]>
git.proxmox.com Git - qemu-server.git/log
Alexandre Derumier [Thu, 6 Sep 2012 08:33:39 +0000 (10:33 +0200)]
add qga_unfreezefs
empty for now, need to implement
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Thu, 6 Sep 2012 08:33:38 +0000 (10:33 +0200)]
add qga_freezefs
empty for now, need to implement
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Thu, 6 Sep 2012 08:33:37 +0000 (10:33 +0200)]
add qemu_snapshot_end
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Thu, 6 Sep 2012 08:33:36 +0000 (10:33 +0200)]
add qemu_snapshot_start
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Thu, 6 Sep 2012 08:33:35 +0000 (10:33 +0200)]
add qm snapshot command
qm snapshot <vmid> <action> <snapshotname> --vmstate --freezefs
action : create|rollback|delete
vmstate : save the vmstate
freezefs: freezefs with qemu-ga
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Thu, 6 Sep 2012 08:33:34 +0000 (10:33 +0200)]
add qemu_volume_snapshot
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Tue, 4 Sep 2012 04:46:53 +0000 (06:46 +0200)]
set correct default for new 'agent' option
We do not enable that by default for now.
Dietmar Maurer [Tue, 4 Sep 2012 04:31:44 +0000 (06:31 +0200)]
rename option 'qga' to 'agent'
Alexandre Derumier [Mon, 3 Sep 2012 07:51:08 +0000 (09:51 +0200)]
add qemu guest agent support
this add chardev and virtio-serial devices for communication with guest agent.
vmconfig option :
qga: 1
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Fri, 31 Aug 2012 09:04:52 +0000 (11:04 +0200)]
bump version to 2.0-55
Dietmar Maurer [Fri, 31 Aug 2012 09:02:47 +0000 (11:02 +0200)]
migrate: disable xbzrle for now.
This is not stable, and sometimes cause endless migration (migration never stops).
Dietmar Maurer [Thu, 30 Aug 2012 10:18:44 +0000 (12:18 +0200)]
bump version to 2.0-54
Dietmar Maurer [Thu, 30 Aug 2012 10:15:07 +0000 (12:15 +0200)]
avoid warning about uninitialized value
Dietmar Maurer [Thu, 30 Aug 2012 07:28:24 +0000 (09:28 +0200)]
migrate: tolerate query-migrate errors
Dietmar Maurer [Wed, 29 Aug 2012 11:42:41 +0000 (13:42 +0200)]
use mbps instead of bps
The new parser automatically converts old bps to mbps.
Dietmar Maurer [Wed, 29 Aug 2012 11:01:52 +0000 (13:01 +0200)]
cleanup: split out parse_size
Dietmar Maurer [Wed, 29 Aug 2012 07:19:21 +0000 (09:19 +0200)]
allow to specify sata as boot disk
Dietmar Maurer [Wed, 29 Aug 2012 06:00:26 +0000 (08:00 +0200)]
fix changelog
Alexandre Derumier [Tue, 28 Aug 2012 10:46:08 +0000 (12:46 +0200)]
livemigrate : activate xbzrle cache
This help migrate for vm with of lot of memory access (like database)
live migration tests working:
kvm 1.2 -> kvm 1.2 (xbzrle set on both side)
kvm 1.1 -> kvm 1.2 (xbzrle on target)
kvm 1.1 -> kvm 1.1 (xbzrle not set, qmp command try to set xbzrle but fail)
failing migration
kvm 1.2 -> kvm 1.1 fail, but this is expected.
I tested with a memory benchmark running on the vm with 4GB ram
without xbzrle : migration take 10min, with many network hang
with xbzrle : migration take 1min, no hang
I display xbzrle counters for debug purpose, we can remove them later
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Tue, 28 Aug 2012 10:46:07 +0000 (12:46 +0200)]
use scsi-generic by default with libiscsi
This add scsi passthrough with libiscsi
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Tue, 28 Aug 2012 10:46:06 +0000 (12:46 +0200)]
scsihw: add megasas controller
It could be useful for windows7 and windows2008 , as they are no more classic lsi driver for these platform
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Mon, 27 Aug 2012 11:45:50 +0000 (13:45 +0200)]
bump version to 2.0-53
Dietmar Maurer [Mon, 27 Aug 2012 11:41:24 +0000 (13:41 +0200)]
fix bug 247: retry qmp open
Dietmar Maurer [Mon, 27 Aug 2012 11:13:36 +0000 (13:13 +0200)]
set default qmp timout to 3 seconds
And allow to pass timeout paramater to vm_qmp_command().
Alexandre Derumier [Fri, 24 Aug 2012 06:57:45 +0000 (08:57 +0200)]
remove qmp socket file on vm_stop_cleanup
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Thu, 23 Aug 2012 08:34:12 +0000 (10:34 +0200)]
bump version to 2.1-52
Dietmar Maurer [Thu, 23 Aug 2012 08:28:41 +0000 (10:28 +0200)]
we call vm_stop to target host,
to be sure that kvm process is killed (but it should kill itself),
and deactivate volumes
I slightly modified this patch (orig. from Alexandre) so that it apply cleanly.
Alexandre Derumier [Tue, 21 Aug 2012 10:21:54 +0000 (12:21 +0200)]
live migration: reduce sleep when remaining memory is low
Reduce sleep to 0.3s when remaining memory is lower than the average transfert in 1 iteration.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Thu, 23 Aug 2012 05:36:48 +0000 (07:36 +0200)]
cleanup: detete trailing whitespace
Alexandre Derumier [Tue, 21 Aug 2012 10:21:53 +0000 (12:21 +0200)]
put target vm in singlestep mode and resume it only when config is moved
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Thu, 23 Aug 2012 05:26:41 +0000 (07:26 +0200)]
pve-bridge: use enviroment variable PVE_MIGRATED_FROM
We use environment variable to pass additional information the
the network script.
Alexandre Derumier [Tue, 21 Aug 2012 10:21:51 +0000 (12:21 +0200)]
add migratedfrom param to start vm with conf file an another node
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Tue, 21 Aug 2012 10:21:50 +0000 (12:21 +0200)]
implement phase2_cleanup
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Tue, 21 Aug 2012 10:21:49 +0000 (12:21 +0200)]
move config file in phase3, when live migration is finished
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Tue, 21 Aug 2012 09:39:17 +0000 (11:39 +0200)]
fix bug #236: use correct shutdown timeout
Dietmar Maurer [Tue, 21 Aug 2012 08:07:09 +0000 (10:07 +0200)]
bump version to 2.0-51
Dietmar Maurer [Tue, 21 Aug 2012 08:03:17 +0000 (10:03 +0200)]
allow only 16 virtio devices
This is the same number we have in 1.9. In future, most users will uses virtio-scsi, so there is no real need to support more devices here.
Alexandre Derumier [Mon, 20 Aug 2012 11:47:56 +0000 (13:47 +0200)]
print_pci_addr : check if $bridges is initialized
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 20 Aug 2012 11:06:59 +0000 (13:06 +0200)]
add bridge hotplug
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Mon, 20 Aug 2012 10:13:15 +0000 (12:13 +0200)]
fix description for max number of devices
Alexandre Derumier [Mon, 20 Aug 2012 09:10:24 +0000 (11:10 +0200)]
add pci-bridge support and bump MAX_NET and MAX_VIRTIO to 32 devices
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Mon, 20 Aug 2012 07:23:00 +0000 (09:23 +0200)]
implement 'qm rescan' to update disk sizes and unused disk info
Dietmar Maurer [Mon, 20 Aug 2012 05:14:19 +0000 (07:14 +0200)]
removed outdated TODO file
I guess it is better to have no TODO, instead of a file with outdated information.
Dietmar Maurer [Fri, 17 Aug 2012 08:34:39 +0000 (10:34 +0200)]
fix bug 242: re-add old monitor code
Dietmar Maurer [Tue, 14 Aug 2012 04:56:00 +0000 (06:56 +0200)]
bump version to 2.0-48
Dietmar Maurer [Wed, 8 Aug 2012 07:25:54 +0000 (09:25 +0200)]
fix for resize: final cleanups
Dietmar Maurer [Wed, 8 Aug 2012 06:26:58 +0000 (08:26 +0200)]
fix for resize: only allow to resize one disk.
Dietmar Maurer [Wed, 8 Aug 2012 05:36:51 +0000 (07:36 +0200)]
fix for resize: remove useless permission checks
Dietmar Maurer [Wed, 8 Aug 2012 05:34:36 +0000 (07:34 +0200)]
fix for resize: avoid conflicting path
Alexandre Derumier [Tue, 7 Aug 2012 10:05:17 +0000 (12:05 +0200)]
qm : add resize
qm resize <vmid> -ide1 +10(G|M|K)
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 6 Aug 2012 09:56:34 +0000 (11:56 +0200)]
add qemu_block_resize
this call storage plugin resize first.
storage plugin will
return undef if we don't need to call qmp block_resize
or
return 1 if we need to call qmp block_resize
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Wed, 1 Aug 2012 11:44:54 +0000 (13:44 +0200)]
correctly use new disksize() method
Dietmar Maurer [Wed, 1 Aug 2012 11:22:43 +0000 (13:22 +0200)]
add size hint to drive options
We now store the size when we create a drive. This is just a hint, but
good enough to display on the GUI. We can now avoid to query the storage in
vmstatus().
Dietmar Maurer [Wed, 1 Aug 2012 11:15:57 +0000 (13:15 +0200)]
Avoid warnings about undefined values
Dietmar Maurer [Wed, 1 Aug 2012 06:58:16 +0000 (08:58 +0200)]
bump version to 2.0-47
Dietmar Maurer [Wed, 1 Aug 2012 05:29:23 +0000 (07:29 +0200)]
remove more unused subs
Alexandre Derumier [Sat, 28 Jul 2012 08:46:56 +0000 (10:46 +0200)]
remove old subs no more used
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 30 Jul 2012 12:58:40 +0000 (14:58 +0200)]
implement virtio-scsi-pci controller
This add the new virtio-scsi controller support.
http://wiki.qemu.org/Features/VirtioSCSI
Guest need kernel >= 3.4 to support. Windows drivers are also available in last virtio-win
Advantages :
- true scsi controller.(like lsi but a lot faster, around 5% slower than virtio-blk)
- multiples disk by controller (256 for now)
- scsi passthrough
- discard support (great for ssd or thinp storages)
- bootable
Hotplug is not yet available in 1.1. (already available in git)
to define lsi or virtio-scsi-pci controller for scsi disk:
scsihw: lsi|virtio-scsi-pci
default is lsi if not defined.
A megasas controller is comming form qemu 1.2, so we'll able to simply add it in the list of scsi controllers
lsi0 and lsi1 controllers have been renamed to generic scsihw0 and scsihw1,
so we can use them for both lsi or virtio-scsi controller type. (and use same pci slot addr).
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Fri, 27 Jul 2012 09:59:42 +0000 (11:59 +0200)]
bug fix: allow to set devices directly (-ide1 /dev/XYZ)
Alexandre Derumier [Fri, 27 Jul 2012 05:59:20 +0000 (07:59 +0200)]
replace file_size_info by volume_size_info.
Currently we only find a volume size by using qemu-img info (in file_size_info).
This doesn't works with "virtual" storage like sheepdog,rbd,iscsidirect,nexenta.
This also doesn't work with classic iscsi direct plugin.
So, we need to define a method for each plugin.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Mon, 16 Jul 2012 08:19:11 +0000 (10:19 +0200)]
migrate: only scan available storages
Alexandre Derumier [Sun, 15 Jul 2012 15:19:07 +0000 (17:19 +0200)]
implement qmp block_set_io_throttle on running vm config update
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Sun, 15 Jul 2012 15:19:06 +0000 (17:19 +0200)]
fixes for qemu_block_set_io_throttle
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Mon, 16 Jul 2012 04:59:53 +0000 (06:59 +0200)]
code cleanup, bump version to 2.0-44
Alexandre Derumier [Fri, 13 Jul 2012 13:12:41 +0000 (15:12 +0200)]
migrate: syncdisk : avoid scanning shared storage
Currently we get list from PVE::Storage (for unused volumes), from all storage.
If something goes wrong with the network on host and thenwe can't communicate with a network shared storage(sheepdog,rbd,..),
the vdisk_list die (timeout) and we cannot migrate the vm on another kvm host.(online or offline).
We don't need to scan shared storage, as they are no disk to sync.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Fri, 13 Jul 2012 10:37:19 +0000 (12:37 +0200)]
migrate: fix warning about uninitialized values
And display acurate byte values instead of KB
Dietmar Maurer [Fri, 13 Jul 2012 10:36:40 +0000 (12:36 +0200)]
fix command timeout
Dietmar Maurer [Fri, 13 Jul 2012 07:25:58 +0000 (09:25 +0200)]
new option for vmstatus to query $full informations from KVM using qmp
Dietmar Maurer [Fri, 13 Jul 2012 06:56:13 +0000 (08:56 +0200)]
use qmp for everything - remove old monitor code
Dietmar Maurer [Fri, 13 Jul 2012 06:42:13 +0000 (08:42 +0200)]
use qmp for vm_devices_list
Dietmar Maurer [Fri, 13 Jul 2012 05:30:13 +0000 (07:30 +0200)]
implement vm_human_monitor_command using qmp
Dietmar Maurer [Fri, 13 Jul 2012 05:06:22 +0000 (07:06 +0200)]
use new QMPClient code
Dietmar Maurer [Thu, 12 Jul 2012 10:28:27 +0000 (12:28 +0200)]
add PVE::QMPClient.pm
This is experimental code, not used currently.
Alexandre Derumier [Fri, 29 Jun 2012 10:34:55 +0000 (12:34 +0200)]
qmp_read_available : correctly handle end of response
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Tue, 26 Jun 2012 04:42:18 +0000 (06:42 +0200)]
fix whitespace errors
Alexandre Derumier [Mon, 25 Jun 2012 08:02:58 +0000 (10:02 +0200)]
convert eject cdrom monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:57 +0000 (10:02 +0200)]
convert migrate monitor commands to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:56 +0000 (10:02 +0200)]
convert cont monitor command (in vm_start) to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:55 +0000 (10:02 +0200)]
convert migrate_set_downtime monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:54 +0000 (10:02 +0200)]
convert migrate_set_speed monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:53 +0000 (10:02 +0200)]
convert system_powerdown and stop nocheck monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:52 +0000 (10:02 +0200)]
convert system_reset monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:51 +0000 (10:02 +0200)]
convert block_io_throttle monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:50 +0000 (10:02 +0200)]
convert balloon monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:49 +0000 (10:02 +0200)]
convert cont monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:48 +0000 (10:02 +0200)]
convert stop monitor command to qmp
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:47 +0000 (10:02 +0200)]
add vm_mon_cmd and vm_mon_cmd_nocheck wrapper
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:46 +0000 (10:02 +0200)]
vm_qmp_command : reworks of the sub
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Mon, 25 Jun 2012 08:02:45 +0000 (10:02 +0200)]
qmp_read_avail : reworks
qmp response could me more than 1 json.
we can have 1 json with event info, and 1 json with return infos.
We die if we receive an error message in response.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Alexandre Derumier [Sat, 23 Jun 2012 07:07:32 +0000 (09:07 +0200)]
check if volume exist on volume update.
for the moment we check only if the real path exist on the host.
This doesn't work for "virtual" device that host doesn't see.(like rbd, virtio-scsi,...).
This add a check if the volid exist in the storage.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Wed, 30 May 2012 10:08:33 +0000 (12:08 +0200)]
fix coding style
Alexandre Derumier [Tue, 29 May 2012 12:01:50 +0000 (14:01 +0200)]
add basic qmp support
this add qmp socket to kvm process
and anew sub copied from vm_monitor_command:
vm_qmp_command ($vmid, $cmdstr, $nocheck)
$cmdstr could be a simple command to be executed, without argument
vm_qmp_command($vmid,"stop");
or a complex hash with arguments
$cmdstr->{execute}="eject";
$cmdstr->{arguments}->{device}="ide1-cd0";
vm_qmp_command($vmid,$cmdstr);
documentation about qmp commands is here
http://git.qemu.org/?p=qemu.git;a=blob;f=qmp-commands.hx;h=
db980fa811325aeca8ad43472ba468702d4a25a2 ;hb=HEAD
Code must be polish a little more, but it's a start.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Wed, 30 May 2012 08:13:51 +0000 (10:13 +0200)]
fix pool permission checks on create
We do not need Permission.Modify on the pool.
Alexandre Derumier [Tue, 29 May 2012 05:56:16 +0000 (07:56 +0200)]
die if vdisk_free fail on delete_drive
replace the warn by a die.
Currently, if we vdisk_free a disk and something goes wrong (network
storage problem by example), the drive is removed from config and we
cannot retry to remove it later.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Thu, 24 May 2012 05:22:41 +0000 (07:22 +0200)]
fix cluster_lock_storage() call
Dietmar Maurer [Wed, 23 May 2012 06:05:36 +0000 (08:05 +0200)]
set RELEASE to 2.1
Dietmar Maurer [Wed, 23 May 2012 05:42:55 +0000 (07:42 +0200)]
avoid warning if vm does not belong to a pool
Dietmar Maurer [Wed, 23 May 2012 05:24:15 +0000 (07:24 +0200)]
minor fixes for newer pve-storage versions
Alexandre Derumier [Wed, 9 May 2012 12:29:29 +0000 (14:29 +0200)]
add diskio throttling option to drive
This add disk io limit to drive options.
I also add the qemu monitor command, but I din't have added yet to Qemu.pm
>From qemu mailing:
Some available features follow as below:
(1) global bps limit.
-drive bps=xxx in bytes/s
(2) only read bps limit
-drive bps_rd=xxx in bytes/s
(3) only write bps limit
-drive bps_wr=xxx in bytes/s
(4) global iops limit
-drive iops=xxx in ios/s
(5) only read iops limit
-drive iops_rd=xxx in ios/s
(6) only write iops limit
-drive iops_wr=xxx in ios/s
(7) the combination of some limits.
-drive bps=xxx,iops=xxx
Known Limitations:
(1) #1 can not coexist with #2, #3
(2) #4 can not coexist with #5, #6
(3) When bps/iops limits are specified to a small value such as 511 bytes/s,
this VM will hang up. We are considering how to handle this senario.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
Dietmar Maurer [Fri, 20 Apr 2012 07:30:42 +0000 (09:30 +0200)]
removed startall/stopall commands
Because we now support startup priorities, and start/stop order must include containers