Because Vagrant no longer determines `all_disks` through `vboxmanage
list hdds`, it can't rely on unattached disks existing in `all_disks`.
This commit fixes that by including the results from `list hdds` to
determine if Vagrant needs to reattach a disk created by Vagrant that
failed to be attached previously.
This commit updates how Vagrant manages disk state with the virtualbox
provider. Instead of using the raw structure from `list_hdds` for
all_disks, it instead uses that data that now lives inside the
controllers attachment structure.
This commit includes all detailed disk information from VirtualBox
inside its storage structure in Vagrant. This gives Vagrant more
detailed access for a disk beyond its uuid and port/device such as its
capacity and the type of disk it's classified as.
This commit updates how the virtualbox provider obtains `all_disks`.
Instead of using the `list_hdds` driver method, which lists al known
harddisks for an entire VirtualBox installation, it instead uses that
array of disks to filter out the proper disks associated with the guest,
given its existance inside the guests controller attachments.
This commit adds a unique error message for an empty box value. It
requires modifications to vagrantfile.rb because some Vagrantfile config
is used before validation occurs.
This fixes an issue where having a box name set to an empty string will
cause all Vagrant commands to fail with an error like:
ArgumentError: Malformed version number string (random box name)
This may be related to #10663.
- removed superfluous cap/shell_expand_guest_path
- solaris11 guest_capability for shell_expand_guest_path now directly returns the linux capability class
Use subject instead of invalid_subject because the validation assertions
test the subject double.
This also adds an additional check when validating the `size` attribute
because it is only required for disks of type `:disk`.
SCSI controllers are a lot like SATA controllers. This commit also
changes some controller detection logic to take boot priority into
account when selecting an appropriate controller.
Move all storage bus logic into the storage controller class. Since most
of the storage controller interaction only cares about the storage
controller name, we can simplify #get_controller and isolate the storage
controller detection-type logic in the StorageControllerArray.
This commit adds a new error message to be raised if a VM has no
supported storage controllers. This lets us differentiate between two
different "controller not found" scenarios:
1. If we are looking for a controller that we're expecting to find (i.e.
one that was recorded in the disk metadata file)
2. If we are poking around for the *best* controller to use in a
configuration task
This preserves the existing behavior of the method but changes the
implementation to use #read_storage_controllers. The caps that call a
mix of #get_port_and_device and #read_storage_controllers may be a
candidate for further refactoring, but this makes sure that we're
fetching storage attachments consistently.
.configure_disks is responsible for determining the disk controller(s)
to use, given the machine configuration and the current disk config.
When a machine has a single controller, use it for all attachments.
When a machine has multiple controllers, attach disks to the SATA
controller and DVDs to the IDE controller.
This commit also returns additional attachment information
(controller/port/device) with the disk metadata.
This makes it easier to check if the required controller can be found,
and automatically raise an error if it is not.
Add a #storage_bus method to the StorageController class as a shorthand
way to check the general storage controller type.
This commit adds a new error type that can be raised whenever a storage
controller of the required type is not found. This indicates that a user
needs to either add the storage controller manually or change their disk
configuration.
It also removes the last hardcoded instance of "SATA Controller" as a
default argument.
This commit adds a new VirtualBox provider helper method to return a
list of storage controllers so Vagrant can find a storage controller
with the desired characteristics (type IDE or SATA).
This still needs to get wired up to the disk cleanup method.
This builds on the existing disk functionality, and adds some special
IDE controller-related flavor.
Considerations for IDE controllers:
- Primary/secondary attachments, so that each port can have two devices
attached
- Adding the ability to address a specific controller name for disk
attachment
This also prevents a user from attaching multiple instances of the same
ISO file, because VirtualBox will assign each of these the same UUID
which makes disconnection difficult. However, if multiple copies of the
ISO are attached to different devices, removing the DVD config will
cause the duplicate devices to be removed.
We may want to consider additional work to make the storage controllers
truly generic.
At the time of writing, the Haiku rsync lacks old-style --compress due
to its external zlib. Pass `-zz` to the guest rsync.
This works with or without the `--compress` flag in the host's
`rsync__args`.
Prior to this commit, the default value of disk_ext was set in the
finalize! method, and was really only valid for the virtualbox provider.
This commit updates that by moving the step into the validate function,
which has access to the machines provider.
This commit changes the SSH communicator to raise an error if Vagrant
doesn't receive an exit status from an SSH command, for example if the
command is terminated by the OOM-killer.
* hostname is a boolean
* a network that sets hostname should have a static ip address
* only one network may set hostname
can be set at `config.vm.network :public_network, hostname: true, ip: "192.168.0.1"`
Prior to this commit, if a created but exited container bound a port,
and a new container grabed that same port (say for an ssh port forward),
when the initial container came back up it would fail because the port
also got bound to the second container. This commit fixes that behavior
by first looking at what containers are already bound prior to creating
a container.
This commit fixes a couple of issues with the shell provisioner when the
WinSSH shell is set to cmd:
- A check for the .bat extension returned by File.extname
- Execute inline scripts with PowerShell when upload_path ends with .ps1
* Ensure script has correct extension
Co-Authored-By: Sophia Castellarin <sophia@hashicorp.com>
Retains the original default value of 15 seconds for SSH connect
timeout. Allows users to modify this timeout via SSH communicator
option. Enforces integer values for timeout and validates custom
values are greater than 0.
If the VirtualBox guest property /VirtualBox/GuestInfo/Net/1/V4/IP
returns an IP address ending in .1, raise an error.
This addresses an issue that was revealed as an NFS error, where Vagrant
was creating an exports file with the wrong IP address. This was thought
to be caused by the presence of a docker0 interface, but it manifested
itself even without Docker installed.
This issue is difficult to reproduce, but hopefully this PR will get us
closer to the root cause.
With the adjustments provided in #11455 the location of injecting
within the call stack has been changed slightly. With the entire
stack now being generated before execution instead of dynamic
hooks being wrapped around actions at run time, this update
ensures that the method is called correctly after the entirety of
the synced folders action has completed.
When user is using podman's docker CLI emulation the containers would
fail to enter running state because the docker driver could not catch
the container name. This commit fixes that by adding a check if podman
docker emulation is used and pick the container hash correctly from the
output.
When buildkit is enabled docker will write out
`writing image <image id>`
When buildkit is not enables docker will write out
`Successfully builld <image id>`
It is more likely that searching for the `writing image` will not
clash with build output than when searching for `Successfully built`.
eg. when installing python packages with pip, it is common to use
the verbage `Successfully built`.
This adjusts how triggers are implemented during a normal run. Any
defined triggers which are applicable are located and injected into
the run stack as the stack is built, including hook type triggers.
Support is included for dynamic hook lookup.
The data type used when defining triggers has also been relaxed to
support symbols, strings, or constants.
This patches the connection instances generated for the winssh
communicator so when a command is executed it is always run
with powershell. This prevents inconsistencies with argument
handling based on what the default shell is set to on the remote
side.
Since powershell is the default, environment variable template
only needs to be set for powershell style. If the shell setting
is updated to `cmd`, the command will be properly prefixed.
Default shell has been updated to powershell, to prevent extraneous
wrapping where it's not required. The `#ready?` check has also been
updated to use a constant value, which is overridden within winssh
as a blank command is invalid.
SFTP is more reliable than SCP when dealing with unknown sshd
configurations on the server side. It also provides a convenient
facility for creating remote directories, so we can remove the
Windows-specific `#create_remote_directory` method.
This commit normalizes our Windows paths to use `/` instead of `\`.
These paths are compatible with both cmd and PowerShell, and are
required if the server-side shell is set to Bash.
The OpenSSH server executes all commands inside a default login shell
which cannot be controlled by the Vagrant configuration. So what ends up
getting executed on the server side looks something like this:
"c:\\program files\\git\\bin\\bash.exe" -c "C:\\Windows\\Temp\\vagrant-ssh20200130-41670-1w5nsjy.bat"
By flipping the direction of the directory slashes, we end up with:
"c:\\program files\\git\\bin\\bash.exe" -c "C:/Windows/Temp/vagrant-ssh20200130-43415-f1d5n2.bat"
This works whether the server-side shell is set to cmd, powershell, or
bash.
Windows commands that run over SSH are wrapped in a script that writes a
special marker to the two output streams (stdout and stderr). This
allows Vagrant to consume the output streams.
Unfortunately, this leads to a sort of chicken-and-egg problem where no
commands can be run before a wrapper script exists. For example, you
can't make a destination directory to upload the wrapper script without
first creating a wrapper script to make the directory. :)
This commit changes the behavior of the WinSSH communicator to assume
that the destination directory already exists for provisioning scripts.
It also moves the default `upload_path` from the shell provisioner
config so we can have OS-specific defaults.
Finally, it introduces a Windows-specific #upload method which will
properly use a Windows path separator on a non-Windows host.
This prevents any unexpected connection related error from breaking
the wait loop while windows reboots. If an underlying problem unrelated
to the guest is causing exceptions, the exception will still be raised
to the user, simply after the loop has exceeded the defined maximum
wait time.
Fixes#11238
Some IdentityReferences cannot be translated to
[System.Security.Principal.SecurityIdentifier] because they don't map to
a SecurityIdentifier.
An example is:
IdentityReference : APPLICATION PACKAGE AUTHORITY\ALL APPLICATION PACKAGES
In such case, it is better to catch the exception and treat it as $false.
This commit creates a custom `read_dhcp_servers` method in the
VirtualBox 6.1 driver to handle changes made in the ouput of
`VBoxManage list dhcpservers`.
Tests for VirtualBox 6.1+ can no longer use the shared examples for the
VirtualBox 4.x driver, because the `read_dhcp_servers` change is not
backwards compatible.
This commit also creates the boilerplate for a VirtualBox 6.x shared
example in case we want to put tests there in the future.
This commit moves out the recovery steps for failures when Vagrant
attempts to resize disks. It wraps itsemf in a begin/rescue in case an
additional error occurs, we can still surface the original error that
caused the resizing to fail.
This commit adds some recovery if Vagrant fails to reattach or clone a
vmdk disk that needs to be resized. It first moves the original disk
(after cloning) to a backup file. If something fails, it will reattach
the original disk to the guest and continue to raise the exception.
This commit ensures that vagrant attaches a new disk to the _next
available_ port, rather than the last one "used". This can occur if a
disk has been removed in a previous run, leaving a port open
Prior to this commit, Vagrant wouldn't validate the `type` option for
any synced_folder configs defined. This commit updates that behavior to
look at the current list of installed synced_folder plugins, and ensure
if a type is defined, it's a valid plugin in that list.
Output from recent Docker has changed:
Stderr: Error response from daemon: conflict: unable to delete 0ba49dd235e5 (cannot be forced) - image is being used by running container 250bbe980448
Fixes#7245
config
Prior to this commit, if a user had configured a provisioner that had a
config with a `name` option, it would not properly set the top level
provisioner classes name config option which would lead to some
understanibly confusing results when trying to `--provision-with`. This
commit fixes that by checking to see if the top level name isn't set,
look to see if that provisioners config defines a name, and use that
instead.
Since the root file system is marked as read-only, attempting to
link the shared directory to `/vagrant` will fail. If the guest
path is on the root file system and APFS is used, create the
link as a firmlink instead.
The Ansible package in FreeBSD is now `py36-ansible`.
If `pkg` itself needs updating, passing only one 'yes' is not enough to
confirm the installation of the package. Use the `-y` option,
specifically made for this purpose. Also add `-q` to force a quiet
output, inline with other guests.
References:
- [`pkg-install(8)`](https://www.freebsd.org/cgi/man.cgi?query=pkg-install&sektion=8)
- freebsd/freebsd-ports@cbf5576945
- By the way: Trim the generate 'pip install' command in Ansible::Cap::Guest::Pip::pip_install
- GH-11048: Fix same regression for FreeBSD and SuSE guests.
- GH-6633: Add RSpec examples to cover ansible_local 'cap' code.
Note: RedHat/Fedora guests are not covered yet by unit tests.
This commit catches the Errno::EPERM raised by the operating system if
the machine folder is inaccessible and displays it as a more friendly
error message.
This can be an issue on macOS Catalina if virtual machine files are kept
in a special directory (Documents/Downloads/Desktop) that Vagrant's
embedded Ruby is not allowed to access.
This ensures that rsync can be installed on an Alpine Linux machine where the
apk cache may not be current.
Display a warning if the vagrant-alpine plugin is installed, since Alpine guest
support has been merged into Vagrant core.
Prior to this commit, if a user had recently checked for updates, there
was no way to force Vagrant to re-check without manually deleting a
state file in the local `.vagrant` data dir. This commit fixes that by
giving users the ability to force check for updates for a given box with
a flag to the `vagrant box outdated` command.
This commit introduces a Darwin-specific template for NFS exports. This is almost identical to the standard BSD template except it puts each NFS export on its own line.
This resolves NFS issues discovered in macOS Catalina.
CentOS 8+ and Fedora 30+ no longer have the alias "nfs" for "nfs-server"
systemd service.
This shouldn't break backward compatibility, since "nfs-server" service
is available on all supported redhat systems that have systemctl binary.
Fixes#10838
Prior to this commit, docker would look for a container ID based on
"Successfully built" string. This output does not exist if a user has
enabled the experimental feature buildkit. This commit updates the build
behavior to match against both kinds of outputs, and instead of using
`scan`, it uses MatchData and groups the container id with match group
name `:id` instead of making hard assumptions with the matches being
contained inside arrays from scan.
At least on macOS combo Catalina + Docker engine 19.03.4 + Docker desktop 2.1.0.4 + vagrant 2.2.6 the original `matches = result.scan(/Successfully built (.+)$/i)` -line fails to generate a match. With this change I can `vagrant up --provider=docker` successfully.
StringBlockEditor already adds the necessary newlines. That extra
newline was making /etc/exports longer and longer, full of empty lines,
because StringBlockEditor doesn't know about it and does not remove it.
This commit fixes an issue where Gem::Version.new could recieve a nil
value if no addtional box updates are available. For some versions of
ruby, this is actually an error case. This commit fixes that by
converting it to an integer to prevent an exception.
Issue reference: https://github.com/rubygems/rubygems/issues/2359
This commit changes the behavior of `vagrant box outdated --global` so
it gets the latest version for the current machine's provider, rather
than the latest version for any provider.
Prior to this commit, if a user didn't supply a box file on disk or a
box url, Vagrant would crash and display a stacktrace with an invalid
file. This commit fixes that by adding some extra handling around the
arguments supplied to the publish command.
Check for modinfo in /sbin if it doesn't appear on the PATH.
If it's not found on the PATH or in /sbin, the command will default back to modinfo so the user sees the error message about adding it to their PATH.
This commit adds the machine guest name in the hostname validation error
message so that it's easier to see which guest in a Vagrantfile has the
validation error.
Sets `DEBIAN_FRONTEND=noninteractive` and tells `dpkg` to take the old
version of config files when installing Ansible, to prevent hangs
waiting for user input.
Fixes#10914
getent queries the system resolver for the hostname - but it's not
the resolver we're interested in. In fact, the hostname-to-be-set
may already exist in DNS (becuase DNS really is a nifty thing and
can do a lot of things which are not that possible with /etc/hosts
alone), in which case getent will "not fail" and vagrant will believe
the hostname had already been set.
Instead, query hostnamectl for the "static" hostname - that's the
one we will be setting, so we're ok IFF hostnamectl returns exactly
what we would be setting.
This commit fixes the original #11027 fix, which assumed that the
hyper-v provider just wasn't properly getting a VM id when it listed
snapshots. In reality, it was just that if you invoke the
`with_target_vm` method with no arguments, it runs on the entire environment.
This meant that the capability snapshot_list attempted to be invoked on
machines that didn't exist yet, which is the original cause for why the
list_snapshot method did not recieve a vm ID. This commit fixes that by
simply skipping the machine if it does not yet exist.
Prior to this commit, the docker action was using the method `prefix` on
an IPv4 and IPv6 address. This works fine for ruby versions 2.5 and
newer, however the ruby shipped with Vagrant is before 2.5, and
therefore the IPv4 and IPv6 classes do not have the prefix method,
resulting in an error. This commit fixes that by using a different
method of determining the prefix.
For bsd the resolve_host_path capability is a no-op. For darwin
hosts, if firmlinks are defined on the system, paths will be properly
resolved to real path.
Using the resolve_host_path capability allows hosts to modify the
shared folder directory while keeping the logic contained within
the specific host implementation.
On OS X 10.15, / is read-only and paths inside of /Users (and elsewhere)
are mounted via a "firmlink" (which is a new invention in APFS). These
must be resolved to their full path to be shareable via NFS.
/Users/johnsmith/mycode => /System/Volumes/Data/Users/johnsmith/mycode
We check to see if a path is mounted here with `df`, and prepend it.
Firmlinks are only createable by the OS and this wasn't supposed to be
visible to applications anyway:
https://developer.apple.com/videos/play/wwdc2019/710/?time=481
Prior to this commit, the docker compose build method would not properly
set build_args if given in a Vagrantfile. This commit fixes that by
using the passed in key `extra_args` from the docker build action.
If VirtualBox is installed but the kernel module is missing or the
service is stopped, the usability test should fail without crashing so
we can fall back to other providers.
Signed-off-by: Anders Kaseorg <andersk@mit.edu>
Prior to this commit, if you ran Vagrant in a different current working
directory other than where a current guest machines location is, the
file provisioner would not take into account the machines local dir, and
would instead use the path where Vagrant was invoked to expand the
`source` path option for a file provisioner. This commit fixes that by
passing the root path `machine.env.cwd` when expanding the source dir.
Prior to this commit, the `pip_install_cmd` option for ansible guest
config was not properly finalized. This commit ensures that if that
value is still UNSET_VALUE, it gets set to empty string in the finalize!
method.
Prior to this commit, if you tried to save a snapshot without giving it
a name, the hyper-v driver would not properly obtain a vm id to save a
snapshot on, resulting in an error. This commit updates the command
snapshot save to hold onto the machines ID in argv rather than `pop` it
off, so that the hyperv driver can obtain the guests id when saving a
snapshot.
Prior to this commit, due to a fix that ignored `.vagrant` with rsync
helper, it broke the ability to watch for changes in subdirectories when
running the rsync-auto command. This commit puts back some of the helper
methods that were there previously for a given watcher path to ensure
that all files and subdirectories are properly watched and synced.
- nfs.service got recently removed in openSUSE Tumbleweed and calling service
restart nfs errors out on Tumbleweed. nfs.service has been an alias to
nfs-client.target for a very long time and can thus be safely substituted.
- all actively supported versions of openSUSE & SLE are using systemd now
=> no reason not to use systemctl
In commit 7980178d19 (#10879) I added a
`usable?` class method to `VagrantPlugins::DockerProvider::Provider`.
However, commit 34e53a5a4b (#10890)
incorrectly changed it to an instance method. This rendered it
ineffective because it’s called on the class, not an instance. Change
it back to a class method.
Signed-off-by: Anders Kaseorg <andersk@mit.edu>
Prior to this commit, if there was a config error inside a provider
block, Vagrant wouldn't grab the right backtrace token on windows since
the api is different for ruby on Windows compared to all other
platforms. This commit ensures that the proper line number is chosen so
the error message is correct.
Prior to this commit, the docker login action assumed that if there was
a password to authenticate with, Vagrant was using a host vm to run
docker. This is likely due to some legacy decisions with how Vagrant
used to manage running docker. This commit fixes that by only grabbing a
host_vm lock if the host_vm is actually in use, otherwise login
normally.
Prior to this commit, `vagrant box prune --force` would not prompt a
user to prune Vagrant boxes, even if that box was in use. There was no
way to prune boxes, ignore the prompt, but keep in-use boxes. This
commit adds a new flag that can be combined with `--force`, that will
keep in-use boxes but prune older boxes without prompting the user.
The workaround for the broken repository should be safe to be removed,
since the last affected Alpine version (<=3.3) EOL'd in November of 2017.
The remaining important commands can be split out into seperate calls
of sudo(), which removes the need for manual exit-code checking
(since it aborts by itself when a command fails) and makes the code
easier to handle in general.
Prior to this commit, the chef_installed capability was looking for a
string that has recently changed in newer versions of chef. This commit
fixes that by instead just looking for the right version that was
configured for the chef client, rather than the specific string that
could change again in the future.
Prior to this commit, the BSD options for the /etc/exports file was not
properly set up in the correct order for a given entry. It was however
fixed for FreeBSD, but never for the original exports template. This
commit fixes that by unifying the BSD and FreeBSD templates.
Prior to this commit, the rsync helper expanded all exclude paths that
should be ignored to be full qualified and regexp escaped. However the
Listen gem expects these ignore paths to be relative to the path passed
into the listener, not a full path. This commit fixes that by using the
path given by the user for the `rsync__exclude` option
Prior to this commit, if a user specified that their `rsync__excludes`
option was an empty array, Vagrant would treat that as if it included
options inside the array rather than ignoring it. This commit fixes that
by only adding the excludes option when it exists and is not empty.
Prior to this commit, if a user ran a `vagrant box update` on their
entire environment and one of the boxes did not have a metadata file,
the rest of the boxes in the update would be skipped. This commit fixes
that by ignoring those boxes and showng a warning, so that the rest of
the boxes could check for updates.
Prior to this commit, Vagrant would list all machine snapshots in a flat
list, without showing which snapshots were associated with which guests.
This commit fixes that by placing some separation to make it clear which
snapshots belong to which guests.
Prior to this commit, the docker compose driver would _always_ path
expand a host volume no matter what. This is not always the correct
option, for example if that host volume is actually a reference to a key
inside a `volumes` hash instead of a path on disk. This commit changes
that by looking to see if the requested host volume is actually a
defined key inside the compose config, and if not, it will path expand
it like before. Otherwise it will leave the key "as is".
This prevents credential information from being persisted into the
local data directory which is used during subsequent runs to determine
folder definition changes.
Prior to this commit, if a snapshot restore was run on an entire
environment with some non-existent guests, Vagrant would attempt to list
their snapshots with a nil id. This commit fixes that by returning an
empty list of snapshots if the machine has not been created yet.
According to ifconfig(8), to list only Ethernet interfaces, excluding
all other interface types, including the loopback interface, the command
to use should be:
ifconfig -l ether
Related to: #8760
This is a follow-up of #10717 to use the same naming convention as on
Linux guests, in order to reduce the diffs.
Also adds the missing capability to `unmount_virtualbox_shared_folder`
on FreeBSD guests.
Use `mask` option for defining subnet on network configuration. Allow
options to be passed through using hash scoping and docker_network and
docker_connect prefixes. Enable public networks. Allow configuration
to define pre-existing networks by name.
Container inspection is caching data on first lookup. This will
result in incorrect data being returned on subsequent lookups if
a different `cid` value is provided. Also removed rescue of
the `Exception` class as this generally should never happen; rescue
of StandardError will be enough.
This commit updates the behavior of how the docker provider creates new
docker networks. It looks at each existing network to see if the
requested subnet has already been configured in the docker engine. If
so, Vagrant will use that network rather than creating a new one. This
includes networks not created by Vagrant. Vagrant will not clean up
these networks if created outside of Vagrant.
Since the virtualbox guest additions seem to only be available for
freeBSD, move the shared folder functionality over to freebsd guests
rather than all BSD guests.
This update was prompted by updates in openssh to the scp behavior
making source directory paths suffixed with `.` no longer valid
resulting in errors on upload. The upload implementation within
the ssh communicator has been updated to retain the existing
behavior.
Included in this update is modifications to the winrm communicator
so the upload functionality matches that of the ssh communicator
respecting the trailing `.` behavior on source paths. With the
communicators updated to properly handle the paths, the file
provisioner was also updated to simply apply previously defined
path update rules only.
Fixes#10675
Prior to this commit, the AddAuthentication hooks still existed in a
deprecated class LoginCommand. This commit fixes that by moving it over
to the vagrant cloud cli namespace instead.
Prior to this commit, the docker action would attempt to compare and
validate synced folders based on their string value, rather than their
actual path value. This commit updates that by path expanding the mounts
when comparing a containers synced folders.
This commit introduces some basic functionality for typed triggers:
- command
- action
Command triggers are triggers that will run before or after a given
sub-command.
Action triggers are for running triggers before or after internal
actions for Vagrant. This could be before or after a provision step,
before or after synced folders, or networking, etc.
On Windows the File::SEPARATOR ends up being `/` which causes
issues with the new way the path is being extracted from the
vbox information. When on Windows (even with WSL), automatically
convert the path to use common forward slash separator. This
fixes path modifications used for storing guest disks.
This fixes issues with box add/update when self hosting with client
certs. The --cert option was not being added to the curl subprocess
in these cases.
Prior to this commit, if a debian system requested an DHCP address using
systemd-network, Vagrant would ignore it and instead use the configured
IP from the virtualbox network action. This commit fixes that by instead
looking if DHCP was requested, and if so, use that option for an IP.
Prior to this commit, Vagrants output would only show that a file
provisioner was running, but had no detail as to what file was being
copied to where. This is especially confusing if a Vagrantfile has
multiple file provisioners. This commit updates that by showing the
source and destination of the file so that it's clearer what is being
copied and to where.
Adds a `reboot` option which allows the guest to be rebooted after
the completion of a shell provisioning task. The guest must support
the `:reboot` capability. Like the `reset` option, the `reboot`
option may be provided without defining `inline` or `file` options
when a reboot may be required between other provisioners.
Fixes#8639
Adds a new `rsync__rsync_ownership` option to rsync based synced folders
which will allow rsync to use the `--chown` flag if it is available. The
`rsync` and `rsync-auto` commands have a new `--rsync-chown` flag which
can be used to force the option on folders when running the commands.
Fixes#7329#7332
This PR adds support for running command passed via the --command
flag as elevated tasks. The option is only valid for commands and
not when setting up a remote session. Logic has also been adjusted
for when communicator restrictions are applied and test coverage
has been added.
Extra options are extracted from the machine configuration for the
network being configured to allow for customized network manager
behavior. The network entries must be filtered to remove non-network
entries (like port forwards) before accessing by index.
Fixes#9546
Rescue and re-wrap any errors encountered when running the post
rsync capability. Rescue this exception type and notify of error
when encountered by rsync auto. Include test coverage.
This commit adds a new rescue to the `publish` command for when the
client detects an invalid version prior to making the API request to
create that version.
Include config option within ssh_config command output template. Default
remote_user value to set username value. Include existence check on
provided config value to ensure file is available. Update tests to
include coverage on all changes. Add new options to the docs.
Also noticed whilst testing that if the `ProxyCommand` uses `%r`, then
it fails with `unable to find remote user`, so added support for
`config.ssh.remote_user` aswell
Both of these commands failed to default the options disabling
the provisioning from ignoring the sentinel file. This resulted
in different behavior than what was seen with the `up` and
`resume` commands which would only provision items with run set
to "always". This defaults the options to proper match the behavior
of `up` and `resume` to be consistent.
This also adds an extra `--no-start` flag to allow users to restore
a snapshot but not start the restored guest immediately.
Fixes#6752
When a large number of shares are defined it may cause the generated
command to exceed the maximum allowed length. To prevent this, only
allow 10 shares to be processed at a time.
Fixes#10483
Elevated commands can fail via winrm under certain conditions like
the machine name being changed. Detect this by checking for a known
exit code combined with known output included within stderr. If found,
attempt to re-execute the command using a machine prefixed username
if possible.
Prior to this commit, the package actions would create a temp dir in
the process of packaging and compressing a Vagrant box. This commit
ensures that the temp dir is removed once the command has completed so
that it doesn't leave around lots of temp directories.
Prior to this commit, if you specified a `version` for the salt provider
but no `install_type` Vagrant would fail to pass the proper parameters
to the bootstrap install script. This commit fixes that by adding some
validation to the salt provider if `version` is specified but not
`install_type`. It also adds some extra context for certain config
validation error messages so that the user knows what option was
incorrect, rather than the message just referring to the option as
*this*.
Prior to this commit, Vagrant would attempt to path expand a file that
didn't exist if it was left out of the passed in arguments and no
`--url` was used for external box uploading. This commit fixes that by
adding some additional validation for the passed in box file.
Prior to this commit, if you went to validate your Vagrantfile and
wanted to ignore the provider, Vagrant would still fail as it checks if
there are any registered providers that are installed and usable. This
commit mocks out all registered providers to bypass that for the
validate command so that Vagrant can just validate the config and ignore
any provider config blocks.
In some cases the E1000 NIC type is the only acceptable value. Since
defaulting causes breakages to existing boxes, leave the default value
as `nil` but check the VirtualBox version in use and print warning to
user if VirtualBox version is vulnerable and E1000 NIC types are
configured for use within defined network adapters.
The dry run import with VirtualBox 6 no longer provides full paths
for disks. Extract base path from suggested settings file location
and use that to generate full path name using provided disk base
name.
Be more restrictive when matching the port forward used for
SSH. Allow matching simply on the guest port, but also include
extra matching criteria for cases where multiple results may
be returned.
The default error action is to stop. When generating the initial
VM configuration during import, if the Compare-VM command fails
it results in a generic error message. Instead the error should
be ignored so the source VM can be inspected and a useful error
message can be returned to the user.
When running a shell provisioner elevated with winrm a scheduled
task is created to bypass permissions issues. If the name of the
computer has changed this may no longer work. To prevent errors
this PR updates the implementation to fetch the computer name
and prepends it to the username before creating the task.
This commit introduces a proper reboot cap for Windows guests. Once it
initiates a reboot on the guest, it calls out to the wait_for_reboot cap
to block on until the guest is finished rebooting.
Prior to this commit, if Windows was slow to reboot, Vagrant would fail
to find the right IP address to upload the wait_for_reboot script to.
This commit fixes this race condition by adding a timeout to ensure that
Vagrant can retry. It also properly catches an exception in the winrm
ready? method for checking if a guest is properly ready for
communications.
This commit adds a new flag to the `vagrant validate` command which
allows users to completely ignore the provider block of a config file.
This is useful for when you are running `vagrant validate` in CI and
don't want to install a valid provider to check the syntax of your
Vagratnfile. When the flag is invoked, a warning will be displayed
saying that the provider block will be ignored and not validated.
This commit is a workaround due to how older debian and ubuntu systems
fail to properly restart networking. Instead of relying on the init
scripts or ifup/down tools to restart each interface, this commit
instead restarts each interface individually
This commit adds some additional logic that falls back to using the
ifdown/ifup tools to restart networking. On Ubuntu 14.04, the init
script was designed to always fail to restart newtorking, so it needs
to use the ifdown/up tools instead. This commit will use the networking
init script as a last resort to restart networking, assuming other
commands haven't broken networking already.
https://bugs.launchpad.net/ubuntu/+source/ifupdown/+bug/1301015
If the type of error changes on retry the messages will effectively
spam the user display with alternating messages. Log each message
sent and only re-display each message once within 10 seconds.
Prior to this commit, Vagrant would prompt for smb username and password
every time, even if only smb_username was defined. This commit changes
that by allowing a "default" username from the Vagrantfile, with the
option of overriding it.
Prior to this commit, the hostname was set with one big bash script and
attempted to determine what tools are available. This commit changes
that by splitting out that tool check on the Vagrant side of things with
the GuestInspection class, and adds back restarting networking to get a
DHCP lease with the change rather than using `dhclient`. This pattern
matches how hostnames are set in the redhat capability.
This commit introduces a new uploader class for uploading files and
splits up some commonly used functionality between it and the downloader
class into a curl helper library.
Prior to this commit, when creating the ControlPath tmp dir for
socket path, Vagrant would simply rely on `rand(1000)` for making unique
dirs for rsyncing files which could result in collisions. This commit
updates that be properly using `Dir.mktmpdir` with a `vagrant-rsync-`
prefix.
This commit introduces a new option to the core trigger feature: `ruby`.
It can be defined to run ruby code when the trigger is configured to
fire. If you give the ruby block an env and machine argument, the
defined ruby code can use those variables internally.
Dynamically generate RDP information when applicable via provider
if supported. When no RDP port is provided ignore RDP in config
and omit from output.
This commit adds a new option `abort`, which when configured, will exit
the Vagrant process completely. If set to `true`, it will exit cleanly
with exit code 0. Otherwise, the exit code can be configured.
Update capability to use guest inspection module for determining
correct actions to execute. When systemd is in use restart the
correct active service, either NetworkManager or networkd. Default
to using the original service restart when systemd service is not
found.
Prior to this commit, the puppet provisioner would not properly set its
environment variables, if any were configured in the Vagrantfile. This
commit separates those properly with semicolons when calling out to
puppet apply.
The AutomaticCheckpointsEnabled option may not always be available
depending on the version in use. Check for support before applying
the configured value. If automatic checkpoints are to be enabled
and support is not available, force an error.
Since we are no longer extracting information based on key value due
to localization issues, use start and end locations to extract data.
This prevents errors when extra information is included like Scope.
Prior to this commit, providers like docker would fail to be brought up
because they do not store box objects like virtualbox or vmware
provider guests. This commit fixes that by making sure the box object
exists before writing the metadata file to disk.
This commit introduces the `--force` flag to the reload command. This
change means that if the flag is included, the halt step of the reload
will forcefully shutdown the virtual machine rather than a graceful
halt.
When performing a box update and the box version has been updated
to be different than the installed version, perform a lookup for
the latest available installed box to allow the update command to
continue successfully
Only move new exports file to destination without sudo when the
file has write access and the directory has write access. Always
use sudo when changing file ownership.
When a guest is created, the box metadata information is stored in the
machine data directory. This allows modifications to happen to the
Vagrantfile definition of the box in use (box name change, box version
change, etc) while still allowing the Machine instance of an active
guest successfully load the box currently backing it.
This occurs with a Linux host when a link-local address is configured
for vboxnet0 (which is the default for VirtualBox 5.2.6).
`connect': Invalid argument - connect(2) for "fe80::ffff:ffff:ffff:ffff" port 80 (Errno::EINVAL)
Adds the `docker_consistency` option, which sets the Docker volume
consistency level. This can be used to greatly improved synced folder
performance, especially on macOS.
See for details: moby/moby#31047
Service status check was failing because it was not being run as root. This resulted in vagrant thinking the service was not running, hence it would always try to start nfs rather than updating exports.
Prior to this commit, the run trigger option wouldn't catch for failures
outside of the #Subprocess.execute raising exceptions. This commit fixes
that by inspecting the exit code result of the subprocess and using the
new `exit_codes` option to determine how to move forward with the
trigger.
After installing msys2, there is another `cmd` on the path which prevents shared folders from being mounted. Explicitly calling `cmd.exe` fixes the issue
When the configured directory for FTP push has too many files, it will
generate an exception and fail due to a stack overflow. When this happens
just rescue out the exception and re-raise a custom error to provide
some context to the user on the actual problem.
Adds the final line of output to tests and properly simulates
failure. Fix with type casting to prevent calling methods on
nil and drop the final entry as it's not valid.
Fixes#9806
The base chef provisioner class will set the node name automatically
if not provided in the configuration. Since the chef apply provisioner
does not provide a node_name option, setting it will invalidate the
configuration. This checks for the node name before attempting to
use it.
Fixes#9901
- Fix#9796: Failing installation on bionic (18.04 LTS)
- Remove support for precise (12.04 LTS), which is out of public support
(On this version, the package python-software-properties contained
the add-apt-repository tool)
- Reduce the number of SSH calls, and avoid unnecessary apt-get calls.
Prior to this commit, when setting up private networks on Ubuntu using
netplan, it assumed that the guest was using systemd, the suggested
default tool to manage networking, and did not take into account devices
that could be managed with NetworkManager. This commit fixes that by
looking at the devices managed on the guest to see if its managed by
NetworkManager, and if so, use that renderer for netplan instead of
networkd.
This commit uses the internal helper SafeExec rather than using the
Kernel class directly. It also adds a test to ensure that if the `-t`
option is used, it will call SafeExec.
Keep track of selected Hyper-V switch using the ID instead of name
to prevent any encoding issues that may occur switching between
PowerShell and Ruby. With the IDs staying consistent, the switch
name can be fetched from the provided ID.
Fixes#9679#8794#9451
To allow users to continue to use the community plugin, this commit adds
a switch that will not load the core trigger feature if it detects the
community plugin is installed. It also allows the user to disable the
warning so they can continue to use the plugin without extra output.
If the guest address is unavailable via Hyper-V inspection, extract
MAC address of network adapters and check neighbor information for
any currently matching known address.
Due to the behavior of URI.parse on Ruby < 2.5 returning the string
value of the parsed URI object may modify the original URL. Specifically
it will remove the `//` prefix characters from SMB paths. When no
host modifications are being made, always return the original value.
Fixes#9636
This commit sets the structured_facts option to nil if its still
UNSET_VALUE at the finalize step. This was causing a bug when the facter
config was set but was not using structured facts.
