linux distributions. Before I give up on OpenBSD though, I'm going to download -current, put on a hard hat, and see if I can't manage to persuade the kernel to Do The Right Thing.

        2. By tedu (71.139.175.127) on

          you either need to patch your bios to make the virtual cpus show up, or start plopping code under arch/i386/i386 to find them.

  3. By Anonymous Coward (69.17.114.101) on

    How about the Apple XServe? I have heard it takes effort to get working. Once Apple comes out with intel based XServes, the used market for G4 and G5 Xserves may be more reasonably priced, so it'd be cool to get these well supported.

    Comments

    1. By Anonymous Coward (151.188.247.78) on

      I've always liked Apple's hardware. It's a shame that it's so pricey, though. But of course, Apple hardware always has been both pricey and closed. Not that certain x86 hardware isn't, too (*cough* BROADCOM *cough* NVIDIA *cough*), but in the x86 world, we have more options available to us at a more reasonable price point.

      I would think that a nice Opteron dual core box would be, perhaps, a better business case. I'm actually a little surprised that Apple didn't choose the Opteron, given its better performance.

Latest Articles

Credits

Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original deadly.org with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]