Skip to content
Commits on Feb 5, 2016
  1. @domenkozar

    Merge pull request #389 from nathan7/virtualbox-sized-baseimage

    domenkozar committed
    Respect deployment.virtualbox.disks.*.size for images with a baseImage
Commits on Feb 2, 2016
  1. @edolstra

    Use hvm-s3 AMIs when appropriate

    edolstra committed
    We were silently ignoring ebsRoot = false for certain instance types.
Commits on Jan 27, 2016
  1. @edolstra

    Show version in the manual

    edolstra committed
  2. @edolstra
Commits on Jan 26, 2016
  1. @nathan7
  2. @nathan7
Commits on Jan 20, 2016
  1. @rbvermaa

    Add command show-arguments that tries to figure out which arguments c…

    rbvermaa committed
    …an be passed to a nixops deployment. Tries to determine the arguments based on the require attribute of nixops networks. Does not work yet for cyclic requires.
Commits on Jan 19, 2016
  1. @rbvermaa

    Update nixops version

    rbvermaa committed
Commits on Jan 18, 2016
  1. @rbvermaa

    Merge pull request #383 from domenkozar/update-changelog

    rbvermaa committed
    Release notes for 1.3.1
  2. @rbvermaa
Commits on Jan 14, 2016
  1. @domenkozar

    Release notes for 1.3.1

    domenkozar committed
Commits on Jan 12, 2016
  1. @edolstra

    Merge pull request #379 from mayflower/jobs-wrapper-removal

    edolstra committed
    fix reference to obsolete `jobs`
Commits on Jan 10, 2016
  1. @globin
Commits on Jan 4, 2016
  1. @rbvermaa
Commits on Dec 18, 2015
  1. @rbvermaa

    Merge pull request #374 from domenkozar/libvirt/improvements

    rbvermaa committed
    Libvirt improvements
Commits on Dec 17, 2015
  1. @domenkozar

    libvirt: add headless mode

    domenkozar committed
  2. @domenkozar

    libvirt: add basic logging

    domenkozar committed
  3. @domenkozar

    libvirt: use more reliable method to retrieve IP address

    domenkozar committed
    Previosly a special path was used to parse DHCP leases. This file
    may not exist (it doesn't for me on NixOS master branch). Instead
    we now run virsh command to parse the ip for the machine.
  4. @domenkozar
Commits on Dec 16, 2015
  1. @domenkozar
Commits on Dec 15, 2015
  1. @domenkozar


    domenkozar committed
Commits on Dec 8, 2015
  1. @aszlig

    backends: Unify call to super.get_ssh_flags().

    aszlig committed
    While we're at it, let's take 9792d9a and 880fc52 even further and pass
    both *args and **kwargs for _all_ backends.
    Signed-off-by: aszlig <>
  2. @aszlig

    Merge pull request #371 from kevincox/ssh-args.

    aszlig committed
    This passes non-keyword arguments in addition to keyword arguments to
    get_ssh_flags() as well, continuing on 880fc52.
  3. @aszlig

    tests/hetzner/rescue-image: Update backport debs.

    aszlig committed
    Although the Debian stable repositories change very rarely, the backport
    repositories however do not. So in the long run we probably need to move
    over to create those debs on our own or provide a third-party mirror for
    In the short run, this should fix the Hetzner backend tests.
    Signed-off-by: aszlig <>
  4. @kevincox

    Pass accept args in backend too.

    kevincox committed
    See 880fc52 for context.
  5. @aszlig

    tests/hetzner/rescue-image: Allow to auto-call.

    aszlig committed
    The Nix expression is in no way dependant on anything else than nixpkgs,
    so let's import <nixpkgs> if no pkgs argument is given.
    This makes it easier for debugging.
    Signed-off-by: aszlig <>
  6. @aszlig

    backends: Pass keyword args to get_ssh_flags().

    aszlig committed
    The superclass has a keyword argument (scp=True/False) which is used by
    upload_file() and download_file().
    So not supporting these arguments is going to fail for these backends
    and instead of passing it via a positional argument I'm using **kwargs
    so we shouldn't run into the same problem if future kwargs are added to
    This has been reported by @kevincox in an unrelated pull request (#369).
    Signed-off-by: aszlig <>
  7. @aszlig

    backends/none: Fix PEP-8 style issue.

    aszlig committed
    While this file has quite a lot of other PEP-8-related errors, things
    like this usually tend to spring into my eyes.
    This particularly fixes E251 (unexpected spaces around keyword /
    parameter equals), even though the line is still too long :-/
    Signed-off-by: aszlig <>
  8. @aszlig

    Merge pull request #364 from @basvandijk.

    aszlig committed
    Also closes #361.
    This makes sure that machines that are already deployed using the "none"
    backend get the right public key before even attempting to connect
    To make sure this works even with new deployments, I successfully ran
    the VM test for the "none" backend.
Commits on Nov 24, 2015
  1. @rbvermaa

    Merge pull request #359 from ninegua/master

    rbvermaa committed
    Change sbin/mount.vboxsf to bin/mount.vboxsf
  2. @basvandijk

    backends/none: don't use _ssh_private_key if its corresponding public…

    basvandijk committed
    … key hasn't been deployed yet
    This should fix #361.
    The boolean attribute none.sshPublicKeyDeployed is added to the
    NoneState and defaults to False. It indicates whether the SSH public key
    has been deployed to the target. If it's False the get_ssh_flags()
    function will not set the corresponding private key as the identity.
    none.sshPublicKeyDeployed will be set to True after activating the
    configuration. From then on get_ssh_flags() will set the identity to the
    corresponding private key allowing nixops to authenticate itself using
    the keypair stored in its DB.
Commits on Nov 17, 2015
  1. @ninegua
  2. @rbvermaa

    Print error message next to the error code for EC2 retries. Code is n…

    rbvermaa committed
    …ot enough information, most useful information is in the message.
Commits on Nov 16, 2015
  1. @aszlig

    scripts/nixops: Properly encode command in op_ssh.

    aszlig committed
    Regression introduced by ffbbe13.
    Using "ssh some_machine 'echo foo > /dev/null'" works, but in our case
    using "nixops ssh some_machine 'echo foo > /dev/null'" bails out with an
    The reason for this is that without allow_ssh_args being set, the
    command will end up getting escaped and passed as a single joined
    argument to OpenSSH.
    However, we don't need to do this and using allow_ssh_args will prevent
    that, especially because we already have splitted off SSH arguments
    using split_openssh_args().
    Signed-off-by: aszlig <>
    Reported-by: Eelco Dolstra <>
Commits on Nov 11, 2015
  1. @edolstra
Something went wrong with that request. Please try again.