Developer bootstrap failure.



  • Doing this in qemu, so be warned that all may not be as it seems.  I'm doing this with the 9-12-2006 dev iso in sullrich's home directory.  Looks like a c-coding error, and that's just out of my league I'm afraid.  :(  Since I can't copy and paste, here's the screenshot:

    ![Q Screenshot 1.png](/public/imported_attachments/1/Q Screenshot 1.png)
    ![Q Screenshot 1.png_thumb](/public/imported_attachments/1/Q Screenshot 1.png_thumb)
    ![Q Screenshot 1.png_thumb](/public/imported_attachments/1/Q Screenshot 1.png_thumb)





  • Somehow the IPSEC patch has been applied twice.

    Try:

    rm -rf /usr/src/lib/libipsec/*

    And start the developer bootstrap over.



  • Done.  We'll know in about 12 hours or so if that did it. ;)



  • @Numbski:

    Done.  We'll know in about 12 hours or so if that did it. ;)

    In the beginning, didn't understand the "12 hours" statement - but I think I get the picture.
    Haven't enough good bear in my place  ;D



  • Football is on.  I'll keep busy.  For some oddball reason fastest CVSUP is taking a really long time to run this time though. ???

    Anyhoo, I'll be patient.  Qemu won't be super fast, I know that.



  • Huh.  That's not right.  It just sits at the fastest cvsup screen for a very looong time, but never actually runs cvsup.



  • Only pointing this out because of how odd it is, but I've worked around it.  The bootstrap script would get to testing cvsup14.freebsd.org, then hang.  /var/db/fastest_cvsup exists, but is empty.  Did that twice, left it sitting for over an hour.

    Manually ran the fastest_cvsup, put the results into /var/db/fastest_cvsup, and commented it from the dev_bootstrap.sh script.



  • Building building building….

    :o

    Up to stage 4.4: builiding everything

    Yay.  Have no idea how much longer this'll take.  ;D



  • @Numbski:

    Building building building….
    Yay.  Have no idea how much longer this'll take.   ;D

    Actually, it really works  ;) - it took nearly 2 hours on a Dell 5150 Dual Core to give a 'home baked' ISO. I ran the Dev in a VMWARE environment.
    I created a second VMWARE BSD host to test the ISO - it worked just GREAT!
    Please note that this was the dev bootstrap-phase – the next time you build an ISO (or whatever) thing go significantly faster….



  • Don't forget to bootstrap bsd installer as well.  There is a file in builder_tools that does this:

    cd /home/pfsense/tools/builder_scripts
    ./cvsup_bsdinstaller



  • @sullrich:

    Don't forget to bootstrap bsd installer as well.  There is a file in builder_tools that does this:

    cd /home/pfsense/tools/builder_scripts
    ./cvsup_bsdinstaller

    This has already being done ones in the intitial "dev_bootstrap.sh" 'sanity ISO build check'…. just before the real ISO building begins.
    A subsequent manual 'cal' to ./cvsup_bsdinstaller gave me an error, when all the installer packages were 'tarred up' in the final phase:

    ...
    ===>  Building package for bsdinstaller-2.0.2006.0918
    Creating package /home/pfsense/tools/builder_scripts/conf/packages
    Registering depends:.
    Creating gzip'd tar ball in '/home/pfsense/tools/builder_scripts/conf/packages.tgz'
    tar: share/dfuibe_lua/lib/[color][b]b[/b][/color]itwise.lua: Cannot stat: No such file or directory
    ...
    tar: share/dfuibe_lua/lib/uinav.lua: Cannot stat: No such file or directory
    pkg_create: make_dist: tar command failed with code 256
    *** Error code 1
    

    ??? Hey, i.e., a (/usr/local/)share/dfuibe_lua/lib/Bitwise.lua exists.

    Btw: this :
    @dev_bootstrap.sh:

    echo "SKIP_RSYNC=yo" >> $HOME_PFSENSE/tools/builder_scripts/pfsense_local.sh

    poluates the pfsense_local.sh file at the end. Is this 'wanted/needed" for later on building ?

    Please note that I understand that 'no help' is (can be) given on Developers issues… (its so bloody huge  :o).



  • SKIP_RSYNC=yo is required if you do not have a pfsense cvs account (most likely).

    To work around the bsdinstaller issues, do this:

    pkg_add -r http://www.pfsense.com/packages/All/bsdinstaller-2.0.2006.0728.tbz
    pkg_add -r http://www.pfsense.com/packages/All/lua50-5.0.2.tbz

    And yes, the developer bootstrap is not supported officially due to the sheer complexity of everything involved in bringing you pfSense.  It requires countless utilities and processes to make everything work together.



  • Thanks everyone, including GeekGod in irc.  Was trying to help out, not create a mess.  Looks like I bit by the cvs probs here.

    cvs [checkout aborted]: invalid change text in /home/pfsense/cvsroot/pfSense/usr/local/www/system_advanced.php,v



  • @sullrich:

    SKIP_RSYNC=yo is required if you do not have a pfsense cvs account (most likely).

    Figured that out when removing these SKIP_RSYNC=yo lines….

    @sullrich:

    To work around the bsdinstaller issues, do this:
    pkg_add -r http://www.pfsense.com/packages/All/bsdinstaller-2.0.2006.0728.tbz
    pkg_add -r http://www.pfsense.com/packages/All/lua50-5.0.2.tbz

    First line : complains that version 2.0.2006.0918 (more recent) is already installed.
    Package lua50-5.0.2.tbz comes in just fine, but…. this doesn't removes the errors.
    Anyway, this is an 'Installer' issue, not a pfSense pb.
    I just 'rebuild world' and the ISO is just fine

    @sullrich:

    And yes, the developer bootstrap is not supported officially …..

    I understand.
    Over & out.



  • About the only other thing you can do at this point is to copy /scripts/lua_installer to /tmp/

    chmod a+rx /tmp/lua_installer
    vi /tmp/lua_installer

    Find where the installer backend is being launched and remove the 2> redirection.

    Now run /tmp/lua_installer to obtain the installer specific error.



  • @Numbski:

    Anyhoo, I'll be patient.  Qemu won't be super fast, I know that.

    There's a good reason why Q (on OSX it's called Q and not Qemu which makes a difference) is kind of slow. It runs in pure emulation mode because the accelerators kqemu (closed source) or QVM86 (OS) didn't make it to OSX as of today.

    Regards
    Daniel S. Haischt


Log in to reply