Child pages
  • 2011.Q4 - Foreverware
Skip to end of metadata
Go to start of metadata

Our stable release is to be based on oi_151a, with backported security fixes. We are aiming to release it before the end of the year.

  • No labels

32 Comments

  1. Anonymous

    Just wanted to send a big heads-up to the developers for their hard work. I'm looking forward to implement the long-awaited stable release of OpenIndiana!
    What is the current guestimate for the release date? Is the goal of "before the end of the year" still valid?

    Best
    ...

  2. Anonymous

    So just to be clear "2011.Q4 - Foreverware" will basically be "oi_151b" a consolidation of fixes over the last few months like:

    https://www.illumos.org/issues/1911 https://www.illumos.org/issues/1673 https://www.illumos.org/issues/1631 https://www.illumos.org/issues/1627 https://www.illumos.org/issues/1333

    Without some of the more recent updates for "resolved" issues, a "Stable" release will not be a solid choice.

    Are you guys still on schedule for a EOY release in 2011?

  3. Anonymous

    For the stable release, is there any possibility that the ISO file for the desktop edition will be packaged to fit on a 700MB CD?

  4. Anonymous

    Really sure does look like OpenIndiana has been abandoned. I've dropped questions on FB and they dont respond to anything from me or anyone else. I'm seeing nothing here is getting answered either. I found an archived msg through google from one of the top ppl on the project on IRC saying looks like the project is dead not long after the last version. Why they wont just announce they dropped it is beyond me. They sure arent answering any questions. AWSOME one more flipping investment of time wasted.

    1. Anonymous

      The mailing list is still active but the wiki is a bit behind, and unfortunately it does give the impression that the project is abandoned.

    2. Anonymous

      As to "wasted" just install oi_151a, and then

      #1 compile the lates illumos-gate via

      https://www.illumos.org/projects/illumos-gate/wiki/How_To_Build_illumos

      #2 make a new BE

      ./usr/src/tools/scripts/onu \
      t nightly`date +%Y-%m-%d` \
      -d $PWD/packages/i386/nightly

      #3 boot into it (via grub) just to test

      init 0

      hard power cycle

      #4 If happy make it your preferred BE (boot environment)

      beadm activate nightly-`date +%Y-%m-%d`

      init 6

      And you are running the latest and greatest illumos-gate build in your openindiana distro (oh you can fallback if you don't like it). To me that's activity (100 resolved issues since 10/19/2011). The developers of illumos are indeed hard work and openindina (whcih uses illumos) benefits greatly.

      1. Anonymous

        Sorry about the formatting in step #2 It should be

        /usr/src/tools/scripts/onu t nightly`date +%Y-%m-%d` -d $PWD/packages/i386/nightly

        1. Anonymous

          /usr/src/tools/scripts/onu -t nightly-`date +%Y-%m-%d` -d $PWD/packages/i386/nightly

  5. Anonymous

    Indeed, even a small note on the status of the project would be very much appreciated!

    1. Anonymous

      The devs are working on the stable version of distro. The first prerelease came out on the 16th of January.

      See: http://wiki.openindiana.org/oi/oi_151a_prestable0+Release+Notes

  6. Anonymous

    I love ZFS.

    I love zones.

    I love the PKG(5) infrastructure.

    I hate dealing with Oracle's restrictions.

    And so, I love OpenIndiana!

    I think the above changes to the original Solaris infrastructure make OI compare favorably with most flavors of linux, and unix. And while Windows makes a decent user support infrastructure, it doesn't make a reliable service infrstructure despite the positive changes made for Server 2008 'Core'.

    So, if you want to grow I am with you...

    quote:
    "Get it to the users. The users will bring money to pay for it."

  7. OpenIndiana can only be summed up as dead?  The stable release was supposed to happen in Q4 2011 and it has been nearly 9 months since the last development release!

    It was a sad, sad day when Sun Microsystems was sold to Oracle.  Only Jonathon Schwartz can be blamed for his bungling, such as paying billions for MySQL and his on again/off again approach to Solaris on x86.  It was a mistake having him lead the company.

    Be that as it may, we are now left with the pieces, and there isn't any viable alternative to OpenSolaris.  The OpenIndiana fork is for all intents and purposes a dead project, and the others that formed around OpenIdiana (Schillix, etc) are using Linux userlands or are not relevant.

    A month ago I chatted online with an Oracle sales rep and asked if I purchased the DVD if I had license to run my commercial websites in a production and the answer was yes.  Wanting verification, I chatted with another sales rep who didn't know and gave me a phone number.  That sales agent couldn't help me and gave me another number.  I was transferred to 4 different sales reps who couldn't answer my question.  The last sales rep gave me an email address to a sales team he said could answer.  They responded and said it wasn't their realm and gave me a phone number - of the guy who gave me their email address!

    So nobody at Oracle can tell me if I purchase their $75 DVD set If I am entitled to run my commercial websites on Solaris 10.  Even though I mentioned the note on the purchase page says you don't have a license to run in production, nobody at Oracle can validate it.

    1. Actually, no, as an observer and casual contributor, I wouldn't say the projects are dead - neither illumos (the OS/Net kernel part which is shared by many Distributions – which in turn combine OS/Net, a packaging engine, and lots of non-kernel Solaris-derived and third-party software), nor OpenIndiana which is the distribution most resembling the classical general-purpose Solaris in its OpenSolaris (Indiana) IPS-packaged variant.

      Apparently, Sun/Oracle did not open up many pieces of the puzzle and those had (and some still have) to be replaced with new code, or with ports of BSD, Linux or GNU codes where appropriate. Also the projects have met some growth issues and are being reorganized, in both development processes, coordination, decision-making and governance, and documentation, in particular to become easier to join and contribute to – without disrupting works of existing contributors and developers.

      So, while the outside shell may seem dead, from activity on the mailing lists and on the Wikis I can see that there is a lot of work under the hood – and it is in the open, anyone can join. We are making it easier to join and start contributing, too. I see that people are coming.

      The released LiveCD is indeed somewhat old; however, there are many improvements (and bug fixes) in the source code repository and package repo added since its release. Some snapshots are published as new "Developer" LiveCDs (i.e. oi_151a3), and about once a month new packages are published in the "/dev" package repository (currently at oi_151a4 level). This means that you can install the existing LiveCD and update it to current right away.

      There is some procedural work, as well as some bugfixes and "cleanup", and of course some testing, to get out of the way before publishing and promoting a build as a truly "Stable" one. AFAIK the "/stable" branch on the package repository is still only planned, as part of such new release.

      However, for most mailing-list posters' the current dev releases are as good and stable as it gets, and while there are sometimes bugs found and squashed – a "stable" release would likely have about as many of them, and a review and request-to-integrate process does delay or block unworthy additions, and causes improvements to be of a higher class. So IMHO this "Stable" would be a label for managers' piece of mind, but not necessarily adding much security or stability in comparison to "Dev" label (though a "ban" on disruptive young changesets could indeed help that).

      I am not sure that the community organization at the moment truly has the resources to honestly support several branches of the OS – like a stable one (backporting critical updates) and a dev one (with all the new features and bugs), but I think it may come to that sooner or later. The more hands on deck – the faster we'll all get somewhere (smile)

      If required, you can use the documented Distribution Constructor to create a new installable LiveCD based on your modified code (which of course includes the possibility of making one based on unchanged current code and packages), and roll your own oi_151a4+ installable media, if needed (i.e. embedding drivers to support your hardware). But I don't know if a download is going to be "officially" provided of any new builds, nor what is the timeframe until a promoted "stable" (or next interim "dev") release.

      Hopefully this helps against some FUD,
      //Jim Klimov (just another external contributor-observer)

      PS: If you want to see first-hand and even influence how the project goes – just come and join the ranks in discussions and development! (wink)

      1. Thanks for your considerate and educational reply.

        Presently I host my domains on OpenBSD, but I've been a Solaris admin for 9 years (17 years on AIX) and like Crossbow network virtualizaton and Containers. I've considered moving to FreeBSD using their new VIMAGE Jails but have ran into a couple of problems, so am also considering using OpenIndiana to use Zones and Crossbow. If I did move to OpenIndiana I would also have to try and port OpenSMTPD which is my current mail server.

        How does OpenIndiana respond to and issue patches for security-related problems?

        I believe there are 200 or so developers for FreeBSD and 20-40 for OpenBSD, do you know how many developers currently work on OpenIndiana?

        1. I think that the discussion should better move to the oi-discuss mailing list - so it will be exposed to more active people. Not many track the comments actively (wink)

          As for security-related updates, I have seen some discussed and integrated on the list. I can't say that this flow is intensive, but there is not so much critical third-party software in the mix, and developers are now more into sorting out the core software's layout and procedures as to become even more open and quick to accept the contributions (including integrations of security fixes).

          It is also hard to guess about the amount of developers, observers and contributors – some are individuals like myself, others work for companies like Nexenta, Joyent, EveryCity and several others, and I am not sure that everybody is exposed during discussions and integrations. I'd guess at least several tens of people with original Solaris background (previously employed in Sun OS-development) are active from around the globe.

          //Jim

  8. I've had OpenIndiana running on a laptop before with the internal disk, however, I just installed OpenIndiana server release to a Seagate GoFlex USB 1T external drive and it doesn't boot.  The installation succeeded and the post-install reboot everything was fine, it came up on the USB drive, but subsequent reboots now is nothing but a continuous loop of "Loading Grub," getting a menu, then it reboots and continues the cycle.

    In the boot menu -as I did with OpenSolaris - I add acpi-user-options=0x2 (this just came to me, I forgot about it).

  9. I spent considerable time installing the Sun Studio compiler which goes to /opt.  Then I installed automake, autoconf, libtool, and m4 which went into /usr/local/bin.  Then I installed db4.  Next I was going to compile OpenSMTPD because all of my prerequisites were now met.

    However!  It said no CC compiler was found, and also when I set my PATH to include /usr/local/bin, to my surprise there is NO MORE /usr/local/bin, and /opt/sunstudio12.1 and the link to /opt/SUNWSpro/sunstudio12.1 is gone which had the compiler.

    Then ONLY thing I did was reboot.  Anybody want to offer an explanation for the disappearance of entire directories? 

    1. Well, it sounds like you either managed to wipe or roll back your installation, or somehow booted a previous boot environment whose rootfs did not yet contain the additions to /opt and /usr/local. Alternately, you might have run a system upgrade (which clones your current rootfs), did not reboot into the new BE and installed the new software – then it remains installed in the old BE and unknown to the new BE after they've split ways post-cloning. What does beadm list say – do you have any newer environments than the one flagged by N (now running), and are discrepancies large (allocated Space unique to that BE compared to the currently active one)?

      For example:

      # beadm list
      BE                         Active Mountpoint Space Policy Created
      oi_151a                    -      -          7.40M static 2012-03-25 02:57
      oi_151a-1                  -      -          7.91M static 2012-03-27 03:31
      oi_151a-2                  -      -          452K  static 2012-03-30 04:15
      oi_151a3-20120503          -      -          12.4M static 2012-05-03 01:17
      oi_151a3-20120503-backup-1 -      -          148K  static 2012-05-16 01:49
      oi_151a4-20120607          NR     /          6.24G static 2012-06-08 00:21
      oi_151a_release            -      -          54.0K static 2012-03-27 03:16

      Here, I am Now running (and will run after the Reboot) the oi_151a4-20120607 boot environment.

      And, again, you're better off asking questions by subscribing to the oi-dev mailing list. There are more eyes and ears than those monitoring the wiki comments (wink)

      HTH,
      //Jim Klimov

      1. It may be a hardware problem.  I tried reinstalling the drive and it just sat at 5% for about 20 minutes, so I grabbed another drive.  Also, before reading your suggestion of the boot environment, I also thought of that and had upgraded, so it may have been on an alternate BE. 

        Anyway, I am installing to a second drive.  Before that happened, the place where I was in the OpenSMTPD installation was missing the Berkeley DB headers, so I have the db-4 package, which I hope suffices.  I also have the db-5 package downloaded from Oracle, but db-4 is in pkg format, so I hope that one works. 

        I will subscribe to the oi-dev mailing list. I haven't used OpenSolaris/Indiana since Oracle dropped it, but have always like the innovation.  At work I do AIX (17 years) and (ahem) RHEL (crying).  There are just a few Solaris servers remaining that are Solaris 8.  Hopefully I get OpenSMTPD working on OI, and may move my commercial websites from OpenBSD (which is also used for my firewall).  FreeBSD is my everyday laptop.

        Thanks for the help.

        1. > Also, before reading your suggestion of the boot environment, I also thought of that and had upgraded, so it may have been on an alternate BE.

          If that is the case, you can reboot back into the un-updated BE where you've done considerable work (beadm activate olderBEname; init 6), and update the result again – into a different branch of ZFS, though.

          You might also want to just mount a different BE (i.e. to copy over some config files or unpackaged programs) – then you can use beadm mount oldBEname /a and the rootfs hierarchy should appear under /a. When done, unmount it with beadm umount oldBEname.

          HTH,
          //Jim Klimov

  10. I sent this to oi-dev:

    I spent considerable time installing Sun Studio and other packages to port some software and when I rebooted I lost all of /opt/sunstudio12.1 and all of /usr/local/bin.

    No problem, I rebuilt and have again spent considerable time rebuilding everything.  Rebooted a number of times without a problem.  Then last night I powered off and today I powered on and I have a console login, but it is unresponsive to any keys.

    This is the most unstable OS I know of.  I used OpenSolaris as my primary OS until Oracle killed it, and switched to OpenBSD.  I was looking at returning by using OI, but am uncertain now.

    Too many problems.  It is inconsistent.  Keyboard doesn't work?  And I know it isn't a H/W problem, because if I reboot to OpenBSD it works perfectly fine.  Same if I boot to FreeBSD.   

    1. Such hardware glitches are not typical for supported hardware; what do you have? Unfortunately, it takes some time to create or port drivers for bleeding-edge chips, because the currently-limited human resources are best spent on OS features (and the HW resources might not include what you have). So, at least, please go to Community HCL page and mark your config as problematic, so it might gain attention.

      Also regarding kbd/mouse over USB, there were reports that power-management might put them to sleep and not wake them up. There were some config tweaks in the mailing lists to disable PM for these components; FWIW:

      (from a recent mail by Udo Grabowski): ...I can circumvent this problem by declaring a always-on in /etc/power.conf, but this should be the default for a USB mouse (and keyboard) devices:

      device-thresholds /pci@0,0/pci1043,82d4@1d/mouse@1 always-on
      device-dependency-property removable-media /dev/fb
      autopm                  enable
      autoS3                  default
      cpu-threshold           1s
      S3-support              default
      # Auto-Shutdown         Idle(min)       Start/Finish(hh:mm)     Behavior
      autoshutdown            30              9:00 9:00               noshutdown
      cpupm  enable event-mode
      cpu_deep_idle enable

      -- I guess your device paths might differ; I didn't test this myself (Jim)

      Apparently, many users, businesses and cloud-hosters now use illumos-based distros, so instability and unpredictability is not typical of them...

      HTH,
      //Jim Klimov

      1. My hardware is not bleeding edge.  It is an old Dell Latitude D510 and am only using the integrated keyboard, and since it is the server edition, I am not using a mouse. 

        The drive I installed to was a 1TB Seagate GoFlex USB drive because the internal drive of the laptop has OpenBSD installed, and the keyboard works when booted.  I have FreeBSD installed on another SeaGate GoFlex USB drive and that one boots and the keyboard works.

        There was an email thread I read yesterday from April 2012 from someone involved in OI (can't remember the name), so that encouraged me that development is still taking place and it isn't stalled or dead.  So I was enthused to try again and then ran into the keyboard issue.

        I have considered installing onto the internal drive (which of course would wipe out my OpenBSD install - but it is test anyway) and see if that presents more stability than what I've been having.  But I have no stability issues with USB drives with FreeBSD (fwiw).

        1. Installed to the internal drive on Tuesday and powered off Tue night.  Wednesday it powered up and was usable, then powered down for the night.  Here we are on Thursday and sitting at a login prompt but no response from the keyboard. 

          The original BE works.  It must be some packages that I'm installing.  Meant to install other packages into a Zone but forgot and put them into the global.  Will start over with the original BE.

          1. Unfortunately, I am not sure what may be wrong with the keyboard (driver? power-mgmt? some USB hub enroute? does re-plugging it help after hangs?), as well as your subscription to the mailing lists - I haven't seen a post from you there, I think, so I reposted an overview of your HW problems and a link to this comment thread.

            Why would a server need a KBD anyway? (wink) Doesn't it have SSH for times when it works, and IPMI/sercon for when it doesn't?
            HTH,
            //Jim

  11. Is VirtualBox in a repository for installation?  I have openindiana.org/dev, sfe, and sfe-encumbered set and have tried to install and search for VirtualBox but it doesn't appear to exist as a package.

    I have tried to install it, and run into "cannot find libfreetype|libXrender|libXi" etc.  That is using a binary downloaded from virtualbox.org and trying with any version all the way back to the 3x branch.  When VB was installed on OpenSolaris, I did a pkgadd and that was it.  What is different with VB installing on OI?

    1. I do not think VirtualBox will ever appear in the OpenIndiana-related IPS repositories, because its PUEL licensing dictates that each user himself must download and freely use the software from the vendor.

      UPDATE2: Hmm, I may be wrong. The base VirtualBox 4.x is GPL'ed, and the Sun/Oracle optimized and extended features are PUELed  (see https://www.virtualbox.org/wiki/VirtualBox_PUEL, https://www.virtualbox.org/manual/ch01.html#intro-installing) – so it seems that the core engine may be redistributable and used however you want (probably from SFE repo?).

      As for library dependencies, I have an OI 148a-based box (updated to oi_151a and privately-built kernels since then), which has a working VirtualBox installation. I think the basic GUI LiveCD installation sifficed to have the libraries in place; you can search the IPS repos for what you need from command-line, using wildcards for file names:

      # pkg rebuild-index
      # pkg search -l libfreetype'*.so'
      INDEX      ACTION VALUE                          PACKAGE
      basename   link   usr/lib/libfreetype.so         pkg:/system/library/freetype-2@2.4.9-0.151.1.3
      ...
      
      # pkg search -l 'libXi*.so'
      INDEX       ACTION VALUE                                           PACKAGE
      basename    link   usr/X11/lib/amd64/libXi.so              pkg:/x11/library/libxi@1.3.2-0.151.1.3
      ...
      basename    link   usr/X11/lib/amd64/libXinerama.so        pkg:/x11/library/libxinerama@1.1-0.151.1.3
      ...
      pkg.fmri    set    openindiana.org/x11/library/libxi       pkg:/x11/library/libxi@1.3.2-0.151.1.3
      pkg.fmri    set    openindiana.org/x11/library/libxinerama pkg:/x11/library/libxinerama@1.1-0.151.1.3
      
      # pkg search -l 'libXrender*.so'
      INDEX      ACTION VALUE                           PACKAGE
      basename   link   usr/X11/lib/amd64/libXrender.so pkg:/x11/library/libxrender@0.9.6-0.151.1.3
      ...

      In the example above I used the -l option to search in the locally-installed packages (the libs my VBox apparently uses). For example, there are some copies provided with Java as well. The search command also suggested that I rebuild my indexes, so I added that above.

      Apparently, for this example you'd have to install the three packages above (recursing for dependencies), then reiterate until your VBox installs and runs successfully – and doesn't ask for more libs (wink) (you can use ldd to check VBox's bins and libs for dynamic dependencies). Install the pkgs found above:

      # pkg install freetype-2 libxi libxrender

      When you have the list of needed packages, I guess a new Wiki page on VirtualBox installation on OI is in order (wink)

      UPDATE1: The renamed IPS packages also define the legacy SUNW* package names that they "replace". You can just inspect the SVR4 dependencies of the SUNWvbox package (the depends file in its manifest), or remember its dependency complaints during installation, and use those legacy package names for searches or IPS installation, i.e.:

      # pkg search SUNWfreetype2 | grep legacy
      legacy_pkg  legacy SUNWfreetype2                       pkg:/system/library/freetype-2@2.4.9-0.151.1.3
      
      ### You can "pkg search '*freetype*' | grep legacy" if you can't guess old pkg name

      Hopefully, since these are IPS packages now under the hood, asking for their installation should pull up all their needed dependency packages too.

      Good luck,
      //Jim Klimov

      1. Thanks for the info.  You are very helpful.

        Those were the only libraries I could think of at the time (one other is fontconfig).  I installed from a USB image, but will do a search from the repository and install all of the missing libraries from there.

        I need VirtualBox because I can't get OpenSMTPD to port to OI and the developer responsible for the portable version won't reply, so I will run OpenBSD on VB.  Otherwise I have successfully built Pound and nginx, so I am set once I get VB installed.

        1. USB images should be functionally identical to ISOs – and there are also several "editions" (Live with GUI, Text without GUI and its libs, AI for networked installations).

          If you plan to use VirtualBox to run a server VM (i.e. with the SMTP relay), take a look at my SMF package to maintain VBox VMs as SMF service instances – this might be convenient for automated systems management, definite dependencies, etc. The project is on http://vboxsvc.sourceforge.net and its discussion is in VBox forum, thread "[Free as in beer] SMF service for VirtualBox VM's": http://forums.virtualbox.org/viewtopic.php?f=11&t=33249. I am not sure that much more on this subject belongs in this comment thread on the OI Wiki (wink)

          HTH,
          //Jim Klimov

  12. This problem is really confusing me.  I am setting up, according to a Sun document on eliminating web page hijacking, and am unable to  get one filesystem to mount in a zone.

    >add fs

    > set dir=/var/apache2

    > set special=/shared/data

    > add options [ro,nodevices,nosuid,noexec]

    > set type=lofs

    > end

    >add fs

    > set dir=/var/apache2/logs

    > set special=/shared/logs

    > add options [rw,nodevices,nosuid,noexec]

    > set type=lofs

    > end

    >add fs

    > set dir=/var/apache2/run

    > set special=/shared/run

    > add options [rw,nodevices,nosuid,noexec]

    > set type=lofs

    > end

    When the zone boots it fails with these messages:

    zone 'web': Could not mkdir /zones/web/root/var/apache2/logs.

    zone 'web': It is on a read-only file system in this local zone.

    zone 'web': Make sure /var/apache2/logs exists in the global zone.

    zone 'web': failed to create mount point

    zoneadm: zone 'web': call to zoneadmd failed

    But /var/apache2/run mounts just fine without an error.  Removing /var/apache2/logs allows it to boot.  I can understand because /var/apache2 is read-only but it mounts run and besides they are lofs mounts.

    1. Just a sanity check: does your global zone "/shared/data" base structure include the "logs" and "run" subdirectories you would then mount over? I think the mountpoints should pre-exist, for lofs too. In fact, I would not expect it to try creating the mountpoints itself in zfs-style; if it does - this must be some relatively recent development.

      1. Good catch.  /shared/data/logs didn't exist but /shared/data/run did.

        One thing to note is "set options=[ ]" has changed to "add options [ ]"