Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Commits on Mar 22, 2011
  1. Subject: Properly guard auth's "system homedirectories"

    pdibowitz authored
    From: Phil Dibowitz <>
    Trac: Refs #131
    * If a directory is in auth_system_homedirs, auth shouldn't touch it in
      any way
    * auth needs to respect a 0 in min_root_keys
    Signed-off-by: Phil Dibowitz <>
Commits on Mar 16, 2011
  1. Subject: Remove .svn override for overlays

    pdibowitz authored
    From: Phil Dibowitz <>
    Trac: Refs #128
    Doh, PhilFail. The existing exclude functionality does work here,
    I had a PEBKAC. Thanks to Chet for pushing me on this.
    This removes that one small part from the last commit.
    Signed-off-by: Phil Dibowitz <>
  2. Subject: Fix bad syslog levels, fix bad quoting, consistently ignore …

    pdibowitz authored
    …svn internal files
    From: Phil Dibowitz <>
    Trac: Refs #127, #128
    - Various places are trying to log with 'warn' which isn't valid
      and causes spine to crash instead of log the warning
    - While I was in various files fixing logging, I fixed " -> ' in a few
    - We ignore .svn for config keys, but not overlays. My ignoring voer
      overlays is not as agressive on overlays as config keys because
      that would break stuff, I just want to be able to do --croot
      with my checkout directory without dropping .svn dirs everywhere.
    From: Phil Dibowitz <>
Commits on Nov 18, 2009
  1. Subject: Fix new Exec code to behave during dryrun.

    cfb authored
    From: Chet Burgess <>
    Trac: Refs #107
    Same as changes to spine_2_1 branch, but now on trunk.
    The introduction of simple_exec caused a few problems. Most noteably
    the fact that the logic for detectin dryrun and intert functions
    was incorrect such that command always ran in dryrun mode even if
    inert was set to 0. After fixing this I discoverd a number of other
    issues around the fact that commands were no longer running during
    dryrun mode even when they needed to.
    More information on a file by file basis.
    M    trunk/lib/Spine/
            -do_rsync needs to support inert since it can be called
             in cases that need to run even during dryrun
            -Fixed typo wraper => wrapper
            -exec_command was only used by RestartServices, updated
             RestartService to just use simple_exec directly. Checked
             the existing TM data tree and found no uses of the
             exec_command function in templates.
    M    trunk/lib/Spine/Util/
            -Fixed detection of dryrun mode (c_dryrun not dryrun).
            -_readlines would error if we were in dryrun, no returns undef
             if we are running in dryrun mode.
    M    trunk/lib/Spine/Plugin/
            -Call do_rsync with inert=1 when copying the content of c_root
             to the temp overlay.
    M    trunk/lib/Spine/Plugin/
            -Call simple_exec directly instead of exec_command.
    Signed-off-by: Chet Burgess <>
Commits on Nov 13, 2009
  1. Subject: Add --quiet mode support and modify quick_template to use it

    rkhardalian authored
    Trac: Refs #4
    - Added support for --quiet mode to support future tools which may need it.
    - Modified quick_template to call spine-mgmt using --quiet.
    Signed-off-by: Rafi Khardalian
Commits on Nov 12, 2009
  1. Subject:

    cfb authored
    From: Chet Burgess <>
    Trac: Refs #100
    -Using /o in regexp isn't always safe and isn't really useful
     anymore (see the trac tickets for more details).
    Signed-off-by: Chet Burgess <>
Commits on Nov 11, 2009
  1. Subject: Prevent service restarts when in dryrun mode

    rkhardalian authored
    Trac: Refs #101
    - Fix to prevent service restarts from being triggered during a dryrun.
  2. Subject: Rename spine to spine-mgmt

    cfb authored
    Author: Chet Burges
    Trac: Refs #96
    Rename the spine binary to spine-mgmt.
    Rename most references of spine to spine-mgmt.
    Move config files to /etc/spine-mgmt/
    Note: Yet another commit message form. Hopefully the
    subject at the top will make the messages look cleaner
    in RSS readers.
    Signed-off-by: Chet Burgess <>
Commits on Nov 6, 2009
  1. Author: Richard Pounder

    cfb authored
    Trac: Refs #95
    Spine::Register::load_plugin always returns SPINE_SUCCESS instead
    of the actual rc value generated from attempting to laod the
    Signed-off-by: Chet Burgess <>
  2. Refs #65

    cfb authored
    If we exceed the max nesting depth while processing includes
    log an error and abort the run.
Commits on Nov 5, 2009
  1. - Added a new argument to getvals to force return of an empty array, …

    rkhardalian authored
    …if desired.
    - Force privfiles call to return an array, #refs 90
  2. While SystemHarden is running there is no indication as to what spine…

    cfb authored
    … is doing.
    SystemHarden will now print an informative message telling you it is running.
  3. Fix for issue #78.

    cfb authored calls syslog() without calling openlog(), which Sys::Syslog considers
    a no-no.
    This spine-config now calls openlog and closelog. Additionally openlog is 
    called with user configurable options for facility, syslog opts, and the ident
    string. These can be configured in the config file under the Spine section
    using SyslogIdent, SyslogFacility, and SyslogOptions.
    Patch provided by Alexander Toth (atoth).
Commits on Nov 4, 2009
  1. Fix for issue #91.

    cfb authored
    There is no reason why /config should be REQUIRED and produce a fatal error
    if it doesn't exist.
  2. Fix for issue #80.

    cfb authored
    Use lstat instead of stat in order to ensure we are testing the permissions
    of the symlink and not the underlying file. The underlying file should be 
    caught about the search anyway. 
    Additionally log if we encounter a symlink with the SUID and/or SGID bit set
    as we can't reliably change the permissions of the symlink.
  3. Fix for issue #89.

    cfb authored
    SystemInfo will now look for a key called network_path that provides the path
    to where the the network directory is located. If the key is not present the
    code will default to simply "network" (the current behavior). 
    The network_path key will need to be placed in the spine_internal directory as
    SystemInfo runs during the DISCOVERY/populate phase before the tree is parsed.
  4. Fix for issue #77.

    cfb authored
    There are 3 primary changes.
    1) There is now a config file option called Parser under the spine section 
    to control what type of TT parser is used. The current values are pureTT
    and lameTT. The code defaults to pureTT. pureTT does not call the 
    convert_lame_to_TT function. 
    2) The convert_lame_to_TT function will now store the name of every key it 
    encounters that has lame syntax in the c_lame_keys key. This key can be used 
    by existing data tree to help clean-up their trees and remove existing 
    lame syntax.
    3) At the end of a spine run the count of lame keys encountered will be 
    printed if any lame keys were encountered. This will help in the clean-up 
    of existing trees.
  5. Fix for issue #82.

    cfb authored
    Remove the need for File::Touch in favor of utime which is a perl
  6. Fix for issue #84

    cfb authored
    RPM treats an extra space after -e as the name of packages to try and remove.
    To fix this we sanatize rpm_opts to ensure we don't have any extra spaces
    that RPM will interpretty as package names.
Commits on Nov 3, 2009
  1. Fix for issue #87.

    cfb authored
    The call to simple_exec for executing dmidecode in the get_hardware_platform
    function of was missing arguments.
  2. Fix for issue #85.

    cfb authored
    Don't remove the template file from disk if invoked via
    Additionally numerous formating changes regarding the use of {}.
Commits on Aug 27, 2009
  1. Fix svn props

    richard authored
Commits on Aug 26, 2009
  1. Ticket #55

    richard authored
    Starting from a blank config and adding each requirement in the code will now give useful errors and stop trying to run if fatal issues are detected.
    More work needed on the Auth plugin. The few changes made here just make sure it doesn't genarate perl errors.
Commits on Aug 25, 2009
  1. Quite a large commit to close Ticket #52 (standardize exec)

    richard authored
    Also removed Spine::Parser as it is not used and will be replaced in 2.2
    Tests show it to work cleanly but as this is a large change heavy QA is needed.
    Currently yum changes are untested.
Commits on Aug 24, 2009
Commits on Aug 19, 2009
  1. Fixes for issues #73.

    cfb authored
    Sort deps by service/command, evaluate all deps
    for the same service/command and only execute the restart
    once if any dep is met.
Commits on Aug 18, 2009
  1. 0xDEC0DE pointed out that sorting @TEMPLATES twice was

    cfb authored
    less than optimal. 
    We now sort it once and use the sorted copy.
  2. Fix for issue #72.

    cfb authored
    sort the list of files returned by find before diffing them. This will 
    ensure a uniform output between identical systems.
Commits on Aug 17, 2009
  1. fix missuse of 'my'

    richard authored
  2. clean out tabs

    richard authored
Commits on Aug 14, 2009
  1. Final round of fixes for issue #7.

    cfb authored
  2. More fixes for issue #7.

    cfb authored
    This is another plugin that makes heavy use of Ticketmasters
    notion of host organization.
  3. Removing ClassInstance per issue #7.

    cfb authored
    ClassInstance calculates the the number of class instances
    in a given cluster/product combination. 
    These are very Ticketmaster specific concepts and don't 
    translate to a generic form.
  4. Fix for issue #49.

    cfb authored
    We are removing the Ticketmaster specific FilerExports plugin and 
    replacing with with a generic NFSExports plugin.
Commits on Aug 5, 2009
  1. Used for testing Spine::Registry and Spine::Data (tests/spine-core.t)…

    richard authored
    …. It might be better to move this to tests/lib/Spine/Plugin since it's not needed for production
Something went wrong with that request. Please try again.