proxmox-mirrors/qemu-server
1
0
Fork 0
mirror of https://git.proxmox.com/git/qemu-server synced 2025-05-05 11:12:56 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
776 Commits 7 Branches 0 Tags 12 MiB
c427973b46
Commit Graph

398 Commits

Author SHA1 Message Date
Dietmar Maurer
c427973b46 implement trivial hotplug 
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2015-01-07 06:42:27 +01:00
Dietmar Maurer
055d554dfa vm_start: apply pending changes 
I move related helper methods into PVE::QemuServer.

Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2015-01-07 06:42:17 +01:00
Dietmar Maurer
ef824322a0 write_vm_config : write pending change 
example:

$conf->{pending}->{virtio1}
$conf->{pending}->{delete} = "net0,net1"

[PENDING]
virtio1: ...
delete: net0,net1

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2015-01-07 06:42:00 +01:00
Dietmar Maurer
e297c4903a parse_vm_config: only allow 'delete' inside [PENDING] 
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2015-01-07 06:41:46 +01:00
Dietmar Maurer
0d732d1615 parse_vm_config: correctly handle $descr 
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2015-01-07 06:41:36 +01:00
Alexandre Derumier
eab09f4e11 parse_vm_config : parse pending changes 
example:

[PENDING]
virtio1:...
delete:net0,net1

$conf->{pending}->{virtio1}
$conf->{pending}->{del}->{net0}
$conf->{pending}->{del}->{net1}

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2015-01-07 06:41:24 +01:00
Stefan Priebe
6fab033c3c this commit: 
commit 1c0c1c17b0
Author: Wolfgang Link <wolfgang@linksystems.org>
Date:   Wed Nov 26 11:11:40 2014 +0100

    shutdown by Qemu Guest Agent if the agent flag in the config is set

    Important: "guest-shutdown" returns only by error a message.

    Signed-off-by: Wolfgang Link <wolfgang@linksystems.org>

breaks live migration as it always tries to load the vm config - even in case of $nocheck. Also it double loads the config ($conf && $config)
Signed-off-by: Stefan Priebe <s.priebe@profihost.ag>
 2014-12-22 17:19:39 +01:00
Dietmar Maurer
67fb9de608 white space cleanups 2014-12-04 13:07:59 +01:00
Alexandre Derumier
2ed5d5724c add custom numa topology support 
numaX: cpus=<id[-id],memory=<mb>[[,hostnodes=<id[-id]>][,policy=<preferred|bind|interleave>]]

example:
-------
sockets:4
cores:2
memory:4096
numa: 1
numa0: cpus=0-1,memory=1024,hostnodes=0-1,policy=interleave
numa1: cpus=2-3,memory=3072,hostnodes=2,policy=bind

qemu command line
-----------------
-object memory-backend-ram,size=1024M,policy=interleave,host-nodes=0-1,id=ram-node0
-numa node,nodeid=0,cpus=0-1,memdev=ram-node0

-object memory-backend-ram,size=3072M,policy=bind,host-nodes=2,id=ram-node1
-numa node,nodeid=1,cpus=2-3,memdev=ram-node1

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-12-04 13:02:47 +01:00
Alexandre Derumier
8a010eaeb1 add numa options v3 
This enable numa support inside the guest, and share the memory and cores across the sockets numa nodes.

numa: 0|1

example:
-------
sockets:2
cores:2
memory:4096
numa: 1

qemu command line
-----------------
-object memory-backend-ram,size=2048,id=ram-node0
-numa node,nodeid=0,cpus=0-1,memdev=ram-node0
-object memory-backend-ram,size=2048,id=ram-node1
-numa node,nodeid=1,cpus=2-3,memdev=ram-node1

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-12-04 13:01:47 +01:00
Wolfgang Link
af9110ddd7 Fix in PVE::QemuServer::snapshot_create 
remove the freezefs flag.
If Qemu Guest Agent flag is set in config the vm filesystem will always be frozen,
unless we save RAM.

also remove param freezefs in PVE::API2 snapshot,
because there is no use for it.

Signed-off-by: Wolfgang Link <wolfgang@linksystems.org>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2014-12-04 12:30:35 +01:00
Dietmar Maurer
108864e591 snapshot_create: fix bug from prev. commit, consider $freezefs parameter 2014-12-02 13:42:51 +01:00
Wolfgang Link
65994ad73c snapshot_create: use guest-fsfreeze-freeze if possible 
Signed-off-by: Wolfgang Link <wolfgang@linksystems.org>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2014-12-02 13:11:42 +01:00
Dietmar Maurer
6f70864360 drive-mirror: avoid division by zero bug 2014-12-01 09:50:12 +01:00
Dietmar Maurer
c8125172c7 qmpclient: improve error handling 2014-12-01 09:50:12 +01:00
Dietmar Maurer
2ea5450311 simplify code by using vm_qmp_command() directly 2014-12-01 09:50:12 +01:00
Dietmar Maurer
7a6c215042 qmpclient: code cleanup, execute qga and qmp in parallel 2014-12-01 09:50:12 +01:00
Wolfgang Link
1c0c1c17b0 shutdown by Qemu Guest Agent if the agent flag in the config is set 
Important: "guest-shutdown" returns only by error a message.

Signed-off-by: Wolfgang Link <wolfgang@linksystems.org>
 2014-12-01 09:50:12 +01:00
Wolfgang Link
c5a07de5af qmpclient: now if the QMP command starts with guest-+ , it will bind dynamicly to the VMID.qga socket 
Signed-off-by: Wolfgang Link <wolfgang@linksystems.org>
 2014-12-01 09:50:12 +01:00
Alexandre Derumier
693d12a2c0 add vm_qga_command 
and reuse vm_qmp_command && qmp_socket with $qga param

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-12-01 09:50:11 +01:00
Dietmar Maurer
f34ebd5259 code cleanups 2014-12-01 09:50:11 +01:00
Alexandre Derumier
2c9e8036f5 savevm-end : wait that savevm is finished 
savevm-end is async, we need to wait that savevm is finished, before continue.

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-12-01 09:50:11 +01:00
Dietmar Maurer
1b0b51ed94 serial: allow to pass arbitrary device names 2014-12-01 09:50:11 +01:00
Wolfgang Link
7626772856 Add check if host has enough real CPUs for starting VM 
To prevent a Qemu CPU emualtion!

Signed-off-by: Wolfgang Link <wolfgang@linksystems.org>
 2014-11-17 11:24:15 +01:00
Dietmar Maurer
883839209f qemu_drive_mirror: cleanup, avoid code duplication 2014-11-10 08:18:39 +01:00
Dietmar Maurer
08ac653fcf qemu_drive_mirror: raise exception if we cannot parse $dst_volid 2014-11-10 07:55:09 +01:00
Dietmar Maurer
b467f79a50 delete trailing whitespace 2014-11-10 06:31:08 +01:00
Alexandre Derumier
21ccdb506a block-job-complete : retry if block job cannot be complete 
Even if we check the busy flag, we can have sometime race condition if new write
are coming between the query-block-job and the block-job-complete.

block-job-complete throw an error "The active block job for device '%(name)' cannot be completed"

we just need to retry in this case.

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-11-10 06:26:25 +01:00
Dietmar Maurer
bcc87408c0 cleanup: do not entry wait loop if block-job-cancel() fails. 2014-11-07 15:31:56 +01:00
Alexandre Derumier
ab6ecffe53 drive-mirror : wait that busy eq false before block-job-complete 
When the drive-mirror is at 100%, and write occurs, the busy flag can change from false->true

- 100% no new writes

transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: false
transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: false
transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: false

- 100% new writes

transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: true
transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: true
transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: true

- 100% no new writes

transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: false
transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: false
transferred: 1073741824 bytes remaining: 0 bytes total: 1073741824 bytes progression: 100.00 % busy: false

So, we need to check that busy is false before doing the block-job-complete.

Also, we force the vm to pause, if it's busy for more than 120s when drive-mirror is at 100%

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-11-07 15:28:13 +01:00
Alexandre Derumier
5fe03fd8ee block-job-cancel : wait that jobs is really finished 
block-job-cancel is async, we need to check that job is really finished

before try to free the volume

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-11-07 15:26:21 +01:00
Alexandre Derumier
f1e053055c enable write zeroes optimisations 
This enable write zeroes optimisation inside guest.

(I have tested them with qcow2,raw (ext4|xfs) , zfs|iscsi and nfs too.

No impact if the block driver don't support it yet (like rbd)

https://lists.nongnu.org/archive/html/qemu-devel/2014-04/msg00009.html

> a) mkfs.ext4 -E lazy_itable_init=0,lazy_journal_init=0 /dev/vdX
>
> QCOW2         [off]     [on]     [unmap]
> -----
> runtime:       14secs    1.1secs  1.1secs
> filesize:      937M      18M      18M
>
> iSCSI         [off]     [on]     [unmap]
> ----
> runtime:       9.3s      0.9s     0.9s
>
> b) dd if=/dev/zero of=/dev/vdX bs=1M oflag=direct
>
> QCOW2         [off]     [on]     [unmap]
> -----
> runtime:       246secs   18secs   18secs
> filesize:      51G       192K     192K
> throughput:    203M/s    2.3G/s   2.3G/s
>
> iSCSI*        [off]     [on]     [unmap]
> ----
> runtime:       8mins     45secs   33secs
> throughput:    106M/s    1.2G/s   1.6G/s
> allocated:     100%      100%     0%

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-10-13 10:34:41 +02:00
Alexandre Derumier
7183bd9a8f add iothread/dataplane support 
new config option:

iothread: 1|0

This enable iothread/dataplane support, to improve io performance on fast storages

Currently block jobs don't work yet, it's planned for qemu 2.2.
So it's better to not expose yet this option in gui.

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-10-10 10:27:41 +02:00
Alexandre Derumier
f78cc802cf vm_devices_list : also list block devices 
This allow scsi disk to be plug|unplug

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-09-01 11:33:54 +02:00
Dietmar Maurer
7946e0fa42 Currently,if we don't have a "machine" option in running config, and we take a vmstate snapshot 
the machine option is write in the snapshot (ok), but also in the running config (bad)

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
 2014-08-29 11:27:27 +02:00
Alexandre Derumier
6c33c345ef allow hotplug of virtio-scsi disks 
It was an old protection, it's working fine now

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-08-29 08:18:08 +02:00
Dietmar Maurer
cd11416f7a clone_vm: auto generate new uuid 2014-08-26 09:20:09 +02:00
Alexandre Derumier
3aefd6fdf9 add Broadwell cpu model 
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-08-20 12:20:18 +02:00
Alexandre Derumier
d2da6d9b8e generate spice devices cmd line after pci bridge 
we should push to $devices array instead $cmd array,

because pci bridges need to be create before spice devices

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-08-13 06:14:03 +02:00
Alexandre Derumier
5cffb2d2f2 bump max hostpci to 4 
user need to passthough 3 devices here
http://forum.proxmox.com/threads/19134-KVM-passtrought-PciExpress-3-card

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-08-06 09:40:32 +02:00
Alexandre Derumier
137483c053 vga=none if x-vga passthrough is enabled 
we need to disable virtual vga card if we passthrough a physical gpu

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-08-06 09:39:38 +02:00
Alexandre Derumier
9040435417 disable kvm cpu signature if x-vga is enabled 
see
http://git.qemu.org/?p=qemu.git;a=commit;h=f522d2acc549dd11f495048330aa5f3f424a7dfa

last nvdia drivers don't install in kvm machine if they detect kvm signature.

This patch hide kvm signature in cpuflags (but don't disable kvm)

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-07-24 06:51:55 +02:00
Kamil Trzciński
8f3e88af00 Reset device only if has_fl_reset is defined 
Signed-off-by: Kamil Trzciński <ayufan@ayufan.eu>
 2014-07-23 06:11:15 +02:00
Stefan Priebe
8fd882a42b snapshot_delete: reorder parent "connection" after prepare 
Signed-off-by: Stefan Priebe <s.priebe@profihost.ag>
 2014-07-17 09:23:40 +02:00
Stefan Priebe
258e646c57 pass snapname to PVE::Storage::vdisk_clone 
Signed-off-by: Stefan Priebe <s.priebe@profihost.ag>
 2014-07-17 09:21:37 +02:00
Dietmar Maurer
2796e7d542 new option smbios1: specify SMBIOS type 1 fields (uuid, ...) 2014-06-26 11:12:25 +02:00
Alexandre Derumier
b1f72af6d4 add pci multifunction unbind support 
we need to unbind each functions of a multifunction pci device

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-06-25 09:30:01 +02:00
Alexandre Derumier
4543ecf07f add pci multifunction support 
multifunction device should be define without the .function

hostpci0: 00:00

example
-------
if 00:00.0
   00:00.1
   00:00.2

exists,

then we generate the multifunction devices

-device (pci-assign|vfio-pci),host=00:00.0,id=hostpci0.0,bus=...,addr=0x0.0,multifunction=on
-device (pci-assign|vfio-pci),host=00:00.1,id=hostpci0.1,bus=...,addr=0x0.1
-device (pci-assign|vfio-pci),host=00:00.2,id=hostpci0.2,bus=...,addr=0x0.2

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-06-25 09:29:53 +02:00
Alexandre Derumier
2e3b7e2a3a add pcie and x-vga passthrough 
hostpci0:  .....,x-vga=on,pcie=1

x-vga require kernel 3.10 with vfio-vga support enable
if x-vga=on, we force vfio-pci device

pcie=1 choose the pciexpress bus (need q35 machine model)

Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
 2014-06-25 09:29:10 +02:00
Stefan Priebe
000fc0a254 added support for vfio-pci passthrough 
Signed-off-by: Stefan Priebe <s.priebe@profihost.ag>
 2014-06-18 07:03:09 +02:00