IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
Libvirt permits spaces in object names but rejects the ‘/’ character.
This change aligns our validator with libvirt’s behavior (and QEMU’s),
preventing names with '/' while still allowing human‑friendly
names with spaces.
Fixes: #740
We should not use guest.os.arch, it doesn't work correctly if the ISO is
i386 and the VM arch is x86_64.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
The function _find_treearch will be used for ISO files as well where we
don't want to have default arch.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Instead of using existing arch use badarch in order to test parts of the
code when we don't know the architecture.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
When running virt-install using waypipe the DISPLAY variable is not
defined and virt-install will complain that it cannot start virt-viewer.
Check for WAYLAND_DISPLAY as well, DISPLAY is defined only when xwayland
is used. In case of waypipe it configures only WAYLAND_DISPLAY.
Move the check before we check for virt-viewer as without display there
is no point to check if virt-viewer is installed or not.
Fixes: https://github.com/virt-manager/virt-manager/issues/884
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Using qemu-bridge-helper QEMU Session VMs are now able to use host
bridge interfaces. Currently only interface named virbr0 is allowed by
default but it is possible to change it in `/etc/qemu/bridge.conf`.
We will still keep the usermode network as default.
Resolves: https://github.com/virt-manager/virt-manager/issues/863
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
The old issue template stopped working as GitHub updated how the
template should look like. While at it create two issue templates, one
for bugs and the other for feature requests.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
`pre-commit` provides an easy way for devs to never miss committing
code without running black, and an easy way to to invoke black from
CI. In the future we can use it to run codespell, and new-to-us
tools like flake8
Signed-off-by: Cole Robinson <crobinso@redhat.com>
virtio-mem block size must be equal to or greater than the transparent
huge page size; otherwise, libvirt raises an error and a test will
fail. For example, on Asahi Linux, the transparent huge page size is 32
MiB, which is greater than 2 MiB, the specified virtio-mem block size.
On Linux 6.13, the configuration with the maximum transparent huge page
size is Arm64 with 64 KiB, and it has 512 MiB transparent huge pages.
Increase the block size to 512 MiB so that the test passes on every
Linux configuration.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Libvirt since v10.6.0 introduces pstore pci device for storing oops/panic
logs in nvram storage.
Let's add it into virt-install, It has 3 suboptions:
* backend: The desired backend, by far only 'acpi-erst' is accepted.
* path: Represents a path in the host that backs the pstore device in
the guest. It is optional, If not specified the libvirt will
auto generates one.
* size: Configures the size of the persistent storage available to the
guest. It is mandatory, unit is kilobytes.
Eg:
virt-install --pstore backend=acpi-erst,path=/tmp/guest_acpi_esrt,size=8
Signed-off-by: Lin Ma <lma@suse.de>
Add a pyproject.toml file with the configuration for the "black" tool,
to prepare reformatting the whole codebase with it:
- extend the line length from the default 88 characters to 100, more
suitable for the virt-manager codebase
- set a custom regexp for the files to format, so also the wrapper
scripts are handled
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
As all major distros support the riscv64 architecture, add it to the
list of prioritized architectures. This will move it up in the
architecture drop-down menu and thereby easier to find when creating
a new VM.
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Since libvirt v9.4.0, It introduces 'poll' settings in domain XML to
override the hypervisor-default interval of polling for iothread.
Let's add it into virt-install.
Eg:
virt-install \
...... \
--iothreads iothreads=2,\
iothreadids.iothread0.id=1,\
iothreadids.iothread1.id=2,\
iothreadids.iothread1.poll.max=123,\
iothreadids.iothread1.poll.grow=456,\
iothreadids.iothread1.poll.shrink=789
It results in the following domain XML snippet:
<iothreads>2</iothreads>
<iothreadids>
<iothread id='1'/>
<iothread id='2'>
<poll max='123' grow='456' shrink='789'/>
</iothread>
</iothreadids>
Signed-off-by: Lin Ma <lma@suse.de>
Libvirt supports setting dynamicMemslots attribute for virtio-mem since
v10.1.0, Let's add it into virt-install. Eg:
virt-install \
......
--vcpu 2 \
--cpu cell0.cpus=0,cell0.memory=4194304,\
cell1.cpus=1,cell1.memory=4194304 \
--memory maxMemory=65536,maxMemory.slots=8 \
--memdev model=virtio-mem,\
target.node=0,\
target.block=2048,\
target.size=8192,\
target.dynamicMemslots=yes \
......
It results in the following domain XML snippet:
<memory model='virtio-mem'>
<target dynamicMemslots='yes'>
......
</memory>
Signed-off-by: Lin Ma <lma@suse.de>
Using internal and external snapshot mode for the same VM has some
limitations. When creating new snapshot default to the same mode as
already existing currently used snapshot. If there is no existing
snapshot default to external snapshot.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
The 'po' subdirectory contains mostly the translation template, the
translations, and other metadata related to translations, and all of
them are not to be fixed directly.
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
Rather than passing the various bits for codespell when running it via
meson, create a configuration file for them. This way it will be
possible to run codespell manually getting the same results.
There is no change to the actual codespell configuration.
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
The commit c5a46646 asks libvirt to fill in a default panic model for us.
Now libvirt domcaps can advertise panic models, Let's fill in the panic
model combo with the advertised values.
Signed-off-by: Lin Ma <lma@suse.de>
The actual default CPU at the QEMU level is often a relatively
poor choice, which is stuck with just baseline functionality
and can sometimes not run modern guests at all.
Whenever possible, prefer maximum mode for a much nicer out of
the box experience.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
This mode has been introduced in libvirt 7.1.0 (March 2021) and
can be already used today with
--cpu mode=maximum
This is however slightly inconvenient to type and is not
consistent with the special treatment that the other modes
(host-passthrough, host-model) get.
Introduce a proper special mode for it.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Scientific Linux is no longer supported as it ended with version 7 so
change it to AlmaLinux that is mentioned to be used instead.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
With setup.py we cleared po/virt-manager.pot file and created a new
fresh content using gettext.
I missed that fact when converting to meson and with the --join-existing
argument the comments for each msgid were growing with references to
their location in source files.
There is no need to clear the file and generate it again because
removing the --join-existing will end up with the same result.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Tells virt-install to not use UEFI, if it would normally default
to it.
This likely isn't too useful in practice, since all occasions we
default to UEFI require it. But it future proofs opt out in case
we ever start defaulting to UEFI in cases where BIOS still works.
Resolves: https://github.com/virt-manager/virt-manager/issues/692
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The way `--boot uefi` is implemented allows a value to be passed,
but the value is completely ignored. So `--boot uefi=off` or
`--boot uefi=FOOBAR` is treated the same as `--boot uefi`
Lets fix this by making `--boot uefi` and `--boot uefi=on` the same,
`--boot uefi=off` will be implemented later, but any other value
throws an error. This is technically an CLI break but I don't
think it's anything to worry about
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Currently we will only apply the default `--cpu mode=host-passthrough`
config when _no_ `--cpu` config is passed. But this means if a user
configures ex. cpu `<topology>`, we don't set `host-passthrough` and
they get the libvirt/qemu default baseline CPU.
Instead, only skip the default config if the user manually
specified a `mode` or `model` value directly
https://issues.redhat.com/browse/RHEL-65371
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Input XML can have non-existent disk images as long as `--preserve`
is used, or those disks are skipped for any reason. We already
handle that correctly in some cases, but others we fail. This
covers another one.
Resolves: https://github.com/virt-manager/virt-manager/issues/563
Signed-off-by: Cole Robinson <crobinso@redhat.com>
With the libvirt testdriver, we can create+start a directory
pool that points to any arbitrary host path. But this doesn't
match the behavior of the real storage driver, where pool start
will fail if the source directory doesn't exist.
Let's mock this by raising an exception if we are inside the
test suite and we see a special string in the pool XML
Signed-off-by: Cole Robinson <crobinso@redhat.com>
`--replace` will attempt to remove any existing VM with the new
clone name before performing the clone operation. There's a report
that it will fail for an existing VM with nvram, since we need
aren't doing the equivalent of `virsh undefine --nvram`
https://issues.redhat.com/browse/RHEL-65105
Making `--replace` remove nvram is easy, but it kinda breaks some
Cloner assumptions about determining ahead of time whether we
need to clone nvram or not. It's fixable but it's messy.
Also, making changes here raises the question about how we should be
handling all the other `virsh undefine` flags that have appeared over
the years.
I think it's better to leave `--replace` as is, and just document
that it's doing the simplest equivalent of
`virsh destroy` + `virsh undefine`. Anything more complicated
than that will cause the clone operation to fail, the user needs
to handle VM removal themselves.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
--reflink only works with raw images, since copying anything else
will hit the qemu-img code path in libvirt storage driver.
This can pop up more nowadays since UEFI support is using qcow2 as
well.
Let's only attempt --reflink for raw disk images. It's basically
an optimization anyways
https://bugzilla.redhat.com/show_bug.cgi?id=2256285
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Default to adding tablet+keyboard if graphics and usb are configured.
This saves us from having to duplicate the arch list.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Our logic here is poorly duplicating libvirt's postparse logic.
Notably it will try to add bus=ps2 on non-x86, and misses obscure
cases like parallels.
https://issues.redhat.com/browse/RHEL-66768
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Was erroring with 'Dist currently only works with Git or Mercurial
repos'. Seems like something about the libvirt checkout ordering
caused problems
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The commit 6baa327d added active_pcr_banks support, but put it under the child
element <tpm>, which is wrong, It should be under sub child element <backend>.
Before:
--tpm model=tpm-tis,backend.type=emulator,backend.version=2.0,\
active_pcr_banks.sha1=on,\
active_pcr_banks.sha256=yes,\
active_pcr_banks.sha384=yes,\
active_pcr_banks.sha512=yes
It results in the following domain xml:
<tpm model='tpm-tis'>
<backend type='emulator' version='2.0'/>
<alias name='tpm0'/>
</tpm>
After:
--tpm model=tpm-tis,backend.type=emulator,backend.version=2.0,\
backend.active_pcr_banks.sha1=on,\
backend.active_pcr_banks.sha256=yes,\
backend.active_pcr_banks.sha384=yes,\
backend.active_pcr_banks.sha512=yes
It results in the following domain xml:
<tpm model='tpm-tis'>
<backend type='emulator' version='2.0'>
<active_pcr_banks>
<sha1/>
<sha256/>
<sha384/>
<sha512/>
</active_pcr_banks>
</backend>
<alias name='tpm0'/>
</tpm>
Signed-off-by: Lin Ma <lma@suse.de>
E.g.
virt-install \
... \
--tpm model=tpm-tis,backend.type=emulator,backend.version=2.0,\
backend.source.type=dir,backend.source.path=/some/dir
It results in the following domain xml:
<backend type="emulator" version="2.0">
<source type="dir" path="/some/dir"/>
</backend>
Signed-off-by: Lin Ma <lma@suse.de>
E.g.
virt-install \
... \
--tpm model=tpm-tis,backend.type=emulator,backend.version=2.0,backend.debug=3
It results in the following domain xml:
<tpm model="tpm-tis">
<backend type="emulator" version="2.0" debug="3"/>
</tpm>
Signed-off-by: Lin Ma <lma@suse.de>
E.g.
virt-install \
... \
--sound model=virtio,streams=4
It results in the following domain xml:
<sound model="virtio" streams="4"/>
Signed-off-by: Lin Ma <lma@suse.de>
E.g.
virt-install \
... \
--sound model=usb,multichannel=yes
It results in the following domain xml:
<sound model="usb" multichannel="yes"/>
Signed-off-by: Lin Ma <lma@suse.de>
E.g.
virt-install \
... \
--disk /tmp/disk0.qcow2,size=16,driver.type=qcow2,blockio.discard_granularity=4096
It results in the following domain xml:
<disk type='file' device='disk'>
<driver name='qemu' type='qcow2' discard='unmap'/>
<source file='/tmp/disk0.qcow2'/>
<target dev='vda' bus='virtio'/>
<blockio discard_granularity="4096"/>
</disk>
Signed-off-by: Lin Ma <lma@suse.de>
Running `meson dist` requires working git but that doesn't seem to be the
default in github actions when running jobs inside container. Any git
invocation fails with the following error:
fatal: detected dubious ownership in repository at '/__w/virt-manager/virt-manager'
To add an exception for this directory, call:
git config --global --add safe.directory /__w/virt-manager/virt-manager
So that's what this patch does to make `meson dist` work correctly.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
We will no longer have `rpm` target with meson. That should not be an
issue as users can run `meson dist -C build` and
`rpmbuild -tb build/meson-dist/virt-manager-{version}.tar.xz`.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
The reason we fork by default, is to force ssh to invoke
ssh-askpass when a password is required, rather than prompt on
a terminal no one is looking at. There's a more thorough
explanation here:
https://github.com/virt-manager/virt-manager/issues/731
With SSH_ASKPASS_REQUIRE=force, we now have a way to force ssh
to use askpass in the above scenario, when ssh and libvirt are new
enough.
The default forking behavior has caused maintenance pain in the
past, and is currently causing issues on macos:
https://github.com/virt-manager/virt-manager/issues/620
Let's flip the default to `--no-fork`. The VIRT_MANAGER_DEFAULT_FORK
env variable is there as an escape hatch incase I really miscalculated.
I don't expect many people are depending on use of askpass either
way, or if they are, they are launching virt-manager from their
desktop and not a terminal, which already gives us the correct
behavior AFAICT>
My suspicion is barely anyone will notice, which is why
I'm ok with changing this now, despite the libvirt support being
brand new.
If this doesn't raise any issues, then we can eventually drop
the forking behavior all together.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
openssh 8.4p1 released in Sep 2020 finally added a feature
to force using SSH_ASKPASS instead of prompting on the commandline
for password, if a password would be required.
https://man.openbsd.org/ssh.1#SSH_ASKPASS_REQUIRE
Getting this behavior is basically what our whole fork dance is
about. Now we can do it with an environment variable
Let the user override it from the environment though, so there's
an escape hatch incase this causes unforseen problems
Signed-off-by: Cole Robinson <crobinso@redhat.com>
- drop all the architectures (and thus building only on x86_64):
virt-manager has no architecture-specific installation bits, and thus
it builds in the same way on every architecture; hence no need to
explicitly test on various architectures
- test on all the supported Fedora versions (Rawhide included): this way
it is possible to check that older versions are still supported, at
least when building
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
- use a single translatable message, instead of splitting the text in
two; it helps translators to have the full context/text
- improve the language a bit
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
Drop our hardcoded model lists, and just ask libvirt to fill in
a model for us. Add an entry to the combo box so users can type
in a non-default value if they want one.
Long term libvirt should be providing all this info to us via
domcapabilities
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Currently `--panic default` for aarch64 doesn't even request
a `<panic/>` device due to quirky xmlbuilder behavior. Fix that,
but more generally just leave model empty and let libvirt fill it
in for us.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This is a special convenience option for filling in `type=hostdev`
config using the same format of lookup string that can be
passed to `--hostdev`
Fixes: https://github.com/virt-manager/virt-manager/issues/500
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This just covers the common PCI case with these new options
source.address.type=
source.address.domain=
source.address.bus=
source.address.slot=
source.address.function=
Signed-off-by: Cole Robinson <crobinso@redhat.com>
I think this is a better default than the current `Only when
Fullscreen`, which is the same as `Never` for VM windowed mode.
I wrote my reasoning here:
https://github.com/virt-manager/virt-manager/issues/747
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Fix no longer valid URL fragments, wrap at 80th column, remove dots that
follow only a few links.
Signed-off-by: Grzegorz Szymaszek <gszymaszek@short.pl>
This is essentially what it has always behaved as, but making it
explicit in the code will now trigger the extra warnings when
detect=on is also used.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Using `detect=on,name=OSNAME` is good for CI safety, incase
a new distro tree has a regression with detection, or if testing
against a new distro that osinfo-db doesn't know about.
But virt-install should still try to inform the user that detection
failed, and suggest filing a bug if the user expected it to work.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The require= behavior should be AUTO for this case, but the
way we were previously initializing variables made this OFF.
I don't think this was intentional. We should have changed
this when we started defaulting to `detect=on`
Signed-off-by: Cole Robinson <crobinso@redhat.com>
It's kinda hard to follow the logic here. cli require= can only be
on|off or unset, but we use an internal 'auto' value which seems
like it can come from the cli.
Try to clean it up with enum like values.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Latest upstream release was back in 2012 and the new
libayatana-appindicator project is present in all distribution supported
by libvirt.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
has_install_phase is an ambiguous name for its intended purpose.
Really this is so API users have a way of knowing if the VM `install`
process requires 2 XML phases or not. Make that naming explicit
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This basically reverts this commit from virt-manager 4.0.0
```
commit 825ec644b859b99d6f9f155b04ef3ac07eca1fc2
Author: Cole Robinson <crobinso@redhat.com>
Date: Tue Mar 1 10:32:01 2022 -0500
installer: Do not force reboot with --cloud-init
Resolves: https://github.com/virt-manager/virt-manager/issues/189
```
After that commit, using `--cloud-init` or `--unattended` would
basically imply `--noreboot`
For some usage of `--cloud-init`, like in the initial report, this
makes sense to do. virt-install will drop you into the cloud image,
you mess without, run `poweroff`, and are surprised when virt-install
reboots the VM. I agreed, and made the change
But changing this was a bad idea.
cloud-init and unattended VMs have 2 XML phases. First XML is booting
off temporary media (the generated cloudinit iso), second XML is the
permanent boot config (booting off disk). We need to force the VM
to fully poweroff, so that the second XML config takes effect. We
make that happen with `<on_reboot>destroy</on_reboot>`, which tells
libvirt/qemu to poweroff the VM when it receives a guest reboot
notification. virt-install then defaults to restarting the VM when
it detects shutdown, to manually replicate the VM requested reboot,
but still get the second XML config to take effect.
The perennial problem with this is that, from outside the VM, we
don't know if the user inside the VM requested `poweroff` or `reboot`.
virt-install emulates the reboot behavior, and provides `--noreboot`
to override it. It's still confusing if a user puts a `poweroff`
command at the end of an anaconda kickstart file, and sees the VM
reboot, but it's been that way forever.
Except with that commit above, we flipped it for --cloud-init and
--unattended. Now, users who do `poweroff` are getting expected
behavior, but users who request `reboot` are not.
We could go further and add a `--yesreboot` option or similar. But
for consistency sake I think we should just revert that behavior,
and tell users to use --noreboot like usual.
Resolves: https://github.com/virt-manager/virt-manager/issues/497
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Trying to size the window based on VM desktop resolution does
not do the correct thing when host fractional scaling is enabled
and using spice.
The best we can do here is ask the viewer widget for its
preferred size
Signed-off-by: Cole Robinson <crobinso@redhat.com>
_adjust_viewer_size has one remaining function: when scaling and
resizeguest is disabled, do
viewer.set_size_request(*desktop_resolution), so the viewer fills
out the scrolled window and scrollbars show up.
After playing with this all a bunch, I discovered two things
* spice already behaves the way we want, so we don't need to
manually intervene. so all that is unnecessary
* vnc has an option to behave the way we want, via the
set_force_size knob
So instead lets rip it all out, and implement the VNC piece by
toggling set_force_size based on whether scaling+resizeguest is
enabled.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
* Move shared init to `_set_display`
* Move per class init to `_init_display`
* Move per class signal setup to `_connect_display_signals`
* Make base class call cleanup of child class via `_close`
* Drop unnecessary spice `_display_channel` handle
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This wires up the guest.convert_to_vnc function to command line,
and documents it.
There's one suboption `qemu-vdagent=on|off`, defaulting to `off`
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This is the beginnings of support for a `virt-xml --convert-to-vnc`
option. Take an existing VM, strip out most of the previous graphics
config, and add VNC graphics.
We try to convert over some of the shared graphic bits, like listen
and port settings, if they were previously specified.
If spice GL was enabled, we convert to egl-headless config
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Wire up guest.convert_to_q35 to the CLI. We add one suboptions,
`num_pcie_root_ports=X`, matching the pre-existing
`--controller num_pcie_root_ports=X` option
Nothing fancy here. See man page docs for details
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This function converts a PC/i400FX XML config to Q35. Mostly
this is deleting any plain PCI addresses and controllers, adding
pcie-root-ports, convert IDE -> SATA, and let libvirt fill in
the rest.
This the implementation piece. CLI additions come later
Signed-off-by: Cole Robinson <crobinso@redhat.com>
this doesn't change any behavior, but it makes things cleaner
by not overloading the size-allocate signal
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This does not seem to be required anymore. We get notified on
viewer size-allocate already, which seems like enough.
Rename the function to _adjust_viewer_size which is more accurate
to what it does, and adjust the callers
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This mattered when we were doing widget adjustments to maintain
scaling ratio. But nowadays it seems totally redundant: we are
basically trying to set an allocation that already happened via
the normal chain of widget events, so it's a no-op.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Checking for resolution = None is a roundabout way to check if
the console was already opened. Make that more clear
Signed-off-by: Cole Robinson <crobinso@redhat.com>
+ Don't check for redirdev devices, let the spice widget tell
us if things aren't configured correctly.
+ Remove some duplication
Signed-off-by: Cole Robinson <crobinso@redhat.com>
We can not reliably use VM desktop resolution for host widget
sizes when the host is using desktop scaling, so don't even try.
This fixes the black bars around the VM console when scaling=never
and autoresize is disabled
Signed-off-by: Cole Robinson <crobinso@redhat.com>
We can make `--xml` fit the common xml cli option paradigm, which
less us drop a whole bunch of special handling in virt-xml
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Adjust cli.py `run_parser` and similar functions to take the
parservalue directly, rather than passing in the whole `options`
structure.
This makes it easier to reason about what the virt-xml action
functions are actually working with.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This adds an Action class which is a container for (most) of the
data we need to perform a virt-xml operation. Basically an instance
of the class represents a pairing like
--edit 1 --disk path=/foo or
--add-device --network wibble,model=virtio
etc
This is a functional no-op, but it moves the code closer to
an architecture that will allow us to perform multiple actions
with a single virt-xml command.
While doing this, we centralize most of the cli validation that is
scattered around, since it's now easier to do it in one place.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Move the validation functions out from amongst the functional XML
editing code, and into their own section. We will be adding more
code alongside in future patches.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Spice has maintained aspect ratio for us forever.
GtkVnc was added in previous commit, provided since May 2021.
If user is running on an old GtkVnc, they just get potentially
stretched out viewing when scaling is enabled, so the fallback
scenario is no big deal IMO
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Right now you can select the option for an offline VM, but once
it starts up, and no spice agent is detected, the option is
set insensitive. This is a pain.
Keep the tooltip, but keep the option always selectable
Signed-off-by: Cole Robinson <crobinso@redhat.com>
We can use this signal to be notified when VM resolution changes.
Log when we actually detect a resolution change
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Current code will take a screenshot when the 'New Snapshot'
dialog is initially launched, and then use that cached content
as the screenshot for the snapshot... which may be totally out
of date by the time the user enters all the snapshot info.
Instead, drop the snapshot preview UI in the 'New Snapshot' wizard
entirely, and use a screenshot that is fetched after the user actually
clicks 'Finish'
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This is from when we used to support directly connecting to
serial console /dev/pty devices. Nowadays all serial connections
are mediated through libvirt stream APIs and we don't need to
care about SIGHUP
Signed-off-by: Cole Robinson <crobinso@redhat.com>
If the VM has implicit TPM state, use the VIR_DOMAIN_UNDEFINE_KEEP_TPM
flag to preserve that state when renaming the VM (if libvirt is new
enough).
The state is stored based on VM UUID and nothing else, and the UUID
is preserved during rename, so we don't need to do any of the same
trickery that's required for nvram duplication.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Fix console config validation for an edge case where both libvirt and
Spice/VNC TCP connection are on localhost. Transport over TCP does not
necessarily mean that libvirt connection is to a remote host.
Compare libvirt and graphics device addresses to localhost individually.
Raise an error only when guest device is bound to localhost but libvirt
connection is non-local (remote).
Validation that prevents fully local TCP seems to go back all the way to
bc13c302de4 ("console: Warn if qemu+tcp URI and listen == 127.0.0.1").
Signed-off-by: Sami Loone <sloone@forcepoint.com>
* add_q35_pcie_controllers already skips adding controllers if
any type=pci already exist, so delete the extra checking
* be more paranoid and only run the live edits when the machine
type actually changed from an expected config
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The commit db8305ad explicitly adds pcie-root and pcie root ports for q35.
If the user selects i440FX chipset instead of Q35(default) from customize
dialog, The pre-defined pcie-root and ports cause failure when starting
i440fx VM installation because they're not applicable to i440fx.
It fails with below error message:
summary=Unable to complete install: 'XML error: The PCI controller with index='0'
must be model='pci-root' for this machine type, but model='pcie-root' was found instead'
details=Traceback (most recent call last):
File "/usr/share/virt-manager/virtManager/asyncjob.py", line 72, in cb_wrapper
callback(asyncjob, *args, **kwargs)
File "/usr/share/virt-manager/virtManager/createvm.py", line 2088, in _do_async_install
installer.start_install(guest, meter=meter)
File "/usr/share/virt-manager/virtinst/install/installer.py", line 737, in start_install
domain = self._create_guest(
File "/usr/share/virt-manager/virtinst/install/installer.py", line 679, in _create_guest
domain = self.conn.createXML(initial_xml or final_xml, 0)
File "/usr/lib64/python3.10/site-packages/libvirt.py", line 4442, in createXML
raise libvirtError('virDomainCreateXML() failed')
libvirt.libvirtError: XML error: The PCI controller with index='0' must be
model='pci-root' for this machine type, but model='pcie-root' was found instead
This patch fixes it by removing the pcie-root and ports for i440fx in
apply_overview.
Resolves: https://github.com/virt-manager/virt-manager/issues/444
Signed-off-by: Lin Ma <lma@suse.com>
Inspired by the x86_64 counterpart.
We're using TCG in this case, so things like host-passthrough
are not available.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
The old capabilities are extremely outdated. The new ones were
captured on an IBM POWER8 machine running Fedora 40.
A few new features are advertised, and the details of the
machine are significantly different, but not much of this is
reflected in the output XML files.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
The old capabilities are extremely outdated. The new ones were
captured on an Ampere Mt. Jade machine running Fedora 40.
Notable differences that are reflected in the output XML files
include the availability of SPICE, as well as EFI firmware and
ACPI support being advertised.
The test script had to be updated too, since both virtiofs and
memfd are now available.
Closes: #714
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
internal snapshots largely only work if the VM is backed by a single
qcow2 disk. But newly supported external snapshots support other
disk formats, and multiple disks.
Drop the old validation checking. Let's just leave it up to libvirt
to give us an error at snapshot creation time
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Libvirt enables blob resources for the virtio video device since 9.2.0.
It accelerates the display path due to less or no copying of pixel data.
E.g.
virt-install \
... \
--video model.type=virtio,blob=on
It results in the following domain xml:
<video>
<model type="virtio" blob="on"/>
</video>
Reviewed-by: Cole Robinson <crobinso@redhat.com>
Signed-off-by: Lin Ma <lma@suse.de>
Set kvm pv-ipi feature by --features argument.
E.g. virt-install --features kvm.pv-ipi.state=off
It results in the following domain xml:
<features>
<kvm>
<pv-ipi state='off'/>
</kvm>
</features>
Reviewed-by: Cole Robinson <crobinso@redhat.com>
Signed-off-by: Lin Ma <lma@suse.de>
This has ambiguous volume labels that will match multiple
osinfo output:
$ osinfo-detect -a tests/data/fakemedia/fake-win-multi.iso
Media is bootable.
Generated by editing `fake-win7.iso` already in tree
$ xorriso -indev fake-win7.iso -outdev test.iso \
-boot_image isolinux keep \
-volid SSS_X64CHK_ -volset_id SSS_X64CHK_
Add a simple test case that confirms _some_ os was detected,
and virt-install doesn't totally choke on it
Signed-off-by: Cole Robinson <crobinso@redhat.com>
+ Drop test_livetests bit which was working around the old behavior
+ Add test_details piece that covers the new behavior, and
ideal behavior that we maintain the selection across VM restarts.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Currently, View->Consoles disables the menu items for serial devices
when the VM is offline. This changes that behavior. This is useful,
since it allows you to pre-select serial console before starting the
VM, which can help ensure you don't miss any serial boot output.
It also makes the UI interaction more intuitive.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Reproducer:
+ Start VM with 1 graphical and 1 serial console
+ Select serial console
+ Power off VM
+ Open console menu, select 'Graphical' option
+ See errors
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Reproducer:
+ Enable libguestfs inspection
+ Open app and VM details 'Overview' window
+ After some time, the `Apply` button is mysteriously activated
+ Trying to switch away from the page will give 'Unapplied errors'
warning box
When libguestfs inspection completes, it triggers a signal which
will refresh the OS info details.py page. Which is fine, but we
should be limiting it to only refreshing the page if its the currently
visible one. Otherwise the `Apply` button can be activated, which
messages up app navigation
Signed-off-by: Cole Robinson <crobinso@redhat.com>
* Remove .gitpublish, we only use github for patches now
* Drop author* fields entirely from setuptools, that seems fine.
* Drop other references
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Tweak the test case added in ba3a098c3b82d0dcf36198e67ae3bcd6ad7c6d4a
to test XML generation instead, to make sure nothing regresses
weirdly there.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
We will not enable hyperv_reset feature as modern Hyper-V versions don't
export it.
We also don't enable hyperv_reenlightenment as it is mostly relevant
when migrating VMs that are running Hyper-V inside and requires other
bits to work correctly.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
We will no longer enable hyperv features based on libvirt and QEMU
versions.
Some tests don't use the kvm-x86_64-domcaps-latest so the code will now
not enable Hyper-V features. There are still other test cases that cover
Hyper-V features so instead of mangling with URI for these tests update
test data.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Examples:
`virt-xml MYVM --add-device --sound FOO1 --sound FOO2` is not rejected
and will add 2 devices.
`virt-xml MYVM --remove-device --sound opt1=FOO1 --sound opt2=FOO2`
is not rejected, only the final `--sound` option is used.
`virt-xml MYVM --edit --sound FOO1 --sound address.type=BAR` is not
rejected. The commandlines will effectively be squashed
`--sound` can be any XML option, but trying to mix and match
options like `--sound` and `--disk` is always rejected.
These bits are unexpected and undocument, but have been around since
the beginning of virt-xml.
I can see us breaking compat on these in the future (particularly the
--remove-device bit), but that would need to be an explicit decision,
and announced appropriately. In the mean time, we should be testing
this so it doesn't accidentally regress.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This reverts b34ae0d0c883972d74deef6cdadb29bdd1bbe487 and
1fef5d8661118a4f00a8689b323d275144b39dbe.
Playing with this some more I found some deeper problems. For example
```console
[:~/src/virt-manager] (main) $ ./virt-xml test-for-virtxml --connect test://`pwd`/tests/data/testdriver/testsuite.xml --print-diff --edit 1 --video model=FOO --edit 2 --sound model=BAR
--- Original XML
+++ Altered XML
@@ -180,14 +180,14 @@
<graphics type="vnc" port="-1" autoport="yes">
<listen type="address"/>
</graphics>
- <sound model="sb16"/>
+ <sound model="BAR"/>
<sound model="es1370"/>
<sound model="ich6"/>
<video>
<model type="vmvga" vram="16384" heads="1" primary="yes"/>
</video>
<video>
- <model type="cirrus" vram="16384" heads="3"/>
+ <model type="FOO" vram="16384" heads="3"/>
</video>
<hostdev mode="subsystem" type="usb" managed="yes">
<source>
```
There's other weirdness too, though it's mostly strange plays on previous weird behavior
* `--edit --video model=vga --edit --sound model=ich9 --video model=qxl` works
* `--add-device --sound model=foo --video model=bar` was previously rejected but now works
* `--remove-device --video model=vmvga --sound model=sb16` was previously rejected by now works
Fixing all this is not trivial. So I think we need to revert and go back to the drawing board.
This makes it easier for users to change video model from QXL to
different model as QXL is special and uses other attributes that are not
supported by any other model. Without this code libvirt would report
error or user would have to request `clearxml`.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
When Spice graphics is used QEMU creates a Spice server and communicates
with Spice client using multiple channels. These channels are used by
the spice devices as well. Without the Spice graphics defined there is
no use for the other devices. In addition libvirt will report error for
such configuration.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
This will test spice related changes. Follow up patches will change this
behavior to remove spice related devices.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
The libvirt test driver supports more APIs now, so we are getting
a different error message.
Extend the grep= string to work with both old and new libvirt
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This will let us match more complex output, and match error message
differences across libvirt versions
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Fix the cloud-init config file for --clouduser-ssh key.
Also change ssh_authorized_keys to `ssh_authorized_keys`, and make sure
default user is not deleted when setting ssh key for root.
Fixes commit 22478f307dd9 (virt-install: Add --cloud-init clouduser-ssh-key=)
Resolves: https://github.com/virt-manager/virt-manager/issues/452
Signed-off-by: Natanael Copa <ncopa@alpinelinux.org>
When there's a pool with many vols (files), and user clicks on "Browse" in
VM's Virtual Disk details, the UI could stuck for up to several minutes.
By using DeviceDisk.paths_in_use_by() for bulk getting names, it could
show browser much faster.
conn.fetch_all_vols() is slow, taking ~0.03s every time it's used. When
handling with a large number of paths, its performance is too bad and
could stuck UI for a very long time.
DeviceDisk.paths_in_use_by is added to help handle many paths with
only one call to conn.fetch_all_vols().
Add adaptations for loongarch with the following features:
Default video Support
UEFI prefer
Usb tablet and usb keyboard
rng and memballoon
sound device
Usb controller
Signed-off-by: Xianglai Li <lixianglai@loongson.cn>
While the shorthand saves a tiny bit of typing, it is confusing
to people coming from QEMU, where -cpu host is the equivalent
of our --cpu host-passthrough. It's better to stick with the
unambiguous names used by libvirt.
Besides, the GUI already uses "host-model" throughout, so
advocating its use in the CLI too increases the internal
consistency.
https://issues.redhat.com/browse/RHEL-40003
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
The test driver was broken for some time and our mdev tests were failing
on most libvirt versions. Recently this was fixed in libvirt 10.4.0 so
skip these tests unless we have new enough libvirt.
This requires extracting some mdev devices from many-devices test case.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
The only thing that GtkSourceView gives us is syntax highlighting and
auto-indent. When this library is not available, we can still offer xml
editing with a plain textview with very little lost functionality.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com>
This won't make any difference on x86, where either SATA or IDE
is used depending on the machine type, but on other architectures
we want virtio-scsi for a reasonable chance that the guest OS
will recognize the disk.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Most existing disk images expect UEFI; those that don't, require
either using direct kernel boot or manually providing a firmware
image, both of which imply user intervention.
Using UEFI by default means that at least images belonging to
the first group, of which more are going to pop up as the
ecosystem matures further, can work out of the box.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Cover additional scenarios: direct kernel boot, use of cloud-init,
installation from CDROM and unattended installation.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
The new capabilities reflect the status of riscv64
virtualization as of Fedora 40, which comes with libvirt
10.1.0 and QEMU 8.2.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Bring back the `Open` label for `Browse Local` option, but
preserver the SAVE dialog behavior (can enter a manual path)
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Use GtkFileChooserNative [1] instead of GtkFileChooserDialog [2]
to integrate better with the platform (e.g. use the portal
implementation when run with GTK_USE_PORTAL=1, resulting in
the KDE Frameworks implementation being used when
xdg-desktop-portal-kde is in use). Quoting from
the GtkFileChooserDialog doc [2]:
> If you want to integrate well with the platform you should use the
> GtkFileChooserNative API, which will use a platform-specific dialog if
> available and fall back to GtkFileChooserDialog otherwise.
Also replace the use of GTK_STOCK_CANCEL [3] and GTK_STOCK_OPEN [4]
which were deprecated in GTK 3.10:
Both, the `accept_label` and `cancel_label` params of
`Gtk.FileChooserNative.new` can be `None` to use the default
text ("Open", "Cancel"). [5]
Adjust the only caller (in `vmmVMWindow#_takeScreenshot`)
that was passing an explicit label/icon name for the
accept button to pass `_("_Save")` as label, rather than
the also deprecated Gtk.STOCK_SAVE [6].
(GtkFileChooserDialog has special handling for Gtk.STOCK_SAVE
etc., but that's not generally the case for native dialogs).
Rename the method param from `choose_button` to `choose_label`
to make clearer that this is a label.
[1] https://docs.gtk.org/gtk3/class.FileChooserNative.html
[2] https://docs.gtk.org/gtk3/class.FileChooserDialog.html
[3] https://docs.gtk.org/gtk3/const.STOCK_CANCEL.html
[4] https://docs.gtk.org/gtk3/const.STOCK_OPEN.html
[5] http://pygobject-doc.gitee.io/pgi-docs/Gtk-3.0/classes/FileChooserNative.html#Gtk.FileChooserNative.new
[6] https://docs.gtk.org/gtk3/const.STOCK_SAVE.htmlFixes#315
Setuptools 62.4.0 adds setuptools.command.build to transparently replace
the old distutils equivalent; hence, use it when available to fully
switch away from distutils.
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
This makes sure that the actions/checkout action uses git to fetch the
sources (and thus resulting in a git repository), rather than using the
REST API (and get a simple directory with sources).
This may be useful to certain services (e.g. newer version of
codecov-cli) as they need to have the sources as proper git repository.
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
Split the installation of the base dependencies & the RPM build
dependencies in two steps, so they can be reordered as needed later on.
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
Replace Gtk.Box.add() with Gtk.Box.pack_start() to place the VTE terminal widget. This enables horizontal expansion/filling the terminal widget when resizing the serial console window.
This event is mainly to refresh the VM XML to figure out the state of
virtio channels and that should happen only for running VMs.
This affects external snapshot behavior in virtManager. If the VM is
offline and user deletes external snapshot libvirt needs to start QEMU
process to delete the snapshot. The QEMU process is started with stopped
CPUs so it is visible as online and paused. It also emits the agent
lifecycle event but there is no other domain lifecycle event so after
the snapshot deletion the VM stays paused in virtManager even if it is
already offline in libvirt.
To mitigate this behavior we can ignore agent lifecycle event for
shutoff VMs.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Add new 2 rows into the snapshotsnew windows, one to select if the
snapshot mode is `internal` or `external` and memory state location when
making external snapshot for running VM.
We will check if libvirt fully supports external snapshots by consulting
host capabilities and add `external` to snapshot mode only if it is
supported.
For external snapshots the memory state is stored in separate file but
libvirt doesn't have any default location so virtManager will get path
to all usable disks and create a dropdown menu for it, but user is still
allowed to modify that path to use whatever path they prefer.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
- Add UI coverage
- Drop redundant systray_instance caching
- Tweaks help test and docs
- Show an error if the systray doesn't embed
Signed-off-by: Cole Robinson <crobinso@redhat.com>
- Rework the ConsolePages abstraction to carry the Gtk.Menu. makes
it all less messy
- Make the console menu the single source of truth for console
embeddability, and error message reporting
- Small misc cleanups here and there
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Currently, when there are multiple graphics devices, the first graphics
device is always selected as the graphical console. For the following
configuration:
<graphics type="egl-headless">
<gl rendernode="/dev/dri/renderD128"/>
</graphics>
<graphics type="spice" autoport="yes">
<listen type="address"/>
<image compression="off"/>
<gl enable="no"/>
</graphics>
virt-manager will report an error:
Cannot display graphical console type 'egl-headless'
The patch will iterate through all graphics devices to try to select the
first embeddable graphics device as graphical console.
Signed-off-by: Feng Jiang <jiangfeng@kylinos.cn>
Signed-off-by: Liming Sun <sunliming@kylinos.cn>
Reviewed-by: Cole Robinson <crobinso@redhat.com>
Now open_graphics_fd() always opens the 1st graphic device, when
there are multiple graphic devices (e.g. spice + egl-headless),
we need to specify the index number for libvirt.
Signed-off-by: Feng Jiang <jiangfeng@kylinos.cn>
Signed-off-by: Ming Xie <xieming@kylinos.cn>
Reviewed-by: Cole Robinson <crobinso@redhat.com>
Not quite sure what's going on here, I think it has something to
do with slowness when the accessibility bus is being polled.
Add some infrastructure to increase the timeout while we wait for
the app to pop up
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Newer libvirt + edk2 will default to nvram in qcow2 format, but
our domain rename code had some .fd assumptions baked in.
Adjust uitests to handle it too
Signed-off-by: Cole Robinson <crobinso@redhat.com>
I have no idea if the generated config is optimal, but this
at least provides a base for us to confirm when defaults change.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The libvirt test driver now has implementations for hotplug routines,
which broke string matching for one case.
Loosen it up to work for old and new libvirt versions.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
With libvirt 9.8.0, the test driver now has a stub getDomainCapabilities
implementation. But we still have some code that needs to handle
a driver with missing domcaps.
Make our magicuri mock return NO_SUPPORT for domcaps, when the URI
doesn't have any domcaps XML passed in. This is enough for our test
purposes.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
When specifying the socket.source option for filesystem devices, like
this:
--filesystem type=mount,driver.type=virtiofs,source.socket=/xyz.sock,target.dir=tag1
virt-install is writing the xml as:
<filesystem type="mount">
<source>
<socket>/xyz.sock</socket>
</source>
<target dir="tag1"/>
<driver type="virtiofs"/>
</filesystem>
This produces an error such as:
ERROR missing source information for device mount_tag1
But the socket should be an attribute of source rather than a child
element. After this patch, the same command results in the following XML
and no error is produced:
<filesystem type="mount">
<source socket="/xyz.sock"/>
<target dir="tag1"/>
<driver type="virtiofs"/>
</filesystem>
Resolves: RHEL-1126
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com>
Nowadays with libvirt split daemons, libvirtd isn't required to
be installed for a first run local connection to succeed, so we
are needlessly blocking the app from 'just working' in many cases.
Especially considering that many distros often have libvirt running
out of the box due to gnome-boxes pulling it in.
Drop the daemon checking entirely.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Enable dependabot for the "main" branch, letting it scan for outdated
GitHub Actions used in workflows on a weekly base.
Signed-off-by: Pino Toscano <ptoscano@redhat.com>
When shim in the guest sees unpopulated EFI NVRAM, like when
we create a new UEFI VM, it invokes fallback.efi to populate
initial NVRAM boot entries. When the guest also has a TPM device,
shim will do a one time VM reset. This reset throws off the
reboot detection that is central to virt-install's install
process.
The main install case that this will usually be relevant is
the combo of UEFI and --cloud-init. The latter usually implies
use of a distro cloud image, which will be using shim, and the
--cloud-init process requires a multi stage install compared
to just a plain import install.
For that case, we disable the default TPM device for the first
boot.
https://bugzilla.redhat.com/show_bug.cgi?id=2133525
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Commit f107e3998908 ("Switch to more traditional logging structure",
2019-06-17) replaced "logging.exception" with "log.exception", effectively
shifting the argument lists 4 characters to the left. The second and
further lines of multiline invocations were not accordingly unindented,
however, which ended up setting a suboptimal precedent as well. Unindent
those lines now.
Reported-by: Cole Robinson <crobinso@redhat.com>
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Libosinfo seems to generate Fedora tree URLs using the "https", not
"http", scheme now; which breaks CI. Update the expected outputs
accordingly.
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Just to be sure, this patch implements the second approach (described in
the previous patch) as well.
Note that there is precedent for suppressing "guest_lookup" exceptions:
refer to the "Error determining machine list" branch from commit
ae7ebc220b15 ("details: Properly limit machine type list by guests
arch/type", 2013-09-01).
(
In fact, that branch gets activated when opening the details window for a
domain that uses a non-default emulator; the "virt-manager --debug" log
contains:
> ERROR (details:613) Error determining machine list
> Traceback (most recent call last):
> File "virtManager/details/details.py", line 605, in _init_details
> capsinfo = caps.guest_lookup(
> File "virtinst/capabilities.py", line 319, in guest_lookup
> raise ValueError(msg)
> ValueError: Host does not support domain type kvm with machine
> 'pc-q35-8.1' for virtualization type 'hvm' with architecture 'x86_64'
)
Fixes: #539
Fixes: 05fcc7410eee ("virtinst: fix caching of domain capabilities", 2022-07-27)
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
When I try to open the details window for a domain that does not use the
system default emulator, I get the following exception:
> Traceback (most recent call last):
> File "virtManager/vmwindow.py", line 40, in get_instance
> cls._instances[key] = vmmVMWindow(vm)
> File "virtManager/vmwindow.py", line 83, in __init__
> self._details = vmmDetails(self.vm, self.builder, self.topwin,
> File "virtManager/details/details.py", line 389, in __init__
> self._init_details()
> File "virtManager/details/details.py", line 807, in _init_details
> vmmAddHardware.build_video_combo(self.vm, video_dev)
> File "virtManager/addhardware.py", line 816, in build_video_combo
> default = DeviceVideo.default_model(vm.xmlobj)
> File "virtinst/devices/video.py", line 47, in default_model
> if (guest.lookup_domcaps().supports_video_virtio() and
> File "virtinst/guest.py", line 656, in lookup_domcaps
> if not self._domcaps or not _compare(self._domcaps):
> File "virtinst/guest.py", line 646, in _compare
> if self.os.machine and not _compare_machine(domcaps):
> File "virtinst/guest.py", line 633, in _compare_machine
> capsinfo = self.lookup_capsinfo()
> File "virtinst/guest.py", line 674, in lookup_capsinfo
> self._capsinfo = self.conn.caps.guest_lookup(
> File "virtinst/capabilities.py", line 319, in guest_lookup
> raise ValueError(msg)
> ValueError: Host does not support domain type kvm with machine
> 'pc-q35-8.1' for virtualization type 'hvm' with architecture 'x86_64'
This is a regression; according to git-bisect, it was introduced in commit
05fcc7410eee ("virtinst: fix caching of domain capabilities", 2022-07-27).
"lookup_capsinfo" (and "guest_lookup" called by it) are unsuitable for
machine type alias checking (or for anything else) if the domain uses an
emulator that differs from the system default emulator. The information
returned by virConnectGetCapabilities() pertains to the system default
emulator. Thus, when using a non-default emulator, we should either not
call "lookup_capsinfo" for machine type alias checking, *or* we should
suppress the exception, and pretend that the alias check was a mismatch.
It turns out that we can avoid the "lookup_capsinfo" call (and thereby the
exception) in practice if we just delay the call until after the direct
(non-alias) comparison.
Fixes: #539
Fixes: 05fcc7410eee ("virtinst: fix caching of domain capabilities", 2022-07-27)
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
It is common on x86 and other architectures to install a guest from
network by using "--boot hd,network" with virt-install - as long as
the hard disk is not bootable yet, the installation will be started
via network, and once it finished, the guest can boot from hd during
the next reboot.
However, this does not work on s390x since this architecture only
supports one single boot device to be passed to the guest. Thus add
a note to the documentation to avoid that people are running again
into this common pitfall.
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2032472
Signed-off-by: Thomas Huth <thuth@redhat.com>
Before this, on clone the serial file would remain the same for the cloned
domain. This doesn't make much sense as the output would be an intermix of two
unrelated output sequences.
Here we apply the the same filename changing algorithm, as with disk files.
Signed-off-by: Oleg Vasilev <oleg.vasilev@virtuozzo.com>
Provide ready to use recipes for explicitly enabling and
explicitly disabling Secure Boot, as well as a pointer to
the more extensive information found on the libvirt website.
Setting loader_secure=yes is only one part of a proper Secure
Boot setup, so stop documenting it in the section about manual
firmware selection to avoid confusion.
https://bugzilla.redhat.com/show_bug.cgi?id=2112154https://bugzilla.redhat.com/show_bug.cgi?id=2149971
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Firmware autoselection is the way to go in most cases, so
recommend that instead of telling users that they should provide
all information manually.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
This commit enables Packit `copr_build` tasks which will run on every PR
and build RPMS using the spec file present upstream with Source0 as
the archive created from HEAD commit of the PR.
Signed-off-by: Lokesh Mandvekar <lsm5@fedoraproject.org>
virt-install prints the total_size value to the progress bar even if it
is meaningless.
This value can be confusing to user, so for execute prosess that doesn't
copy files (total_size = 0B), we hide the total_size value.
For example, 'Creating domain...' doesn't need to print the total_size
value.
Signed-off-by: Toshiki Sonoda <sonoda.toshiki@fujitsu.com>
Signed-off-by: Haruka Ohata <ohata.haruka@fujitsu.com>
When a sparse file is created during a disk allocation,
virt-install prints not the created disk size but a sparse file size.
Therefore, we fix to print the created disk size during disk allocation
instead of the size of the sparse file by updating the meter with the
self.capacity.
Signed-off-by: Toshiki Sonoda <sonoda.toshiki@fujitsu.com>
Signed-off-by: Haruka Ohata <ohata.haruka@fujitsu.com>
Reproducer:
Reproducer:
./virt-install --connect test:///default \
--location tests/data/fakemedia/fake-f26-netinst.iso
Before:
Starting install...
Retrieving 'vmlinuz' | 0 B 00:00:00 ...
Retrieving 'initrd.img' | 0 B 00:00:00 ...
After:
Starting install...
Retrieving 'vmlinuz' | 9 B 00:00:00 ...
Retrieving 'initrd.img' | 9 B 00:00:00 ...
progress.end() currently only reports the total amount of bytes
that were last written to the UI. It should report the total amount
that's been passed to update().
Reported-by: Toshiki Sonoda <sonoda.toshiki@fujitsu.com>
Signed-off-by: Cole Robinson <crobinso@redhat.com>
* Move browse_reason handling entirely into storagebrowser.py
* Open code some of the browse_local logic at the few callers
Signed-off-by: Cole Robinson <crobinso@redhat.com>
When creating a new VM, in the customize wizard we can't depend on
index= value being set (virtinst doesn't do it for example).
For example, this causes a backtrace when adding two virtio-scsi
controllers via the Customize wizard, or adding an extra
virtio-scsi controller to an aarch64 CDROM install.
Reported-by: Charles Arnold <carnold@suse.com>
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The sheepdog project is no longer actively developed, Libvirt removed
the support for sheepdog storage backend since v8.8.0, Let's drop it.
Signed-off-by: Lin Ma <lma@suse.com>
Nowadays it could be as simple as `virt-install --install fedora36`.
Trying to represent the interdepencies here is not worth it, but
let's keep a simple string around to avoid the default parser
usage string, which is huge
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This commit added support for cpu physical address bits control, It's
useful for VMs with huge amount of ram.
E.g.
--cpu Cascadelake-Server,maxphysaddr.mode=emulate,maxphysaddr.bits=46
Signed-off-by: Lin Ma <lma@suse.com>
Anything passed to --boot should imply --install no_install=yes
in the absence of other --install options. This is historically
what we've done but we regressed in 4.1.0
Resolves: https://github.com/virt-manager/virt-manager/issues/426
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Each bit here is part of the CLI API, we need to be sure we are
covering each one. Extend the test suite to hit one case we are missing
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Using the source.socket of virtiofs needs a virtiofsd daemon launched
outside of libvirtd, So the filesystem UI doesn't support it yet. If
users need it they can set it manually in the XML editor.
But if we view the filesystem info of such a VM on the details page,
It fails with this error message:
Traceback (most recent call last):
File "/usr/share/virt-manager/virtManager/details/details.py", line 1713, in _refresh_page
self._refresh_filesystem_page(dev)
File "/usr/share/virt-manager/virtManager/details/details.py", line 2241, in _refresh_filesystem_page
self.fsDetails.set_dev(dev)
File "/usr/share/virt-manager/virtManager/device/fsdetails.py", line 193, in set_dev
self.widget("fs-source").set_text(dev.source)
TypeError: Argument 1 does not allow None as a value
This patch fixes above issue by leaving the 'source path' info blank in
case of source.socket.
In this case, Considering that showing 'target path' info without source
info is kind of meaningless, So this patch leaves the 'target path' info
blank as well.
Signed-off-by: Lin Ma <lma@suse.com>
Properly setting the metadata cache size can provide better performance
in case of using big qcow2 images.
This patch introduces two driver options:
* driver.metadata_cache.max_size
* driver.metadata_cache.max_size.unit
E.g. --disk ...,driver.type=qcow2,\
driver.metadata_cache.max_size=2,\
driver.metadata_cache.max_size.unit=MiB
BTW, Metadata cache size control is currently supported only for qcow2.
Regarding how to properly caluclate the cache size of qcow2, Please refer
to qemu's documentation.
Signed-off-by: Lin Ma <lma@suse.com>
We need to change the flow from
* parse all the strings
* set capabilities defaults
* build installer
* fill in all guest defaults
To
* parse boot and metadata strings
* set capabilities defaults
* build installer
* set --name default
* parse all the remaining strings
* fill in all guest defaults
Because --disk parsing depends on --name for some path generation.
So this fixes --disk names when --name is implicitly specified by
--install or --osinfo
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The diff_compare rstrip() should never have been added, let's fix
it once and for all but dealing with missing newlines in the diff
helper
Signed-off-by: Cole Robinson <crobinso@redhat.com>
It allows to set the thread pool size to optimize spawning worker threads
for the default event loop in real time environment. For example:
--iothreads defaultiothread.thread_pool_min=8,\
defaultiothread.thread_pool_max=16
Signed-off-by: Lin Ma <lma@suse.com>
+ ./setup.py configure --default-hvs qemu,xen,lxc
error: Multiple top-level packages discovered in a flat-layout: ['po', 'ui', 'man', 'data', 'virtinst', 'virtManager'].
To avoid accidental inclusion of unwanted files or directories,
setuptools will not proceed with this build.
If you are trying to create a single distribution with multiple packages
on purpose, you should not rely on automatic discovery.
Instead, consider the following options:
1. set up custom discovery (`find` directive with `include` or `exclude`)
2. use a `src-layout`
3. explicitly set `py_modules` or `packages` with a list of names
To find more information, look for "package discovery" on setuptools docs.
Downstream bug report: https://bugzilla.redhat.com/2113754
These two optional attributes allow setting lower and upper boundary for
number of worker threads for given IOThread. For example:
--iothreads iothreads=2,\
iothreadids.iothread0.id=1,\
iothreadids.iothread1.id=2,\
iothreadids.iothread1.thread_pool_min=8,\
iothreadids.iothread1.thread_pool_max=16
Signed-off-by: Lin Ma <lma@suse.com>
Otherwise going host-passthrough -> custom can cause libvirt validation
error due to libvirt fills the default value(migratable='on') for the
host-passthrough in domain XML.
Signed-off-by: Lin Ma <lma@suse.com>
Remove the open coded version logging in cli.py and virt-manager
connection.py, and move it into virtinst connection open
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The various version numbers reported by libvirt are all relevant pieces
of information when debugging problem reports.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Knowledge of the version number is not always available from the user
bug report. They may merely have a log file from an automated system.
Including the version number in the log message is key information.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The domain capabilities XML always contains the canonical machine name.
When the selected machine type for a guest is an alias, the check
comparing it to the domain capabilities machine will always fail. This
prevents the domain capabilities result from being cached.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
The new format is too long and wordy IMO. It was added to handle
qemu-vdagent ambiguity, since without it we would print the same string
for spicevmc and qemu-vgagent channel. Let's just special case
qemu-vdagent to solve that problem
Signed-off-by: Cole Robinson <crobinso@redhat.com>
This provides the UI support for the qemu-vdagent channel which allows
clipboard sharing with VNC graphics (see previous commit for more
information).
The channel name in the device list was changed slightly in order to
avoid confusion. Due to the fact that both the spice-vdagent and the
qemu-vdagent specify the same virtio name (com.redhat.spice.0), both of
these channels were showing up in the device list as "Channel spice",
which is a bit confusing.
In order to disambiguate these, channels now show up in the device list
as "Channel {type} ({name})" instead of "Channel {name}". So for
example, a qemu-vdagent channel would show up as:
Channel Qemu vdagent (spice)
Whereas a spice-vdagent channel would show up as:
Channel Spice agent (spice)
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com>
This allows support for host/guest clipboard sharing when using vnc
guests (and possibly other graphics types in the future). This channel
is similar to the spicevmc channel, but it contains a couple additional
options to enable/disable clipboard sharing and specify the mouse mode.
In the case of spice, these settings are specified on the 'graphics'
element, but for qemu-vdagent, they are specified on the channel. For
example:
--channel=qemu-vdagent,source.clipboard.copypaste=on,source.mouse.mode=client
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com>
* More info about translating as part of Fedora: https://fedoraproject.org/wiki/L10N/Translate_on_Weblate
* The up to date translation `.pot` template is stored in the [`translations` branch](https://github.com/virt-manager/virt-manager/tree/translations) and synced with the `main` branch before release.
* The up to date translation `.pot` template is stored in the `main` branch
* Translations are submitted by Weblate as pull requests, usually merged to the
`main` branch before release and whenever needed (e.g. before updating the
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
