commit 4637be95c2a5f9776aad86ab155484f9dc405a9f
Author: Matt Tyson <mtyson@redhat.com>
Date:   Wed May 23 14:28:25 2018 +1000

    Automatic commit of release 25.4
    
    Change-Id: Ib57e59373ce6fc27c8484873d6e1faf7d3b6c4e5

commit d328127e626b7a8483fcdab5c860f074d54de34a
Author: Matt Tyson <mtyson@redhat.com>
Date:   Tue May 22 14:11:11 2018 +1000

    Beaker 25.4 release notes
    
    Change-Id: Ifc1f3df3a8207d3bdd6bce4fa82cf9e37e1152be

commit 48cda1680e51910ec54e147574de80e94a7f7007
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Tue May 22 13:08:05 2018 +1000

    beakerd: avoid holding open a database transaction in the outer loop
    
    At the top of the outer loop functions beakerd was issuing a query
    without any corresponding session.begin(), which means the connection
    and its corresponding transaction was being left open until the next
    invocation of a beakerd outer loop function.
    
    If beakerd starts to get bogged down processing a large number of New
    recipes, as in bug 1574772, this means that the transaction outside the
    loop can be held open for many hours, which then makes everything worse
    as InnoDB has to maintain a very long transaction history.
    
    Change-Id: I27503a50735b2a7b9fa29fee2b96c4fee53499c8

commit 2a80be902a9115205dbaeedf4af97778cfd65bff
Author: Róman Joost <rjoost@redhat.com>
Date:   Fri May 18 16:25:31 2018 +1000

    Revert "Parameter to verify import with STATUS"
    
    This reverts commit 8014f2a04d638ec0a7ba26939a727b1d39728dbb.
    
    Change-Id: I8926869c46ec7072eecf037c28244ce8e793ad2d

commit 2e53b92ecd1fbfc4d46d000f01210cc825e01874
Author: Róman Joost <rjoost@redhat.com>
Date:   Fri May 18 16:25:26 2018 +1000

    Revert "Import separate RHEL8 composes"
    
    This reverts commit 1f41746173b84a8c707677f3e3cd4a047da471b5.
    
    Change-Id: I7a065ba3e7db61ac9af7feeef8abaafa8ba309ce

commit 9aac29b3081bb26f6ac6dec0c6bb87dab95b945e
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Thu May 17 20:26:56 2018 +1000

    Revert "bkr workflow-reserve: reserve a system"
    
    This reverts commit 33c577b567b48f28e4af0f1e3d156cc07a373fe3
    as it has introduced some performance problems due to calling
    System.distro_tree_for_inventory() frequently. See bug 1579289.
    
    Change-Id: I3f6666744bac2e410df098134f3e81a79e6daa45

commit 12f019679a63792c07aa564b25aa27a6d6dc0ea8
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Mon May 14 16:15:05 2018 +1000

    avoid loading system.installations in the provisioning code path
    
    Similar to commit 1908d023, although this case is easier because there
    is no custom ordering for that relationship.
    
    Change-Id: I42539e75480c071f973ee5db02448b39620a0c56

commit 5fca62b77a22e2b94ec8de2bc9288284decb688b
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Mon May 14 16:13:25 2018 +1000

    optimise beakerd queries to find Processed and Scheduled recipe sets
    
    Bug: 1573081
    Change-Id: I2c9b96c6a40cccaa25fc5f15ce898f301e4278a7

commit 353a3eb6e1c7199b5007e1f9f4c0d75353479f93
Author: Róman Joost <rjoost@redhat.com>
Date:   Mon May 14 16:04:13 2018 +1000

    Fix interface error
    
    Our MySQL Python bindings throw an interface error if the underlying database
    throws an error they don't know how to handle:
    
    InterfaceError: (_mysql_exceptions.InterfaceError) (-1, 'error totally whack')
    [SQL: 'INSERT INTO retention_tag (id, expire_in_days, needs_product) VALUES (1,
    0, 0)']
    
    The first interface error uses the same fix as in
    https://gerrit.beaker-project.org/#/c/5951/.
    
    The second interface error is related to the first, but since the current code
    base does not have the attribute anymore, the patch addresses it in the
    migration itself. The downgrade was therefore changed to adding the column as
    non-nullable, which assigns a Boolean value for existing entries and then alter
    the column to be NULLable.
    
    Change-Id: I8487efa03f46bc7b6daf880615c4ffd1e842da96

commit 10822e2b5b68f7174fa5724ae91710b8519c80c8
Author: Róman Joost <rjoost@redhat.com>
Date:   Mon May 14 15:40:21 2018 +1000

    Provide a default when inserting
    
    MariaDB will raise an error if the default value is not given, but the database
    expects one. See: https://mariadb.com/kb/en/library/insert-default-duplicate-values/
    
    Change-Id: Ie6b61cbded907c39230cb9a88e379c44c3c4f606

commit d881b33157818f2902a45ec51149615a8d28df0d
Author: Róman Joost <rjoost@redhat.com>
Date:   Mon May 14 15:27:24 2018 +1000

    Drop the index first before dropping the colum
    
    MariaDB 10 now requires us to drop the index first, before dropping the column.
    Otherwise the column drop will result in:
    
        OperationalError: (_mysql_exceptions.OperationalError) (1072, "Key column 'fw_version' doesn't exist in table") [SQL: u'ALTER TABLE device DROP COLUMN fw_version']
    
    Change-Id: Ib07d09f86394912f7a9e511df8b649cc17f76fbe

commit 717bfc4c68f07a60010c7a05c2e28a5cb4fdf5c0
Author: Róman Joost <rjoost@redhat.com>
Date:   Mon May 14 14:08:01 2018 +1000

    Truncate Activity field_name to 40 characters
    
    This will otherwise result in an exception on MariaDB.
    
    Bug: 1577729
    Change-Id: I87f4c9f84055a65887db815dd380259819805a9b