file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
#
|
|
|
|
# The Qubes OS Project, https://www.qubes-os.org/
|
|
|
|
#
|
|
|
|
# Copyright (C) 2018 Rusty Bird <rustybird@net-c.com>
|
|
|
|
#
|
|
|
|
# This library is free software; you can redistribute it and/or
|
|
|
|
# modify it under the terms of the GNU Lesser General Public
|
|
|
|
# License as published by the Free Software Foundation; either
|
|
|
|
# version 2.1 of the License, or (at your option) any later version.
|
|
|
|
#
|
|
|
|
# This library is distributed in the hope that it will be useful,
|
|
|
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
# Lesser General Public License for more details.
|
|
|
|
#
|
|
|
|
# You should have received a copy of the GNU Lesser General Public
|
|
|
|
# License along with this library; if not, see <https://www.gnu.org/licenses/>.
|
|
|
|
#
|
|
|
|
|
|
|
|
''' Driver for handling VM images as files, without any device-mapper
|
|
|
|
involvement. A reflink-capable filesystem is strongly recommended,
|
|
|
|
but not required.
|
|
|
|
'''
|
|
|
|
|
|
|
|
import collections
|
|
|
|
import errno
|
|
|
|
import fcntl
|
|
|
|
import glob
|
|
|
|
import logging
|
|
|
|
import os
|
|
|
|
import re
|
|
|
|
import subprocess
|
|
|
|
import tempfile
|
|
|
|
from contextlib import contextmanager, suppress
|
|
|
|
|
|
|
|
import qubes.storage
|
|
|
|
|
|
|
|
BLKSIZE = 512
|
|
|
|
FICLONE = 1074041865 # see ioctl_ficlone manpage
|
2018-02-16 22:47:37 +01:00
|
|
|
LOGGER = logging.getLogger('qubes.storage.reflink')
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
|
|
|
|
class ReflinkPool(qubes.storage.Pool):
|
|
|
|
driver = 'file-reflink'
|
|
|
|
_known_dir_path_prefixes = ['appvms', 'vm-templates']
|
|
|
|
|
|
|
|
def __init__(self, dir_path, setup_check='yes', revisions_to_keep=1,
|
|
|
|
**kwargs):
|
|
|
|
super().__init__(revisions_to_keep=revisions_to_keep, **kwargs)
|
|
|
|
self._volumes = {}
|
|
|
|
self.dir_path = os.path.abspath(dir_path)
|
|
|
|
self.setup_check = qubes.property.bool(None, None, setup_check)
|
|
|
|
|
|
|
|
def setup(self):
|
|
|
|
created = _make_dir(self.dir_path)
|
|
|
|
if self.setup_check and not is_reflink_supported(self.dir_path):
|
|
|
|
if created:
|
|
|
|
_remove_empty_dir(self.dir_path)
|
|
|
|
raise qubes.storage.StoragePoolException(
|
|
|
|
'The filesystem for {!r} does not support reflinks. If you'
|
|
|
|
' can live with VM startup delays and wasted disk space, pass'
|
|
|
|
' the "setup_check=no" option.'.format(self.dir_path))
|
|
|
|
for dir_path_prefix in self._known_dir_path_prefixes:
|
|
|
|
_make_dir(os.path.join(self.dir_path, dir_path_prefix))
|
|
|
|
return self
|
|
|
|
|
|
|
|
def init_volume(self, vm, volume_config):
|
|
|
|
# Fail closed on any strange VM dir_path_prefix, just in case
|
|
|
|
# /etc/udev/rules/00-qubes-ignore-devices.rules needs updating
|
|
|
|
assert vm.dir_path_prefix in self._known_dir_path_prefixes, \
|
|
|
|
'Unknown dir_path_prefix {!r}'.format(vm.dir_path_prefix)
|
|
|
|
|
|
|
|
volume_config['pool'] = self
|
|
|
|
if 'revisions_to_keep' not in volume_config:
|
|
|
|
volume_config['revisions_to_keep'] = self.revisions_to_keep
|
|
|
|
if 'vid' not in volume_config:
|
|
|
|
volume_config['vid'] = os.path.join(vm.dir_path_prefix, vm.name,
|
|
|
|
volume_config['name'])
|
|
|
|
volume = ReflinkVolume(**volume_config)
|
|
|
|
self._volumes[volume_config['vid']] = volume
|
|
|
|
return volume
|
|
|
|
|
|
|
|
def list_volumes(self):
|
|
|
|
return list(self._volumes.values())
|
|
|
|
|
|
|
|
def get_volume(self, vid):
|
|
|
|
return self._volumes[vid]
|
|
|
|
|
|
|
|
def destroy(self):
|
|
|
|
pass
|
|
|
|
|
|
|
|
@property
|
|
|
|
def config(self):
|
|
|
|
return {
|
|
|
|
'name': self.name,
|
|
|
|
'dir_path': self.dir_path,
|
|
|
|
'driver': ReflinkPool.driver,
|
|
|
|
'revisions_to_keep': self.revisions_to_keep
|
|
|
|
}
|
|
|
|
|
|
|
|
@property
|
|
|
|
def size(self):
|
|
|
|
statvfs = os.statvfs(self.dir_path)
|
|
|
|
return statvfs.f_frsize * statvfs.f_blocks
|
|
|
|
|
|
|
|
@property
|
|
|
|
def usage(self):
|
|
|
|
statvfs = os.statvfs(self.dir_path)
|
|
|
|
return statvfs.f_frsize * (statvfs.f_blocks - statvfs.f_bfree)
|
|
|
|
|
2018-03-19 22:36:45 +01:00
|
|
|
def included_in(self, app):
|
|
|
|
''' Check if there is pool containing this one - either as a
|
|
|
|
filesystem or its LVM volume'''
|
|
|
|
return qubes.storage.search_pool_containing_dir(
|
|
|
|
[pool for pool in app.pools.values() if pool is not self],
|
|
|
|
self.dir_path)
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
class ReflinkVolume(qubes.storage.Volume):
|
2018-09-09 22:01:15 +02:00
|
|
|
def __init__(self, *args, **kwargs):
|
|
|
|
super().__init__(*args, **kwargs)
|
|
|
|
self._path_vid = os.path.join(self.pool.dir_path, self.vid)
|
|
|
|
self._path_clean = self._path_vid + '.img'
|
|
|
|
self._path_dirty = self._path_vid + '-dirty.img'
|
|
|
|
self.path = self._path_dirty
|
|
|
|
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
def create(self):
|
|
|
|
if self.save_on_stop and not self.snap_on_start:
|
|
|
|
_create_sparse_file(self._path_clean, self.size)
|
|
|
|
return self
|
|
|
|
|
|
|
|
def verify(self):
|
|
|
|
if self.snap_on_start:
|
2018-03-11 16:35:00 +01:00
|
|
|
img = self.source._path_clean # pylint: disable=protected-access
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
elif self.save_on_stop:
|
|
|
|
img = self._path_clean
|
|
|
|
else:
|
|
|
|
img = None
|
|
|
|
|
|
|
|
if img is None or os.path.exists(img):
|
|
|
|
return True
|
|
|
|
raise qubes.storage.StoragePoolException(
|
2018-09-09 22:01:11 +02:00
|
|
|
'Missing image file {!r} for volume {}'.format(img, self.vid))
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
def remove(self):
|
|
|
|
''' Drop volume object from pool; remove volume images from
|
|
|
|
oldest to newest; remove empty VM directory.
|
|
|
|
'''
|
2018-03-11 16:35:00 +01:00
|
|
|
self.pool._volumes.pop(self, None) # pylint: disable=protected-access
|
2018-09-09 22:01:17 +02:00
|
|
|
self._cleanup()
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
self._prune_revisions(keep=0)
|
|
|
|
_remove_file(self._path_clean)
|
|
|
|
_remove_file(self._path_dirty)
|
2018-03-11 16:34:58 +01:00
|
|
|
_remove_empty_dir(os.path.dirname(self._path_dirty))
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
return self
|
|
|
|
|
2018-09-09 22:01:17 +02:00
|
|
|
def _cleanup(self):
|
|
|
|
for tmp in glob.iglob(glob.escape(self._path_vid) + '*.img*~*'):
|
|
|
|
_remove_file(tmp)
|
|
|
|
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
def is_outdated(self):
|
|
|
|
if self.snap_on_start:
|
|
|
|
with suppress(FileNotFoundError):
|
|
|
|
# pylint: disable=protected-access
|
|
|
|
return (os.path.getmtime(self.source._path_clean) >
|
|
|
|
os.path.getmtime(self._path_clean))
|
|
|
|
return False
|
|
|
|
|
|
|
|
def is_dirty(self):
|
|
|
|
return self.save_on_stop and os.path.exists(self._path_dirty)
|
|
|
|
|
|
|
|
def start(self):
|
2018-09-09 22:01:17 +02:00
|
|
|
self._cleanup()
|
2018-03-12 17:38:56 +01:00
|
|
|
if self.is_dirty(): # implies self.save_on_stop
|
|
|
|
return self
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
if self.snap_on_start:
|
|
|
|
# pylint: disable=protected-access
|
|
|
|
_copy_file(self.source._path_clean, self._path_clean)
|
2018-03-12 17:38:56 +01:00
|
|
|
if self.snap_on_start or self.save_on_stop:
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
_copy_file(self._path_clean, self._path_dirty)
|
|
|
|
else:
|
|
|
|
_create_sparse_file(self._path_dirty, self.size)
|
|
|
|
return self
|
|
|
|
|
|
|
|
def stop(self):
|
|
|
|
if self.save_on_stop:
|
|
|
|
self._commit()
|
|
|
|
else:
|
|
|
|
_remove_file(self._path_dirty)
|
2018-03-11 16:35:00 +01:00
|
|
|
_remove_file(self._path_clean)
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
return self
|
|
|
|
|
|
|
|
def _commit(self):
|
|
|
|
self._add_revision()
|
|
|
|
self._prune_revisions()
|
|
|
|
_rename_file(self._path_dirty, self._path_clean)
|
|
|
|
|
|
|
|
def _add_revision(self):
|
2018-02-16 22:47:39 +01:00
|
|
|
if self.revisions_to_keep == 0:
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
return
|
|
|
|
ctime = os.path.getctime(self._path_clean)
|
2018-03-21 17:00:13 +01:00
|
|
|
timestamp = qubes.storage.isodate(int(ctime))
|
|
|
|
_copy_file(self._path_clean,
|
|
|
|
self._path_revision(self._next_revision_number, timestamp))
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
def _prune_revisions(self, keep=None):
|
|
|
|
if keep is None:
|
|
|
|
keep = self.revisions_to_keep
|
|
|
|
# pylint: disable=invalid-unary-operand-type
|
2018-03-21 17:00:13 +01:00
|
|
|
for number, timestamp in list(self.revisions.items())[:-keep or None]:
|
|
|
|
_remove_file(self._path_revision(number, timestamp))
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
def revert(self, revision=None):
|
2018-09-09 22:01:13 +02:00
|
|
|
if self.is_dirty():
|
|
|
|
raise qubes.storage.StoragePoolException(
|
|
|
|
'Cannot revert: {} is not cleanly stopped'.format(self.vid))
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
if revision is None:
|
2018-03-21 17:00:13 +01:00
|
|
|
number, timestamp = list(self.revisions.items())[-1]
|
|
|
|
else:
|
|
|
|
number, timestamp = revision, None
|
|
|
|
path_revision = self._path_revision(number, timestamp)
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
self._add_revision()
|
2018-03-21 17:00:13 +01:00
|
|
|
_rename_file(path_revision, self._path_clean)
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
return self
|
|
|
|
|
|
|
|
def resize(self, size):
|
|
|
|
''' Expand a read-write volume image; notify any corresponding
|
|
|
|
loop devices of the size change.
|
|
|
|
'''
|
|
|
|
if not self.rw:
|
|
|
|
raise qubes.storage.StoragePoolException(
|
2018-09-09 22:01:11 +02:00
|
|
|
'Cannot resize: {} is read-only'.format(self.vid))
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
if size < self.size:
|
|
|
|
raise qubes.storage.StoragePoolException(
|
2018-09-09 22:01:11 +02:00
|
|
|
'For your own safety, shrinking of {} is disabled'
|
|
|
|
' ({} < {}). If you really know what you are doing,'
|
2018-03-11 16:34:56 +01:00
|
|
|
' use "truncate" manually.'.format(self.vid, size, self.size))
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
try: # assume volume is not (cleanly) stopped ...
|
|
|
|
_resize_file(self._path_dirty, size)
|
|
|
|
except FileNotFoundError: # ... but it actually is.
|
|
|
|
_resize_file(self._path_clean, size)
|
|
|
|
|
|
|
|
self.size = size
|
|
|
|
|
|
|
|
# resize any corresponding loop devices
|
|
|
|
out = _cmd('losetup', '--associated', self._path_dirty)
|
|
|
|
for match in re.finditer(br'^(/dev/loop[0-9]+): ', out, re.MULTILINE):
|
|
|
|
loop_dev = match.group(1).decode('ascii')
|
|
|
|
_cmd('losetup', '--set-capacity', loop_dev)
|
|
|
|
|
|
|
|
return self
|
|
|
|
|
|
|
|
def _require_save_on_stop(self, method_name):
|
|
|
|
if not self.save_on_stop:
|
|
|
|
raise NotImplementedError(
|
2018-09-09 22:01:11 +02:00
|
|
|
'Cannot {}: {} is not save_on_stop'.format(
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
method_name, self.vid))
|
|
|
|
|
|
|
|
def export(self):
|
|
|
|
self._require_save_on_stop('export')
|
|
|
|
return self._path_clean
|
|
|
|
|
|
|
|
def import_data(self):
|
|
|
|
self._require_save_on_stop('import_data')
|
|
|
|
_create_sparse_file(self._path_dirty, self.size)
|
|
|
|
return self._path_dirty
|
|
|
|
|
|
|
|
def import_data_end(self, success):
|
|
|
|
if success:
|
|
|
|
self._commit()
|
|
|
|
else:
|
|
|
|
_remove_file(self._path_dirty)
|
|
|
|
return self
|
|
|
|
|
|
|
|
def import_volume(self, src_volume):
|
|
|
|
self._require_save_on_stop('import_volume')
|
|
|
|
try:
|
|
|
|
_copy_file(src_volume.export(), self._path_dirty)
|
|
|
|
except:
|
|
|
|
self.import_data_end(False)
|
|
|
|
raise
|
|
|
|
self.import_data_end(True)
|
|
|
|
return self
|
|
|
|
|
2018-03-21 17:00:13 +01:00
|
|
|
def _path_revision(self, number, timestamp=None):
|
|
|
|
if timestamp is None:
|
|
|
|
timestamp = self.revisions[number]
|
|
|
|
return self._path_clean + '.' + number + '@' + timestamp + 'Z'
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
2018-03-21 17:00:13 +01:00
|
|
|
@property
|
|
|
|
def _next_revision_number(self):
|
|
|
|
numbers = self.revisions.keys()
|
|
|
|
if numbers:
|
|
|
|
return str(int(list(numbers)[-1]) + 1)
|
|
|
|
return '1'
|
|
|
|
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
@property
|
|
|
|
def revisions(self):
|
2018-03-21 17:00:13 +01:00
|
|
|
prefix = self._path_clean + '.'
|
2018-09-09 22:01:14 +02:00
|
|
|
paths = glob.iglob(glob.escape(prefix) + '*@*Z')
|
|
|
|
items = (path[len(prefix):-1].split('@') for path in paths)
|
|
|
|
return collections.OrderedDict(sorted(items,
|
|
|
|
key=lambda item: int(item[0])))
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
@property
|
|
|
|
def usage(self):
|
|
|
|
''' Return volume disk usage from the VM's perspective. It is
|
|
|
|
usually much lower from the host's perspective due to CoW.
|
|
|
|
'''
|
|
|
|
with suppress(FileNotFoundError):
|
|
|
|
return _get_file_disk_usage(self._path_dirty)
|
|
|
|
with suppress(FileNotFoundError):
|
|
|
|
return _get_file_disk_usage(self._path_clean)
|
|
|
|
return 0
|
|
|
|
|
|
|
|
|
|
|
|
@contextmanager
|
|
|
|
def _replace_file(dst):
|
|
|
|
''' Yield a tempfile whose name starts with dst, creating the last
|
|
|
|
directory component if necessary. If the block does not raise
|
|
|
|
an exception, flush+fsync the tempfile and rename it to dst.
|
|
|
|
'''
|
|
|
|
tmp_dir, prefix = os.path.split(dst + '~')
|
|
|
|
_make_dir(tmp_dir)
|
|
|
|
tmp = tempfile.NamedTemporaryFile(dir=tmp_dir, prefix=prefix, delete=False)
|
|
|
|
try:
|
|
|
|
yield tmp
|
|
|
|
tmp.flush()
|
|
|
|
os.fsync(tmp.fileno())
|
|
|
|
tmp.close()
|
|
|
|
_rename_file(tmp.name, dst)
|
|
|
|
except:
|
|
|
|
tmp.close()
|
|
|
|
_remove_file(tmp.name)
|
|
|
|
raise
|
|
|
|
|
|
|
|
def _get_file_disk_usage(path):
|
|
|
|
''' Return real disk usage (not logical file size) of a file. '''
|
|
|
|
return os.stat(path).st_blocks * BLKSIZE
|
|
|
|
|
|
|
|
def _fsync_dir(path):
|
|
|
|
dir_fd = os.open(path, os.O_RDONLY | os.O_DIRECTORY)
|
|
|
|
try:
|
|
|
|
os.fsync(dir_fd)
|
|
|
|
finally:
|
|
|
|
os.close(dir_fd)
|
|
|
|
|
|
|
|
def _make_dir(path):
|
|
|
|
''' mkdir path, ignoring FileExistsError; return whether we
|
|
|
|
created it.
|
|
|
|
'''
|
|
|
|
with suppress(FileExistsError):
|
|
|
|
os.mkdir(path)
|
|
|
|
_fsync_dir(os.path.dirname(path))
|
|
|
|
LOGGER.info('Created directory: %s', path)
|
|
|
|
return True
|
|
|
|
return False
|
|
|
|
|
|
|
|
def _remove_file(path):
|
|
|
|
with suppress(FileNotFoundError):
|
|
|
|
os.remove(path)
|
|
|
|
_fsync_dir(os.path.dirname(path))
|
|
|
|
LOGGER.info('Removed file: %s', path)
|
|
|
|
|
|
|
|
def _remove_empty_dir(path):
|
2018-03-11 16:34:58 +01:00
|
|
|
try:
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
os.rmdir(path)
|
|
|
|
_fsync_dir(os.path.dirname(path))
|
|
|
|
LOGGER.info('Removed empty directory: %s', path)
|
2018-03-11 16:34:58 +01:00
|
|
|
except OSError as ex:
|
|
|
|
if ex.errno not in (errno.ENOENT, errno.ENOTEMPTY):
|
|
|
|
raise
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
def _rename_file(src, dst):
|
|
|
|
os.rename(src, dst)
|
|
|
|
dst_dir = os.path.dirname(dst)
|
|
|
|
src_dir = os.path.dirname(src)
|
|
|
|
_fsync_dir(dst_dir)
|
|
|
|
if src_dir != dst_dir:
|
|
|
|
_fsync_dir(src_dir)
|
|
|
|
LOGGER.info('Renamed file: %s -> %s', src, dst)
|
|
|
|
|
|
|
|
def _resize_file(path, size):
|
|
|
|
''' Resize an existing file. '''
|
|
|
|
with open(path, 'rb+') as file:
|
|
|
|
file.truncate(size)
|
2018-03-11 16:34:55 +01:00
|
|
|
os.fsync(file.fileno())
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
|
|
|
|
def _create_sparse_file(path, size):
|
|
|
|
''' Create an empty sparse file. '''
|
|
|
|
with _replace_file(path) as tmp:
|
|
|
|
tmp.truncate(size)
|
|
|
|
LOGGER.info('Created sparse file: %s', tmp.name)
|
|
|
|
|
|
|
|
def _copy_file(src, dst):
|
|
|
|
''' Copy src to dst as a reflink if possible, sparse if not. '''
|
|
|
|
if not os.path.exists(src):
|
|
|
|
raise FileNotFoundError(src)
|
|
|
|
with _replace_file(dst) as tmp:
|
|
|
|
LOGGER.info('Copying file: %s -> %s', src, tmp.name)
|
|
|
|
_cmd('cp', '--sparse=always', '--reflink=auto', src, tmp.name)
|
|
|
|
|
|
|
|
def _cmd(*args):
|
|
|
|
''' Run command until finished; return stdout (as bytes) if it
|
|
|
|
exited 0. Otherwise, raise a detailed StoragePoolException.
|
|
|
|
'''
|
|
|
|
try:
|
|
|
|
return subprocess.run(args, check=True,
|
|
|
|
stdout=subprocess.PIPE,
|
|
|
|
stderr=subprocess.PIPE).stdout
|
|
|
|
except subprocess.CalledProcessError as ex:
|
2018-09-09 22:01:11 +02:00
|
|
|
msg = '{} err={!r} out={!r}'.format(ex, ex.stderr, ex.stdout)
|
file-reflink, a storage driver optimized for CoW filesystems
This adds the file-reflink storage driver. It is never selected
automatically for pool creation, especially not the creation of
'varlibqubes' (though it can be used if set up manually).
The code is quite small:
reflink.py lvm.py file.py + block-snapshot
sloccount 334 lines 447 (134%) 570 (171%)
Background: btrfs and XFS (but not yet ZFS) support instant copies of
individual files through the 'FICLONE' ioctl behind 'cp --reflink'.
Which file-reflink uses to snapshot VM image files without an extra
device-mapper layer. All the snapshots are essentially freestanding;
there's no functional origin vs. snapshot distinction.
In contrast to 'file'-on-btrfs, file-reflink inherently avoids
CoW-on-CoW. Which is a bigger issue now on R4.0, where even AppVMs'
private volumes are CoW. (And turning off the lower, filesystem-level
CoW for 'file'-on-btrfs images would turn off data checksums too, i.e.
protection against bit rot.)
Also in contrast to 'file', all storage features are supported,
including
- any number of revisions_to_keep
- volume.revert()
- volume.is_outdated
- online fstrim/discard
Example tree of a file-reflink pool - *-dirty.img are connected to Xen:
- /var/lib/testpool/appvms/foo/volatile-dirty.img
- /var/lib/testpool/appvms/foo/root-dirty.img
- /var/lib/testpool/appvms/foo/root.img
- /var/lib/testpool/appvms/foo/private-dirty.img
- /var/lib/testpool/appvms/foo/private.img
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T03:04:05Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T04:05:06Z
- /var/lib/testpool/appvms/foo/private.img@2018-01-02T05:06:07Z
- /var/lib/testpool/appvms/bar/...
- /var/lib/testpool/appvms/...
- /var/lib/testpool/template-vms/fedora-26/...
- /var/lib/testpool/template-vms/...
It looks similar to a 'file' pool tree, and in fact file-reflink is
drop-in compatible:
$ qvm-shutdown --all --wait
$ systemctl stop qubesd
$ sed 's/ driver="file"/ driver="file-reflink"/g' -i.bak /var/lib/qubes/qubes.xml
$ systemctl start qubesd
$ sudo rm -f /path/to/pool/*/*/*-cow.img*
If the user tries to create a fresh file-reflink pool on a filesystem
that doesn't support reflinks, qvm-pool will abort and mention the
'setup_check=no' option. Which can be passed to force a fallback on
regular sparse copies, with of course lots of time/space overhead. The
same fallback code is also used when initially cloning a VM from a
foreign pool, or from another file-reflink pool on a different
mountpoint.
'journalctl -fu qubesd' will show all file-reflink copy/rename/remove
operations on VM creation/startup/shutdown/etc.
2018-02-12 22:20:05 +01:00
|
|
|
raise qubes.storage.StoragePoolException(msg) from ex
|
|
|
|
|
|
|
|
def is_reflink_supported(dst_dir, src_dir=None):
|
|
|
|
''' Return whether destination directory supports reflink copies
|
|
|
|
from source directory. (A temporary file is created in each
|
|
|
|
directory, using O_TMPFILE if possible.)
|
|
|
|
'''
|
|
|
|
if src_dir is None:
|
|
|
|
src_dir = dst_dir
|
|
|
|
dst = tempfile.TemporaryFile(dir=dst_dir)
|
|
|
|
src = tempfile.TemporaryFile(dir=src_dir)
|
|
|
|
src.write(b'foo') # don't let any filesystem get clever with empty files
|
|
|
|
|
|
|
|
try:
|
|
|
|
fcntl.ioctl(dst.fileno(), FICLONE, src.fileno())
|
|
|
|
return True
|
|
|
|
except OSError:
|
|
|
|
return False
|