]> git.proxmox.com Git - qemu-server.git/log
qemu-server.git
7 years agoswitch to 'ha-manager set'
Thomas Lamprecht [Fri, 18 Nov 2016 16:42:17 +0000 (17:42 +0100)]
switch to 'ha-manager set'

ha-manager enabled/disabled will get removed

Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
7 years agovmstate snapshot : activate|deactivate volume
Alexandre Derumier [Mon, 14 Nov 2016 12:23:02 +0000 (13:23 +0100)]
vmstate snapshot : activate|deactivate volume

This fix vmstate snasphot on krbd volume

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agoqemu_volume_snapshot_delete : fix krbd snapshot delete
Alexandre Derumier [Mon, 14 Nov 2016 12:01:21 +0000 (13:01 +0100)]
qemu_volume_snapshot_delete : fix krbd snapshot delete

like for snapshot, we need to check if krbd is enabled, to known
if we need to use qmp delete-drive-snapshot or storage command directly

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agoVZDump/QemuServer: set bless clas correctly
Stefan Priebe [Wed, 16 Nov 2016 19:10:40 +0000 (20:10 +0100)]
VZDump/QemuServer: set bless clas correctly

Signed-off-by: Stefan Priebe <s.priebe@profihost.ag>
7 years agoqxlnum: use $winversion
Alexandre Derumier [Fri, 11 Nov 2016 08:32:11 +0000 (09:32 +0100)]
qxlnum: use $winversion

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agohv_vendor_id : for winversion >= 6 , not >=7.
Alexandre Derumier [Fri, 11 Nov 2016 08:32:10 +0000 (09:32 +0100)]
hv_vendor_id : for winversion >= 6 , not >=7.

as It was previously.

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agocleanup windows version handling and hyperv enlightments
Alexandre Derumier [Tue, 8 Nov 2016 01:56:01 +0000 (02:56 +0100)]
cleanup windows version handling and hyperv enlightments

This cleanup windows guest os version handling,
with normalizing ostype with numbers in a new windows_version sub.

if($ostype eq 'wxp' || $ostype eq 'w2k3' || $ostype eq 'w2k') {
      $winversion = 5;
} elsif($ostype eq 'w2k8' || $ostype eq 'wvista') {
      $winversion = 6;
} elsif ($ostype =~ m/^win(\d+)$/) {
      $winversion = $1;
}

so we can simply do test on windows version with lower or upper version

Hyperv enlightments configuration is centralized
in a new add_hyperv_enlighments sub.

Also disable hyperv with win < 8 + ovmf.

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agoremove format_description from $met_fmt->{model}
Dietmar Maurer [Sat, 5 Nov 2016 15:28:59 +0000 (16:28 +0100)]
remove format_description from $met_fmt->{model}

7 years agoremove unnecessary format_description from cputype
Dietmar Maurer [Sat, 5 Nov 2016 14:56:12 +0000 (15:56 +0100)]
remove unnecessary format_description from cputype

7 years agoregister new standard option 'pve-qm-image-format'
Dietmar Maurer [Thu, 3 Nov 2016 12:21:53 +0000 (13:21 +0100)]
register new standard option 'pve-qm-image-format'

7 years agobump version to 4.0-94
Dietmar Maurer [Thu, 3 Nov 2016 11:54:05 +0000 (12:54 +0100)]
bump version to 4.0-94

7 years agofix add_throttle_desc - pass correct units
Dietmar Maurer [Thu, 3 Nov 2016 11:48:24 +0000 (12:48 +0100)]
fix add_throttle_desc - pass correct units

7 years agobump version to 4.0-93
Dietmar Maurer [Thu, 3 Nov 2016 09:26:51 +0000 (10:26 +0100)]
bump version to 4.0-93

7 years agoClose #1195: support iops/bps_max_length throttling options
Wolfgang Bumiller [Thu, 3 Nov 2016 07:17:28 +0000 (08:17 +0100)]
Close #1195: support iops/bps_max_length throttling options

7 years agofix a perl warning when failing to parse a new drive string
Wolfgang Bumiller [Thu, 3 Nov 2016 07:17:15 +0000 (08:17 +0100)]
fix a perl warning when failing to parse a new drive string

7 years agofix a warning: discard is not a number
Wolfgang Bumiller [Thu, 3 Nov 2016 07:17:14 +0000 (08:17 +0100)]
fix a warning: discard is not a number

7 years agoallow dedicated migration network, bug #1177
Thomas Lamprecht [Mon, 31 Oct 2016 08:42:31 +0000 (09:42 +0100)]
allow dedicated migration network, bug #1177

Without this patch we use the network were the cluster traffic runs
for sending migration traffic. This is not ideal as it may hinder
cluster traffic. Further some users have a powerful network which
would be perfect for migrations, with this patch they can run the
migration traffic over such a network without having the corosync
traffic on the same network.

The network is configurable through /etc/pve/datacenter.cfg which
got a new property, namely migration. migration has two
subproperties: type (replaces the old migration_unsecure property)
and network.

For the case of a network failure or that a VM has to be moved over
another network for arbitrary other reasons I added the
migration_type and migration_network parameters to qm migrate (and
respectively vm_start as this gets used on migration).
They allow overwriting the datacenter.cfg settings.

Fixes bug #1177

Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
7 years agochange default value for cpuunits to 1024
Dietmar Maurer [Thu, 27 Oct 2016 05:23:32 +0000 (07:23 +0200)]
change default value for cpuunits to 1024

Use the same default as containers.

7 years agobump version to 4.0-92
Dietmar Maurer [Fri, 21 Oct 2016 07:14:18 +0000 (09:14 +0200)]
bump version  to 4.0-92

7 years agoClose #351: add more info to backup log
Emmanuel Kasper [Thu, 20 Oct 2016 09:20:13 +0000 (11:20 +0200)]
Close #351: add more info to backup log

Before patch:
INFO: exclude disk 'scsi1' (backup=no)
INFO: skip unused drive 'local:401/vm-401-disk-3.raw' (not included into backup)
INFO: skip unused drive 'local:401/vm-401-disk-1.raw' (not included into backup)

After patch applied:
INFO: include disk 'scsi0' local:401/vm-401-disk-4.qcow2
INFO: exclude disk 'scsi1' local:401/vm-401-disk-2.raw (backup=no)
INFO: include disk 'scsi2' pve4tank:vm-401-disk-1
INFO: skip unused drive 'local:401/vm-401-disk-3.raw' (not included into backup)
INFO: skip unused drive 'local:401/vm-401-disk-1.raw' (not included into backup)

7 years agosnapshot_list: add bash completion for vmid
Dietmar Maurer [Wed, 19 Oct 2016 09:57:53 +0000 (11:57 +0200)]
snapshot_list: add bash completion for vmid

7 years agoadd qm listsnapshot call
Dominik Csapak [Fri, 7 Oct 2016 08:51:13 +0000 (10:51 +0200)]
add qm listsnapshot call

the api call is already there, just not used in qm.pm
copied mostly from pct.pm

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agoFix #1174: remove pve-qm-drive
Wolfgang Bumiller [Tue, 18 Oct 2016 08:38:58 +0000 (10:38 +0200)]
Fix #1174: remove pve-qm-drive

Let 'cdrom' use the pve-qm-ide format, as it's supposed to
be an alias to ide2.
We're not using the 'alias' schema property since the qemu
configs still use a custom parser (due to the
pending-changes system and the filename-to-volume-id
conversion for legacy support) which does not deal with
schema aliases.

7 years agoFix typo
Emmanuel Kasper [Thu, 13 Oct 2016 12:33:28 +0000 (14:33 +0200)]
Fix typo

7 years agoimprove error messages
Wolfgang Bumiller [Mon, 17 Oct 2016 12:49:05 +0000 (14:49 +0200)]
improve error messages

7 years agoenable drive-mirror with iothread for qemu 2.7 v2
Alexandre Derumier [Mon, 17 Oct 2016 10:20:45 +0000 (12:20 +0200)]
enable drive-mirror with iothread for qemu 2.7 v2

changelog : check running qemu binary version

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agoadd get_running_qemu_version
Alexandre Derumier [Mon, 17 Oct 2016 10:20:43 +0000 (12:20 +0200)]
add get_running_qemu_version

return current running qemu process version

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agocpu hotplug : add new cpu hotplug method for qemu 2.7
Alexandre Derumier [Mon, 17 Oct 2016 10:18:57 +0000 (12:18 +0200)]
cpu hotplug : add new cpu hotplug method for qemu 2.7

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agocpu hotplug : add cpu hot-unplug support
Alexandre Derumier [Mon, 17 Oct 2016 10:18:56 +0000 (12:18 +0200)]
cpu hotplug : add cpu hot-unplug support

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agocpu hotplug : add coldplugged cpu to qemu command line
Alexandre Derumier [Mon, 17 Oct 2016 10:18:55 +0000 (12:18 +0200)]
cpu hotplug : add coldplugged cpu to qemu command line

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agocpu hotplug : add print_cpu_device
Alexandre Derumier [Mon, 17 Oct 2016 10:18:54 +0000 (12:18 +0200)]
cpu hotplug : add print_cpu_device

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agonumaX : use cpus option multiple time if cpulist
Alexandre Derumier [Mon, 10 Oct 2016 07:36:55 +0000 (09:36 +0200)]
numaX : use cpus option multiple time if cpulist

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agobump version to 4.0-91
Dietmar Maurer [Fri, 7 Oct 2016 06:21:51 +0000 (08:21 +0200)]
bump version to 4.0-91

7 years agofix #1131: activate volume before copying efidisk
Dominik Csapak [Thu, 6 Oct 2016 09:56:00 +0000 (11:56 +0200)]
fix #1131: activate volume before copying efidisk

before copying the efidisk image to a storage,
we first have to activate the volume

also, add the -n flag to qemu-img convert (prevents
creation of the target volume)

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agobump version to 4.0-90
Dietmar Maurer [Wed, 5 Oct 2016 05:13:52 +0000 (07:13 +0200)]
bump version to 4.0-90

7 years agofix #1111: qm showcmd wrong escape sequence
Wolfgang Link [Wed, 21 Sep 2016 13:14:18 +0000 (15:14 +0200)]
fix #1111: qm showcmd wrong escape sequence

7 years agoAvoid to parse empty property string.
Wolfgang Link [Wed, 28 Sep 2016 07:19:19 +0000 (09:19 +0200)]
Avoid to parse empty property string.

Check if string exists before you try to parse it. Remove unnecessary
eval.

7 years agofix manual page, bump version to 4.0-89
Dietmar Maurer [Thu, 29 Sep 2016 10:21:41 +0000 (12:21 +0200)]
fix manual page, bump version to 4.0-89

7 years agobump version to 4.0-88
Dietmar Maurer [Fri, 16 Sep 2016 05:50:02 +0000 (07:50 +0200)]
bump version to 4.0-88

7 years agorestore: better error handling for vdisk deletion
Fabian Grünbichler [Thu, 15 Sep 2016 11:24:35 +0000 (13:24 +0200)]
restore: better error handling for vdisk deletion

when restoring into an existing VM, we don't want to die
half-way through because we can't delete one of the existing
volumes. instead, warn about the deletion failure, but
continue anyway. the not deleted disk is then added as
unused automatically.

7 years agoforbid migration of template with local base image
Fabian Grünbichler [Thu, 15 Sep 2016 11:23:56 +0000 (13:23 +0200)]
forbid migration of template with local base image

7 years agoforbid restore into existing template
Fabian Grünbichler [Thu, 15 Sep 2016 11:23:12 +0000 (13:23 +0200)]
forbid restore into existing template

7 years agobump version to 4.0-87
Dietmar Maurer [Thu, 8 Sep 2016 10:34:41 +0000 (12:34 +0200)]
bump version to 4.0-87

7 years agoadd seabios bootsplash and use it
Dominik Csapak [Thu, 8 Sep 2016 07:25:57 +0000 (09:25 +0200)]
add seabios bootsplash and use it

this adds a bootsplash image in /usr/share/qemu-server
and if this file exists, use it for seabios

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agouse efidisk0 for efivars
Dominik Csapak [Thu, 8 Sep 2016 09:03:01 +0000 (11:03 +0200)]
use efidisk0 for efivars

if efidisk0 is defined, use it as a efivars disk,
to permanently store efivars (such as boot options)

we check if the files exist, and act accordingly

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agocopy efivars disk on create
Dominik Csapak [Thu, 8 Sep 2016 09:03:00 +0000 (11:03 +0200)]
copy efivars disk on create

when we create the efidisk0 over the api,
we discard the size, and create a 128kbyte vdisk
and copy it there with qemu-img convert

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agoadd efidisk0 to config
Dominik Csapak [Thu, 8 Sep 2016 09:02:59 +0000 (11:02 +0200)]
add efidisk0 to config

just a simple disk (only size, format and volid) for
efivars disk

also do not add it to command line in foreach_drive

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agobump version to 4.0-86
Dietmar Maurer [Mon, 29 Aug 2016 08:11:40 +0000 (10:11 +0200)]
bump version to 4.0-86

7 years agohostpci: bring back multifunction pass-through shortcut
Wolfgang Bumiller [Wed, 24 Aug 2016 07:55:49 +0000 (09:55 +0200)]
hostpci: bring back multifunction pass-through shortcut

7 years agodisable drive-mirror when iothread is enabled
Alexandre Derumier [Sat, 20 Aug 2016 08:02:29 +0000 (10:02 +0200)]
disable drive-mirror when iothread is enabled

drive-mirror is not working with qemu 2.6 when iothread is enabled.

with virtio-blk : mirror is working, but block-job-completed crash the vm
with virtio-scsi : mirror hang at start.

This should be fixed in qemu 2.7

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agomemory: hugepages: map numa node IDs to host and guest correctly
Wolfgang Bumiller [Fri, 29 Jul 2016 11:26:29 +0000 (13:26 +0200)]
memory: hugepages: map numa node IDs to host and guest correctly

foreach_dimm() provides a guest numa node index, when used
in conjunction with the guest-to-host numa node topology
mapping one has to make sure that the correct host-side
indices are used.

This covers situations where the user defines a numaX with
hostnodes=Y with Y != X.

For instance:
  cores: 2
  hotplug: disk,network,cpu,memory
  hugepages: 2
  memory: 2048
  numa: 1
  numa1: memory=512,hostnodes=0,cpus=0-1,policy=bind
  numa2: memory=512,hostnodes=0,cpus=2-3,policy=bind

Both numa IDs 1 and 2 passed by foreach_dimm() have to be
mapped to host node 0.

Note that this also reverses the foreach_reverse_dimm() numa
node numbering as the current code, while walking sizes
backwards, walked the numa IDs inside each size forward,
which makes more sense. (Memory hot-unplug is still working
with this.)

7 years agohugepages: use hostnodes value as numanode for topology
Alexandre Derumier [Fri, 29 Jul 2016 03:17:34 +0000 (05:17 +0200)]
hugepages: use hostnodes value as numanode for topology

also disallow more than 1 hostnode value

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agoqemu-img convert : use default cache=unsafe instead writeback
Alexandre Derumier [Mon, 1 Aug 2016 06:51:00 +0000 (08:51 +0200)]
qemu-img convert : use default cache=unsafe instead writeback

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agohugepages: check if numa node exist only if hugepages are enabled
Alexandre Derumier [Fri, 29 Jul 2016 03:17:33 +0000 (05:17 +0200)]
hugepages: check if numa node exist only if hugepages are enabled

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
7 years agofix verbose qm status output
Dominik Csapak [Mon, 18 Jul 2016 08:50:31 +0000 (10:50 +0200)]
fix verbose qm status output

we did not check if some values were hash refs in
the verbose output.

this patch adds a recursive hash print sub and uses it

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agoFix #1057: make protection a fast-plug option
Wolfgang Bumiller [Tue, 19 Jul 2016 07:17:36 +0000 (09:17 +0200)]
Fix #1057: make protection a fast-plug option

Otherwise you need to shutdown a VM to disable protection,
which is inconvenient for a few tasks such as for instance
deleting an unused disk.

7 years agoadd lock check for move_disk API call
Fabian Grünbichler [Fri, 22 Jul 2016 05:53:53 +0000 (07:53 +0200)]
add lock check for move_disk API call

this API call changes the config quite drastically, and as
such should not be possible while an operation that holds a
lock is ongoing (e.g., migration, backup, snapshot).

7 years agoonly activate the newly created volume
Fabian Grünbichler [Wed, 13 Jul 2016 10:44:13 +0000 (12:44 +0200)]
only activate the newly created volume

instead of the whole list.

7 years agodeactivate new volumes after clone to other node
Fabian Grünbichler [Wed, 13 Jul 2016 10:44:12 +0000 (12:44 +0200)]
deactivate new volumes after clone to other node

this might otherwise lead to volumes activated on the
source and target node, which is problematic for at least
LVM and Ceph.

7 years agopass datacenter.cfg's mac_prefix to random_ether_addr
Wolfgang Bumiller [Wed, 13 Jul 2016 14:25:44 +0000 (16:25 +0200)]
pass datacenter.cfg's mac_prefix to random_ether_addr

7 years agobump version to 4.0-85
Wolfgang Bumiller [Mon, 11 Jul 2016 12:49:03 +0000 (14:49 +0200)]
bump version to 4.0-85

7 years agoFix #1051: typo: vpcus -> vcpus
Wolfgang Bumiller [Mon, 11 Jul 2016 12:44:24 +0000 (14:44 +0200)]
Fix #1051: typo: vpcus -> vcpus

Otherwise only root can change this setting.

7 years agobump version to 4.0-84
Wolfgang Bumiller [Mon, 11 Jul 2016 09:53:46 +0000 (11:53 +0200)]
bump version to 4.0-84

7 years agofix 1046: add non-snapshotted disks as unused
Fabian Grünbichler [Mon, 4 Jul 2016 11:04:14 +0000 (13:04 +0200)]
fix 1046: add non-snapshotted disks as unused

7 years agobuildsys: only deb should be phony; start by removing debs
Wolfgang Bumiller [Mon, 4 Jul 2016 06:33:57 +0000 (08:33 +0200)]
buildsys: only deb should be phony; start by removing debs

7 years agodisable usb hotplug for now
Dominik Csapak [Mon, 4 Jul 2016 13:24:20 +0000 (15:24 +0200)]
disable usb hotplug for now

we have a few problems with hotplug at the moment:
qemu may add usb hubs when adding usb devices but fails to remove them
when removing the usb device (this is a qemu bug)

also when starting a guest with a usb device we add ehci and uchi
controllers, which we cannot hot unplug

with those devices, it is impossible to live migrate the guest
to another host, meaning even if you remove all usb devices,
the migrate fails

so we deactivate usb hotplugging for now

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agoremove old move disk snapshot check
Fabian Grünbichler [Fri, 1 Jul 2016 12:18:32 +0000 (14:18 +0200)]
remove old move disk snapshot check

this is no longer needed, if the moved disk was referenced
by a snapshot we have already died early, and the config is
locked.

7 years agofix #1040: warn early about moving a snapshotted disk
Fabian Grünbichler [Fri, 1 Jul 2016 12:18:31 +0000 (14:18 +0200)]
fix #1040: warn early about moving a snapshotted disk

die if "delete source disk" is enabled, because a disk that
is referenced by snapshots must not be deleted.

7 years agocollect errors from all local volumes
Fabian Grünbichler [Fri, 24 Jun 2016 07:46:37 +0000 (09:46 +0200)]
collect errors from all local volumes

and then die with more meaningful output, instead of on the
first encountered error.

7 years agoadd comments and rename volhash
Fabian Grünbichler [Fri, 24 Jun 2016 07:46:36 +0000 (09:46 +0200)]
add comments and rename volhash

7 years agoswitch order of disk checks
Fabian Grünbichler [Fri, 24 Jun 2016 07:46:35 +0000 (09:46 +0200)]
switch order of disk checks

to make log message more meaningful.
'storage' < 'snapshot' < 'config'

7 years agocollect and log origin of found local volumes
Fabian Grünbichler [Fri, 24 Jun 2016 07:46:34 +0000 (09:46 +0200)]
collect and log origin of found local volumes

just knowing that local disks prevent a migration is not
very helpful, so be a bit more verbose here.

7 years agonew make upload target
Wolfgang Bumiller [Tue, 28 Jun 2016 07:16:11 +0000 (09:16 +0200)]
new make upload target

7 years agofix dependency for upload target
Dietmar Maurer [Tue, 28 Jun 2016 06:20:50 +0000 (08:20 +0200)]
fix dependency for upload target

7 years agoupdate changelog
Dietmar Maurer [Tue, 28 Jun 2016 06:18:23 +0000 (08:18 +0200)]
update changelog

7 years agofix #146 add name to backup log.
Wolfgang Link [Mon, 27 Jun 2016 12:37:14 +0000 (14:37 +0200)]
fix #146 add name to backup log.

7 years agobump version to 4.0-83
Wolfgang Bumiller [Wed, 22 Jun 2016 09:33:13 +0000 (11:33 +0200)]
bump version to 4.0-83

7 years agofix syntax in pve-q35.cfg
Dominik Csapak [Tue, 14 Jun 2016 08:50:41 +0000 (10:50 +0200)]
fix syntax in pve-q35.cfg

there was a wrong } instead of a ]

either it worked or nobody noticed,
but this patch fixes it

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agoimplement usb hotplug
Dominik Csapak [Tue, 14 Jun 2016 08:50:40 +0000 (10:50 +0200)]
implement usb hotplug

this patch introduces working usb hotplugging

you can now add a usb device while a vm is running

this does not work with spice at the moment, only
with usb passthrough

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agoadd qemu_usb_hotplug
Dominik Csapak [Tue, 14 Jun 2016 08:50:39 +0000 (10:50 +0200)]
add qemu_usb_hotplug

this adds a function for hotplugging usb devices

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agoget usbdevices in vm_devices_list
Dominik Csapak [Tue, 14 Jun 2016 08:50:38 +0000 (10:50 +0200)]
get usbdevices in vm_devices_list

since usb devices do not have their own
"query" command in qmp, we have to use
qom-list /machines/peripheral

which essentially gets a list of peripheral devices of
the vm

there we only get the usb devices

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agouse the refactored usb functions
Dominik Csapak [Tue, 14 Jun 2016 08:50:37 +0000 (10:50 +0200)]
use the refactored usb functions

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agorefactor usb functions out of QemuServer.pm
Dominik Csapak [Tue, 14 Jun 2016 08:50:36 +0000 (10:50 +0200)]
refactor usb functions out of QemuServer.pm

this moves most of the usb functionality into its own
module, making the QemuServer.pm smaller

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agorefactor PCI into own file
Dominik Csapak [Tue, 14 Jun 2016 08:50:35 +0000 (10:50 +0200)]
refactor PCI into own file

to reduce QemuServer.pm size
also move the $device hash out of any function

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agoadd hugepages option
Alexandre Derumier [Wed, 15 Jun 2016 03:04:02 +0000 (05:04 +0200)]
add hugepages option

vm configuration
----------------
hugepages: (any|2|1024)

any: we'll try to allocate 1GB hugepage if possible, if not we use 2MB hugepage
2: we want to use 2MB hugepage
1024: we want to use 1GB hugepage. (memory need to be multiple of 1GB in this case)

optionnal host configuration for 1GB hugepages
----------------------------------------------
1GB hugepages can be allocated at boot if user want it.
hugepages need to be contiguous, so sometime it's not possible to reserve them on the fly

/etc/default/grub : GRUB_CMDLINE_LINUX_DEFAULT="quiet hugepagesz=1G hugepages=x"

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Acked-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
7 years agobump version to 4.0-82
Dietmar Maurer [Mon, 20 Jun 2016 09:49:04 +0000 (11:49 +0200)]
bump version to 4.0-82

7 years agoadd dependency on DBus
Fabian Grünbichler [Mon, 20 Jun 2016 09:32:13 +0000 (11:32 +0200)]
add dependency on DBus

required for our systemd-run replacement (systemd itself
only recommends DBus and falls back to an internal IPC
mechanism if unavailable)

7 years agodrop unncessary cdromhash
Fabian Grünbichler [Fri, 17 Jun 2016 14:12:29 +0000 (16:12 +0200)]
drop unncessary cdromhash

7 years agouse foreach_drive instead of foreach_volid
Fabian Grünbichler [Fri, 17 Jun 2016 14:12:28 +0000 (16:12 +0200)]
use foreach_drive instead of foreach_volid

foreach_volid recurses over snapshots as well, resulting in
lots of repeated checks (especially for VMs with lots of
snapshots and disks).

a potential vmstate volume must be checked explicitly,
because foreach_drive does not care about those.

7 years agofix whitespace/indent
Fabian Grünbichler [Fri, 17 Jun 2016 14:12:27 +0000 (16:12 +0200)]
fix whitespace/indent

7 years agodon't repeat storage check for each volid
Fabian Grünbichler [Fri, 17 Jun 2016 14:12:26 +0000 (16:12 +0200)]
don't repeat storage check for each volid

7 years agoadd @param to foreach_drive
Fabian Grünbichler [Fri, 17 Jun 2016 14:12:25 +0000 (16:12 +0200)]
add @param to foreach_drive

7 years agobump version to 4.0-81
Dietmar Maurer [Thu, 16 Jun 2016 10:11:13 +0000 (12:11 +0200)]
bump version to 4.0-81

7 years agofix perl scope issues
Wolfgang Link [Thu, 16 Jun 2016 09:22:30 +0000 (11:22 +0200)]
fix perl scope issues

Add parameter array to foreach_volid to use is in the functions.
correct typos.

7 years agoadd check for snapshots at migration
Dietmar Maurer [Thu, 16 Jun 2016 08:21:57 +0000 (10:21 +0200)]
add check for snapshots at migration

We cannot migrate snapshots on local disks, for example lvmthin snapshots.

7 years agoAdd LVM and LVMThin to QemuMigration
Wolfgang Link [Mon, 13 Jun 2016 13:54:14 +0000 (15:54 +0200)]
Add LVM and LVMThin to QemuMigration

Offline migration on LVM and LVMThin are possible offline.

7 years agoupdate changelog
Dietmar Maurer [Fri, 10 Jun 2016 08:39:59 +0000 (10:39 +0200)]
update changelog

7 years agofix unnecessary root check for unused disks
Dominik Csapak [Fri, 10 Jun 2016 07:26:14 +0000 (09:26 +0200)]
fix unnecessary root check for unused disks

we needed root@pam rights to remove an unused disk
from a vm (instead of the correct Storage rights)

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
7 years agobump version to 4.0-80
Dietmar Maurer [Thu, 9 Jun 2016 16:13:24 +0000 (18:13 +0200)]
bump version to 4.0-80

7 years agosplit old style pipe open call
Wolfgang Bumiller [Thu, 9 Jun 2016 14:54:46 +0000 (16:54 +0200)]
split old style pipe open call