Permalink
Commits on Apr 14, 2014
  1. Version bump to 6.1.3

    joshbronson committed Apr 14, 2014
    Change-Id: Ifb2a5ac562dc420d2b37b7e474d868c178a1aa8f
  2. fix: resolve move appropriately

    joshbronson committed Apr 14, 2014
    Some of the Ironfan code relies on cluster definitions remaining fixed
    after all DSL definitions are done, which means resolve! should be
    called a final time just after evaluating the DSL.
    
    Furthermore, we'd like to support re-opening realm definitions, which
    will entail calling resolve! on a realm. This will re-instantiate any
    cluster objects pointed to by the realm, so we'll need to re-register
    those clusters with Ironfan in order for the registry to continue
    working.
    
    Change-Id: I6b1203004c9b6e0ef5108b184e3df4381b079ef5
  3. Version bump to 6.1.2

    joshbronson committed Apr 14, 2014
    Change-Id: I89bbd28dce91bb57d268f5b382bceddeb3000423
Commits on Apr 7, 2014
  1. fix: do things a bit less safely during launch

    joshbronson committed Apr 7, 2014
    Ironfan swallows a lot of errors when launching. This makes it swallow
    fewer. The errors are still swallowed by Ironfan.parallel, so 'cluster
    launch' has been updated to exit with non-zero status in that case.
    
    Change-Id: I31952e469d3b6d8d801af974593e960a428b8c0a
  2. A base CentOS image can pass 'knife cluster bootstrap' with an

    Erik Mackdanz committed Apr 7, 2014
    empty run list.
    
    Change-Id: I0772766feda2cf7e3f21327c82cef3451608c36c
  3. Fix: Reraise SecurityGroupLimitExceeded

    David Snyder committed Apr 7, 2014
    - limit error was being swallowed causing a more esoteric error to be dumped later
    
    Squashed commit of the following:
    
    commit 507530469c2acd90c96a18ebdfbc7614721d788a
    Author: David Snyder <david@infochimps.com>
    Date:   Mon Apr 7 12:09:19 2014 -0500
    
        Don't swallow SecurityGroupLimitExceeded
    
        Change-Id: If9525c67c958d39886a805d9ff265821a413c114
    
    Change-Id: I64c5f73a3a2c597f90a5dd44c14bcffb196c7836
Commits on Apr 4, 2014
  1. Version bump to 6.1.0

    joshbronson committed Apr 4, 2014
    Change-Id: I368c9f118ce170ccbae14cb35cc0db3af295c80f
  2. bumping gorillib to ~> 0.6.0

    joshbronson committed Apr 4, 2014
    Change-Id: Ice150803a069cf1a7226fd935192c7909d8d7419
Commits on Apr 3, 2014
  1. version bump to 6.0.7

    joshbronson committed with Gerrit Code Review Apr 2, 2014
    Change-Id: Ib32da57cddfbd40df2bac8b7eaa42f407f6bbf0e
  2. feature: let gorillib float and remove json dep

    joshbronson committed with Gerrit Code Review Apr 2, 2014
    Change-Id: I1da2325728c44b99fabce3258499093bbcd5d393
Commits on Apr 1, 2014
  1. Fix: Duplicate Discovery

    Travis Dempsey committed Apr 1, 2014
    The output of `knife cluster show p1` was highlighting an
    error with the broker discovery process: when referencing
    more than one cluster at a time, the ec2 machine/ebs_volumes
    provider would load! more than once resulting in a false
    discovery of 'duplicate' machines/volumes'.
    
    This patch only load!s once, while still checking servers and
    ebs_volumes for respective duplicates.
    
    Squashed commit of the following:
    
    commit fc8b451645355ac48c1da9ebeab535fca57a6b34
    Author: Travis Dempsey <travis@infochimps.com>
    Date:   Tue Apr 1 14:39:10 2014 -0500
    
        only load! servers and ebs_volumes once
    
        Change-Id: Idd53b947e3068b8fb0a6dd9b76eb0a6ab72cd1de
    
    Change-Id: I971d292167ba413a1ca7003d5a64bc252f30dc9e
Commits on Mar 27, 2014
  1. Merge "fix: ironfan discovery too slow"

    David Snyder committed with Gerrit Code Review Mar 27, 2014
Commits on Mar 26, 2014
  1. fix: ironfan discovery too slow

    Chris Howe committed Mar 26, 2014
    - made test cases wor
    - pruned list of clusters before calling broker.discover!
    - tweaked discovered item caching
    
    Squashed commit of the following:
    
    commit 0d9b0b49472d566bb29dab4265d2e9a985c8b8d2
    Author: Chris Howe <howech@infochimps.com>
    Date:   Wed Mar 26 12:03:01 2014 -0500
    
        fixed the broken test cases
    
        Change-Id: Ibdcc1d87ce551c5b6cad13b801a3f7c2023c2843
    
    commit fc37cee29600cda5a5e8af3ef7b4f8a8727bc33c
    Author: Chris Howe <howech@infochimps.com>
    Date:   Wed Mar 26 11:13:59 2014 -0500
    
        pruned list of clusters before calling broker.discover!, tweaked discovered item caching.
    
        Change-Id: Ie2e51ca6055b6f0352520c09906a4de9db538a41
    
    Change-Id: Ic097c3687b1bae2946e51673e7a9c7b69b4291d2
  2. Fix: hostname resolves to private IP

    Erik Mackdanz committed Mar 25, 2014
    Change-Id: I1ad5bd6fd2b4569fa5c9a5b8dc678fd5d7047bcf
Commits on Mar 21, 2014
  1. Merge "version bump to 6.0.5"

    David Snyder committed with Gerrit Code Review Mar 21, 2014
  2. Merge "fix: string cluster names"

    David Snyder committed with Gerrit Code Review Mar 21, 2014
  3. fix: add unf as a gemspec dependency

    joshbronson committed Mar 21, 2014
    Change-Id: I978ce0c23ff1d7568c29d3270db78f0c46d3eac1
  4. version bump to 6.0.5

    joshbronson committed Mar 21, 2014
    Change-Id: I145892d8534da23067510259f973afc75649595e
  5. fix: string cluster names

    joshbronson committed Mar 21, 2014
    Previously, string cluster names in realms did not work.
    
    Change-Id: I1368ad13719f2222afcc8f71d9d26a4ef6bd1d65
  6. Merge "version bump to 6.0.5"

    joshbronson committed with Gerrit Code Review Mar 21, 2014
  7. Merge "Fix: cloud-init in openstack uses top-level userdata"

    joshbronson committed with Gerrit Code Review Mar 21, 2014
  8. Fix: cloud-init in openstack uses top-level userdata

    Erik Mackdanz committed Mar 21, 2014
    Change-Id: I79821b729326999c728e0803536d18c0738c35d6
  9. version bump to 6.0.5

    joshbronson committed Mar 21, 2014
    Change-Id: I4aa6825f0f587328b7f67adc04717de5dada1771
  10. fix: non-vpc security groups no longer use -1 ip_protocol

    joshbronson committed Mar 21, 2014
    Change-Id: I2735222e6a93f59c4cfe87f7984cff9567de3d14
  11. Fix: ensure new fog works with ec2

    Erik Mackdanz committed Mar 21, 2014
    One fix involves how security groups were being cached in ironfan
    with a prepended 'vpc-deadbeef:' but lookups are later performed
    without the prefix.  The fix is to cache the security group with
    both prefixed and non-prefixed keys, to guarantee non-breakage in
    other cases.
    
    The second fix is to text matched in an error message about duplicate
    security groups that we were matching/ignoring.  The error text
    changed somewhere upstream, so we widened the net on how the text
    is matched.
    
    Change-Id: Ie98d558fd95e1e54fe67592e10dc61ce8aac53a2
  12. Fix: apply realm changes for OpenStack too

    Erik Mackdanz committed Mar 21, 2014
    Change-Id: I88806a36308e140e568ac31b1f5555c9894f11ee
  13. Fix: openstack support requires some bumped deps in ironfan

    Erik Mackdanz committed Mar 21, 2014
    Change-Id: I397a53f13171983790a74d991b1d5b450d0b91cc
Commits on Mar 20, 2014
  1. version bump to 6.0.4

    joshbronson committed Mar 20, 2014
    Change-Id: I0b5ede31a99f64e5e2d99aea5f625aa4a4f26f80
  2. fix: ec2/openstack keypair creation

    joshbronson committed Mar 20, 2014
    via Chris Howe. We were looking for the right keypair when trying
    to pull it down to use it, but the wrong one when creating it.
    
    Change-Id: I965394a7fdb6b48dd4e5f60d7296b41916dd8432
Commits on Mar 19, 2014
  1. version bump to 6.0.3

    joshbronson committed with Gerrit Code Review Mar 19, 2014
    Change-Id: I7704fbe77ea21de185a2f0f31d9fa6ded7d02b34
  2. fix: component security group corrections

    joshbronson committed Mar 19, 2014
    1. Security groups shouldn't be created if 'create_security_groups' is
       unset. This will protect us from creating too many rules in VPC.
    2. The component class needed some updates for real realms. This was
       the source of some spurious security groups.
    
    Change-Id: Ie9f55e3da4aad81b5bcb2943848d18d057937fe2
  3. fix: ssh keypair names should include realm and cluster name

    Chris Howe committed Mar 19, 2014
    Squashed commit of the following:
    
    commit cef6df75ee50f5c8be7ea763b8938e069633867a
    Author: Chris Howe <howech@infochimps.com>
    Date:   Wed Mar 19 11:55:15 2014 -0500
    
        Added a full_cluster_name method that includes the realm name and the cluster name
    
        Change-Id: Ifbf8c184090d2d809ca3850803d0d7cdc6ff96d8
    
    Change-Id: I6621ed898f867a5388409da6da1856d16bad2c76
  4. version bump to 6.0.2

    joshbronson committed Mar 19, 2014
    Change-Id: I1b68b820b2e8a098fce21fbefe57db5f83f424b6
  5. fix: fixing cluster and facet role names

    joshbronson committed Mar 18, 2014
    and delete Gemfile.lock
    
    Change-Id: I24cf9ad6c5d1006c0de4a04b2f8014c134df61b5