commit 3019f0480fe1cfcb0f174ac54af90fa8a7c64752
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Tue May 30 16:20:42 2017 +1000

    Automatic commit of release 24.3

commit 9c093eb8f615dac67be050d264403db30c21e535
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Mon May 29 11:35:15 2017 +1000

    release notes for 24.3
    
    Change-Id: If1d90c0b7a529aa016d3d42e70870693b3aff557

commit 66a63d25412b4e49ec62fd655eba0d2191909e4a
Author: Jonathan Toppins <jtoppins@redhat.com>
Date:   Mon May 1 18:14:06 2017 -0400

    server update-inventory: always force correct hostname
    
    When running the inventory update task the running task needs to
    communicate back to the beaker server which machine entry the updated
    inventory should be applied to. This is done by locally (on the machine
    running the task) pulling the configured hostname. This however can
    result in many issues, lets say the kickstart script is custom and so
    overwrites /etc/hosts resulting in the hostname being reported always as
    host1.example.com. Or the dhcp server being used does not issue
    hostnames in sync with the entries in beaker.
    
    All of these systems are outside the control of beaker, and can result
    in the update inventory task failing simply because the task cannot tell
    the server which machine entry the information should be associated with.
    
    Oddly enough the inventory update task allows for someone to override
    the hostname used with a task parameter. We change the beaker server
    to always issue this task parameter, eliminating all these subtle issues
    for how this task could fail.
    
    Bug: 1438666
    Change-Id: Ie0eaaa54611a1ac0ea452ad975d2d2adaade4c39
    Signed-off-by: Jonathan Toppins <jtoppins@redhat.com>

commit 969ac8b6a1d8f707a463751a11b7129ae3043fd9
Author: Jonathan Toppins <jtoppins@redhat.com>
Date:   Mon Apr 10 18:31:34 2017 -0400

    server update-inventory: make dryrun not check if already scheduled
    
    Normally when update-inventory is run a check is performed to determine
    if an inventory update is already scheduled for the system. If an update
    is, the server will respond with a 409 error saying an update is already
    scheduled.
    
    However, if dryrun is true the check is still performed. This seems
    unnecessary and prevents someone from running the command just to
    generate the xml. This change makes it so the check is not performed
    when dryrun is True.
    
    Bug: 1335394
    Change-Id: I71cc8af01f4087b7b7a250d74d1004a876f4148f
    Signed-off-by: Jonathan Toppins <jtoppins@redhat.com>

commit c51b5bdfc25874326b077f62565f4a9bb73cffa8
Author: Jonathan Toppins <jtoppins@redhat.com>
Date:   Mon Apr 10 23:57:54 2017 -0400

    bkr-update-inventory: report human error message to user
    
    update-inventory allows the user to specify N number of servers on
    the command line to trigger an inventory update for. This means that
    the update-inventory command needs to catch any errors received back
    from the server so they don't cause the program to terminate early.
    
    Therefore, we need to reuse the exception handling used in
    Client/src/bkr/client/main.py so that we print the human readable
    error message instead of just the HTTP status code and its textual
    version.
    
    Bug: 1335394
    Change-Id: I58ae3a8c7670dde44e72bfc2dc846d014e30a763
    Signed-off-by: Jonathan Toppins <jtoppins@redhat.com>

commit 5c83fda1026c9fa7cb8c3df22ab6977d0e4b137e
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Mon May 22 17:25:25 2017 +1000

    use mirrorlist not metalink for Fedora updates repo
    
    This matches what the official Fedora kickstarts are doing. Starting
    from Fedora 26 Anaconda will no longer silently accept metalink URLs in
    the --mirrorlist option so we need to get it right.
    
    Bug: 1442146
    Change-Id: Ia82b6506d9628a670e780c41e445d2405e2bb6ee

commit 5c3e7c45015fb43da764af99ab57d2658082f59e
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Mon May 22 16:36:44 2017 +1000

    don't reject host filtering options when --machine is given
    
    Bug: 1234323
    Change-Id: I48a8e7e12dff8fd34777294c3c43ec283ee9d611

commit ff8f090ab760fc9085382cac8691ad1a037c10ae
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Thu May 18 17:15:11 2017 +1000

    anamon3: read files in binary mode
    
    This way we only ever deal in raw bytes, not decoded strings. The file
    contents are passed back to Beaker and stored as raw bytes.
    
    Also dropped the unnecessary MD5 stuff (beaker-proxy has always ignored
    this parameter).
    
    Bug: 1442147
    Change-Id: I5f70f0c7b262218de08319e3f2ad8a1993818af6

commit 5ba2001cdd9837a45ab810d8d3eb7aa7f0b711c3
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Thu May 18 16:42:43 2017 +1000

    add httpd alias for anamon3
    
    Bug: 1442147
    Change-Id: Iad6f706d214e10eddde8f1b133012370d1e96738

commit a700b73c840e16577093b7e51a677963befea933
Author: Dan Callaghan <dcallagh@redhat.com>
Date:   Thu Apr 20 15:48:11 2017 +1000

    return correct id in system JSON representation
    
    Previously Power.id was overwriting System.id if the user was permitted
    to see power settings, because the System.__json__() method does
    data.update(self.power.__json__()).
    
    Bug: 1386074
    Change-Id: I0ee4c762c637953a16434f0d34f88b22a04a734f

commit e4a99ff0503ed76c90c24de4222447ed6a445590
Author: Johnny Bieren <jbieren@redhat.com>
Date:   Wed Mar 22 11:02:06 2017 -0400

    Make anamon script work for both python 2 and python 3
    
    (jbieren author of anamon3 modification)
    Use the right branch this time
    Add kickstart expected file changes
    modify solution to deal with quoting and still be safe
    Fix whitespace issue to match expected output
    
    Bug: 1442147
    Change-Id: I7ea8e597d45bd9c79982b05a7ccffe999b88d229