Commits on Sep 18, 2008
  1. Creating my own branch.

    cfb committed Sep 18, 2008
Commits on Aug 30, 2008
  1. Nic found a blechy segfault. This fixes it.

    rtilder committed Aug 30, 2008
Commits on Aug 13, 2008
  1. - Fix to compare release numbers numerically.

    rkhardalian committed Aug 13, 2008
Commits on Aug 6, 2008
Commits on Aug 5, 2008
  1. From: Chet Burgess <>

    pdibowitz committed Aug 5, 2008
    Attached is a patch that converts all usage of reporting errors via c_failure
    to using the c->error call instead. This results in a nice stack trace like
    error output that shows the flow of errors.
    The main spine program still uses the presence of c_failure to determine if
    there was some type of parsing failure in creating of the Spine::Data object.
    The constructor for Spine::Data handles setting c_failure if a failure has
    occurred during object creation. 
    This patch also adds significant style cleanup to all files affected by the
    core of the patch.
    This patch closes issues 29 and 23.
    Minor typo fixes by Phil Dibowitz <>
    Signed-off-by: Phil Dibowitz <>
  2. From: Alexander Toth <>

    pdibowitz committed Aug 5, 2008
    A comment in says "But we always at --archive, just to be safe." 
    It should probably be "always add --archive," considering the code context.
    With re-diff by Chet Burgess <>
    Signed-off-by: Phil Dibowitz <>
  3. Nic pointed out the the messaging on this error message could be

    rtilder committed Aug 5, 2008
    significantly more informative about what to do.
  4. - Applied patch for issue #28, submitted by cfburgess.

    rkhardalian committed Aug 5, 2008
Commits on Jul 14, 2008
  1. From: Chet Burgess <>

    pdibowitz committed Jul 14, 2008
    While my patch works great it does break backwards compatibility with with
    existing data trees. I had hoped that it would be possible to adjust existing
    trees to account for this. 
    I have however since discovered that this will be difficult for certain older,
    pre spine, version of the code (specially the 1.6.6 and earlier branches of
    As such I am resubmitting a patch that restores the old functionality to
    c_is_virtual. The c_is_virtual key should be considered deprecated and as soon
    as all older trees are upgrades (of which there are exactly 2) I will submit a
    new patch to remove the c_is_virtual entirely and stick purely with the
    c_virtual_type variable.
    Issue 26 opened to undo horrible hack in the near future.
    Additionally, style cleanups through out the file from Phil Dibowitz
    Signed-off-by: Phil Dibowitz <>
Commits on Jul 9, 2008
  1. From: Chet Burgess <>

    pdibowitz committed Jul 9, 2008
    I have re-written the is_virtual function in the SystemInfo plug-in to properly
    detect VMware, Xen Para and Xen HVM. This necessity breaking backwards compatibility
    to implement cleanly. As TM is the only existing user we have internally agreed to
    deal with changing our tree to function with the new code. 
    Additionally the new is_virtual plug-in no longer automagically includes a role for
    virtual support as we now have multiple types of virtual systems and the inclusion of
    roles is something that should be done in the data tree not in the code.
     * Also, various style cleanups by Phil Dibowitz <>
    Signed-off-by: Phil Dibowitz <>
  2. From: Chet Burgess <>

    pdibowitz committed Jul 9, 2008
    OK so ultimately the Hardware plugin should be able to provide this information.
    That being said I took a look at the current state of the hardware plugin and it is
    ugly. The plugin won't even run complaining about numerous syntax errors and
    undefined variables. It will probably take a good amount of effort to get it working
    at this point.
    So as a temporary hack (since we are in need of a solution to this) I have added
    support to the SystemInfo plugin to use dmidecode to return the "Product Name" string
    in the c_hardware_platform key. If for some reason dmidecode does not return a string
    (like it will on Xen VMs) the key will be set to "UNKNOWN".
    Signed-off-by: Phil Dibowitz <>
Commits on Jul 7, 2008
  1. From: Chester Burgess <>

    pdibowitz committed Jul 7, 2008
    Looks like someone forgot a ; when porting this plugin from the old 1.x
    tree. Additionally the variable name that the plugin sets has been changed
    breaking backwards compatibility with exiting config trees.
    Signed-off-by: Phil Dibowitz <>
Commits on Jul 2, 2008
  1. spine-mgmt/spine

    pdibowitz committed Jul 2, 2008
      - Minor fix to help text - s@/etc/nodefile@/etc/nodename@
      - Fixes Issue 19
    Signed-off-by: Phil Dibowitz <>
Commits on Jun 27, 2008
  1. - Let the user specify profile/actiongroups and actions case insensit…

    pdibowitz committed Jun 27, 2008
      (resolves Issue #8)
    Thanks to Chet for testing.
    Signed-off-by: Phil Dibowitz <>
Commits on Jun 21, 2008
  1. makeing it so that YAML and JSON actually work

    richard committed Jun 21, 2008
Commits on Jun 11, 2008
  1. From: Kieren Hynd <>

    pdibowitz committed Jun 11, 2008
    - Add a --version option
    - Print frozen information when called with --last
    (With some minor style fixes on both this and surrounding code
     by Phil Dibowitz <>)
    Signed-off-by: Phil Dibowitz <>
Commits on Jun 5, 2008
  1. From: Chet Burgess <>

    pdibowitz committed Jun 5, 2008
    The use of stat's rdev on 2.4 kernels is unreliable. Here we only use it on
    non-2.4 kernels. See issue #15.
    An earlier version of this patch was committed and reverted, and this is a
    follow up with additional considerations requested by Ryan Tilder.
    This one looks good and has been tested, so I'm committing.
    Signed-off-by: Chet Burgess <>
    Signed-off-by: Phil Dibowitz <>
Commits on May 26, 2008
  1. I can't believe I'm having to revert a commit to fix a problem with a…

    rtilder committed May 26, 2008
    … 5 year
    old distribution.  The file type tweak is there for a reason: I needed it in
    Come up with logic that will work around the particular distribution
    versions if you must but don't remove the logic completely.
Commits on May 25, 2008
  1. commiting cfburgess/Phils patach from Issue 15.

    richard committed May 25, 2008
Commits on May 22, 2008
  1. Fix issue 18, looks like it got broken during the conversion. The who…

    richard committed May 22, 2008
    …le thing needs to be changed some time
Commits on May 17, 2008
  1. Just formatting changes

    richard committed May 17, 2008
  2. _evaluate_key allowed objects that were not array refs to be returned…

    richard committed May 17, 2008
    … which would cause _get_values to bail out.
    This was the only place that _evaluate_key directly added data to a key (rather then only changing and returning extra data) so I opted to remove this and change _get_values to handle a returns other then array refs.
    I am thinking through large changes the might tidy this up which will appear in my branch soon.
Commits on May 16, 2008
Commits on May 14, 2008
Commits on May 12, 2008
  1. since it's perfectly OK to have an overlay-only tree or directory we …

    richard committed May 12, 2008
    …should not need to skip the concat. Not sure what the foo was for originally
Commits on May 1, 2008
  1. If the overlay_root key is not set, no changes are applied, but we ne…

    pdibowitz committed May 1, 2008
    …ver tell
    the user anything is wrong. This patch adds a warning.
    This patch is based on a patch by Chet Burgess
    <>, but fixed up for spacing, line length, and
    quote type by me.
    Signed-off-by: Phil Dibowitz <>
Commits on Mar 18, 2008
  1. - Fixes issue #12 (restart_deps not executing at all).

    rkhardalian committed Mar 18, 2008
Commits on Mar 12, 2008
  1. From: Todd Jimenez <>

    pdibowitz committed Mar 12, 2008
    Spine::Plugin::Auth does not sort ssh authorized keys when it emits them, thus
    diffs can be much larger than they need to be.
    This patch sorts them prior to emiting.
    Signed-off-by: Phil Dibowitz <>
Commits on Mar 9, 2008
  1. From: Chet Burgess <>

    pdibowitz committed Mar 9, 2008
    This patch fixes several bugs with the FirstRun plugin:
      - It didn't register any hooks
      - It returned an invalid constant
    It also fixes the default conf file's use of the plugin.
    Signed-off-by: Phil Dibowitz <>
Commits on Mar 8, 2008
  1. Chet pointed out the help is wrong here, so I'm fixing it.

    pdibowitz committed Mar 8, 2008
    - Phil
Commits on Feb 14, 2008
  1. Extending "internal_dirs" to be exensible and configurable.

    pdibowitz committed Feb 14, 2008
    Simply add a 'spine_local_internals_dirs' key in /spine_internals/config/
    with a list of directories to process *as-if* they were part of spine_internals.
    This list should *not* include the "config" part, but the path *should* have a "config" directory under it (though spine will ignore it if it doesn't).
    In other words, if you want to include:
    then just do
    inside of spine_local_internals_dirs and spine will do the rest for you. Remember, like the rest of spine-internals you will NOT have the rest of your keys defined, this is special early boot-strapping stuff only. Abusing this will cause the bullet to go squarely into your foot.
    - Phil
Commits on Jan 18, 2008
  1. trunk/lib/Spine/Plugins/

    rtilder committed Jan 18, 2008
        - Damn.  Missed a few more tidbits
  2. trunk/lib/Spine/Plugin/

    rtilder committed Jan 18, 2008
        - Merge in minor code change from 1.6.8 that was missed
Commits on Jan 15, 2008
  1. trunk/lib/Spine/Plugin/

    rtilder committed Jan 15, 2008
        - Apply Jeff Schroeder's patch to correct my oversight.  See

Commits on Jan 10, 2008
  1. trunk/lib/Spine/

    rtilder committed Jan 10, 2008
        - Revert changes from yesterday since they didn't work anyway