core-admin/qubes/storage/reflink.py

430 lines
14 KiB
Python
Raw Normal View History

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)
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):
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(
'Missing image file {!r} for volume {!s}'.format(img, self.vid))
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
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)
_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
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):
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)
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
2018-02-16 22:47:39 +01:00
if _get_file_disk_usage(self._path_clean) == 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)
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
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):
if revision is None:
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()
_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(
'Cannot resize: {!s} is read-only'.format(self.vid))
if size < self.size:
raise qubes.storage.StoragePoolException(
'For your own safety, shrinking of {!s} is disabled'
' ({:d} < {:d}). If you really know what you are doing,'
' 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(
'Cannot {!s}: {!s} is not save_on_stop'.format(
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
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
@property
def _path_clean(self):
return os.path.join(self.pool.dir_path, self.vid + '.img')
@property
def _path_dirty(self):
return os.path.join(self.pool.dir_path, self.vid + '-dirty.img')
@property
def path(self):
return self._path_dirty
@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):
prefix = self._path_clean + '.'
paths = glob.glob(glob.escape(prefix) + '*@*Z')
items = sorted((path[len(prefix):-1].split('@') for path in paths),
key=lambda item: int(item[0]))
return collections.OrderedDict(items)
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):
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)
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)
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:
msg = '{!s} err={!r} out={!r}'.format(ex, ex.stderr, ex.stdout)
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