Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Commits on Sep 11, 2012
  1. Merge pull request #39 from galthaus/pull-req-release-fred-master-2bc…

    Rob Hirschfeld authored
    …d4bf5b7
    
    Pull in some of the SUSE Fixes [9/13]
Commits on Sep 7, 2012
  1. @rhafer @galthaus

    Try to update keystone endpoints when needed (bnc#765693)

    rhafer authored galthaus committed
    By comparing the various endpoint URLs the :add_endpoint_template
    action in the keystone_register provider will try to determine if
    an endpoint needs to be updated.
Commits on Jun 21, 2012
  1. @galthaus

    Merge pull request #25 from galthaus/pull-req-release-fred-master-69b…

    galthaus authored
    …04c391b
    
    sqlite removal [1/2]
  2. @galthaus
Commits on Jun 12, 2012
  1. @galthaus

    Merge pull request #23 from saschpe/master

    galthaus authored
    Add a default system user and make use of it (instead of hard-coded values)
  2. @saschpe
Commits on May 21, 2012
  1. @galthaus

    Merge pull request #21 from galthaus/pull-req-master-b6c9b9d219

    galthaus authored
    Work around a chef 0.10.10 issue - clean up keystone UI (still more to do) [2/5]
  2. @galthaus
Commits on May 15, 2012
  1. @galthaus

    Merge pull request #20 from galthaus/pull-req-master-9a16ecbf7d

    galthaus authored
    Merge essex-hack back into development. [14/26]
Commits on Apr 4, 2012
  1. @galthaus

    Merge pull request #19 from ninja-cat/fix-readding-keystone-service

    galthaus authored
    Fix readding keystone service on every chef-client run
  2. @ninja-cat
Commits on Mar 13, 2012
  1. @galthaus

    Merge pull request #18 from ninja-cat/fix-endpoint-issue

    galthaus authored
    Fix endpoint issue
  2. @ninja-cat

    Fix endpoint adding: remove wrong string to integer conversion.

    ninja-cat authored
    String with hexadecimal values becomes 0 after this conversion.
Commits on Mar 9, 2012
  1. @galthaus

    Merge pull request #17 from galthaus/c8d876d5633bcf43c84dee788364c8bb…

    galthaus authored
    …3b81d619
    
    More tweaks and updates to keystone and glance, nova still needs love. [2/3]
  2. @galthaus
  3. @galthaus

    CHange booleans to True/False.

    galthaus authored
    Comment out the idle time.
  4. @galthaus
Commits on Mar 8, 2012
  1. @galthaus

    Merge pull request #16 from galthaus/b258fa740f960fe249f56e48f66f1389…

    galthaus authored
    …6804703d
    
    more nits on keystone [1/2]
  2. @galthaus
  3. @galthaus
  4. @VictorLowther

    Merge pull request #15 from galthaus/84354efa92e9a1437a77ef04e3aff65e…

    VictorLowther authored
    …af314011
    
    Make keystone errors fatal and handle error messages better [1/3]
  5. @galthaus
  6. @galthaus

    Merge pull request #14 from galthaus/52e128118711122d6d8d5d15ce4807ba…

    galthaus authored
    …dc5f22da
    
    Attempt to fix endpoints and nova.  Add nova-volume. [5/8]
  7. @galthaus
Commits on Mar 7, 2012
  1. @VictorLowther

    idle_timeout does not appear to do what it says on the tin.

    VictorLowther authored
    Based on the name, one would think that it controls the time after
    which keystone would give up trying to talk to the databases, and 30
    seconds seems to be a reasonable timeout for that.  However, that does
    not appear to be the case based on the keystone code we are using:
    
        102     def get_engine(self):
        103         """Return a SQLAlchemy engine."""
        104         connection_dict = sqlalchemy.engine.url.make_url(CONF.sql.connection)
        105
        106         engine_args = {'pool_recycle': CONF.sql.idle_timeout,
        107                        'echo': False,
        108                        'convert_unicode': True
        109                        }
    
    Instead, it seems that idle_timeout from the conf file is being
    translated into the pool_recycle parameter to the sqlalhemy
    connection, which is the time that sqlalchemy will forcibly close
    connections.  Since we are not a terribly busy environment, I can
    easily see where Keystone does not actaully talk to mysql for more
    than 30 seconds, after which point the connections are forcibly closed
    and we start getting SQL errors from Keystone.
    
    For now, just comment out idle_timeout, which will wind up having
    sqlalchemy use the defaults and never forcilbly close connections.
  2. @VictorLowther
  3. @VictorLowther
  4. @ninja-cat
Commits on Mar 6, 2012
  1. @galthaus

    Merge pull request #12 from galthaus/3a499bb4dae6cc8343bb349f0550b500…

    galthaus authored
    …08001ba2
    
    Update keystone and glance to work better and better together [2/3]
  2. @galthaus
  3. @galthaus

    Fix access object creation

    galthaus authored
  4. @galthaus

    Fix role creation

    galthaus authored
  5. @galthaus
  6. @VictorLowther

    Merge pull request #11 from galthaus/2890e9f65b9f643ec501464f0c0f0b31…

    VictorLowther authored
    …c5e30b2c
    
    Make keystone idempotent for essex. [1/3]
  7. @galthaus
Something went wrong with that request. Please try again.