If the Data Center cluster compatibility
level is lower than the highest level
supported by engine, display a warning
icon in a new column in the Data Centers tab.

Hovering over the column should show the
alert message prompting the user to upgrade
the data center.

Change-Id: Ib0e4004c2834a314a707e62e5e7661a3e9c7c115
Bug-Url: https://bugzilla.redhat.com/1413150
Signed-off-by: Ravi Nori <rnori@redhat.com>

If the Cluster's cluster compatibility level
is lower than the highest level supported by
engine, display a warning icon in a new column
in the Clusters tab.

Hovering over the column should show the alert
message prompting the user to upgrade the Cluster.

Change-Id: I0f263ff9eddc89fc6b425aab87eb51927cba7c74
Bug-Url: https://bugzilla.redhat.com/1413150
Signed-off-by: Ravi Nori <rnori@redhat.com>

In commit I1e891ac39e , the columns job_id and step_id were removed from
command_entities table, therefore they need to be removed from the
taskcleaner script

Change-Id: If2d1218667649e6a7efd8e933dfb9f91b59eb972
Bug-Url: https://bugzilla.redhat.com/1425705
Signed-off-by: Fred Rolland <frolland@redhat.com>
(cherry picked from commit 4aa7884)

Change-Id: Ia2bb1b42452ea7ea5574505970e99d22b1f2f4e6
Signed-off-by: Sandro Bonazzola <sbonazzo@redhat.com>

In hyperconverged mode: nfs = off, cifs = off
optimize for virt = on

Change-Id: I26f0a02eb39dd34e1975695dfecf0bde9e849edf
Bug-Url: https://bugzilla.redhat.com/1420687
Signed-off-by: Sahina Bose <sabose@redhat.com>

Fix Affinity rule enforcer manager to dismiss
hosted engine VMs when choosing VM candidates for migration.

Change-Id: I83bcbc9ae0a70b078fcda377f17241892802873d
Bug-Url: https://bugzilla.redhat.com/1421771
Signed-off-by: Yanir Quinn <yquinn@redhat.com>
(cherry picked from commit 4106384)

Detect if the total memory or the number of CPUs
were changed in the Edit dialog.
If they were, NUMA nodes are changed to fit the new values.

Change-Id: I67b796a128c0af17dd54056b862798cd0d650bcf
Bug-Url: https://bugzilla.redhat.com/1291064
Signed-off-by: Andrej Krejcir <akrejcir@redhat.com>
(cherry picked from commit 54fc01a)

…re changed"

This reverts commit 9ce0636.
Reverted commit on gerrit: https://gerrit.ovirt.org/#/c/69946/

Change-Id: Ic8b10a33a1bf50421527b25d19fc1f4606a4d70d
Bug-Url: https://bugzilla.redhat.com/1291064
Signed-off-by: Andrej Krejcir <akrejcir@redhat.com>
(cherry picked from commit 5db67a7)

For migration or disaster recovery purposes we let the user filter
out the hosted-engine hosts to redeploy from a backup on a
different hosted-engine env. In order to be sure that all the
hosted-engine runs with a local up-do-date configuration we
have to remove them from the engine DB asking to the user to
redeploy them to be sure that they point to the right hosted-engine
storage domain.
In order to remove them at restore time avoiding foreign key errors,
we need to update all the existing VMs related to these hosts.
Set status=Down, exit_reason=AdminShutdown, exit_status=Normal,
run_on_vds=NULL, migrating_to_vds=NULL for all the VM that were running
on the hosted-engine hosts we are filtering out at restore time.

Change-Id: I85a6fcd18005393e69ed319e7b56709e7adc5b9b
Bug-Url: https://bugzilla.redhat.com/1416466
Signed-off-by: Simone Tiraboschi <stirabos@redhat.com>

… 4.0

Show a warning when setting deployment of hosted engine through the UI
and the compatibility version of the cluster is less than 4.0

Change-Id: I15f4856558762507845055a0b88eac1e8f948d18
Bug-Url: https://bugzilla.redhat.com/1383156
Signed-off-by: Andrej Krejcir <akrejcir@redhat.com>

Restoring HE environment from backup could lead into
two hosts, thinking they have spm_id==1 at same moment.
One of those hosts will be newly deployed HE host with
default spm_id==1, another one host will be old
host from the database. This patch changes
spm_id of the host with value '1' to some unused value.

Change-Id: Ib7b10f57a9350adf3da73580c4a69e5ce317502e
Bug-Url: https://bugzilla.redhat.com/1417518
Signed-off-by: Denis Chaplygin <dchaplyg@redhat.com>
(cherry picked from commit b440a7e)

Fix installing logutils.jar on downstream build when using minimead.py
which replaces pom version with specific version of the build.

Change-Id: I2099186df489841b95d1a086a05f03e9bb525d9d
Bug-Url: https://bugzilla.redhat.com/1408193
Signed-off-by: Martin Perina <mperina@redhat.com>

… 4.0

Removes the check to prevent deployment of HE on clusters with compatibility
vesrion less than 4.0.
The host must have repositories for ovirt 4.0 or newer,
otherwise the deployment will fail.

Change-Id: Iec29070e37058a3d02246e13563ad650a24bd6a1
Bug-Url: https://bugzilla.redhat.com/1383156
Signed-off-by: Andrej Krejcir <akrejcir@redhat.com>

For migration or disaster recovery purposes we let the user
filter out the hosted-engine storage domain
and the engine VM to redeploy from a backup on a different
hosted-engine env.
Due to another bug the user could create a floating
disk on the hosted-engine storage domain and then
connect it to the engine VM.
In this case we should still remove the hosted-engine
storage domain filtering out other VMs disks as well
without failing.

Change-Id: I2be23b172ea592f9b6be08477fad0f4bf29a30ad
Bug-Url: https://bugzilla.redhat.com/1416459
Signed-off-by: Simone Tiraboschi <stirabos@redhat.com>

Adding certificate url to UPLOAD_IMAGE_NETWORK_ERROR message.

Change-Id: I66383147495d0d814b0e9959fc3314763d8003a1
Bug-Url: https://bugzilla.redhat.com/1425774
Signed-off-by: Daniel Erez <derez@redhat.com>

FileUpload widget: support translation for
"Choose File" and "No file chosen" messages.

Change-Id: I488ec54ecb5c630bc5570db2c079a60a625b0d76
Bug-Url: https://bugzilla.redhat.com/1356562
Signed-off-by: Daniel Erez <derez@redhat.com>

The more relevant changes in the new version of the model are the
following:

* Add IPv6 details to the `NicConfiguration` type.

* Add NFS 4.2 support.

The more relevant changes in the new version of the metamodel are the
following:

* Fix generation of XML and JSON descriptions for enum types.

* Merge link with element if it has same name.

Change-Id: Ic037208b32005cff18bef94b9d1bd3489dec6c18
Related-To: https://bugzilla.redhat.com/1424821
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>

Previously there was no explicit timeout for communication to external
network provider. This way there was a long timeout applied to the
communication. Since some objects are locked during the communication,
some objects was locked for a long period of time if the external
network provided did not respond.

An explicit timeout for the communication is added, to avoid locking for
a long period of time.

The library com.woorea.openstack is used for the communication with
external network providers. The version of this library is increased to
3.1.2, which introduces the possibility to add the timeout for
communication.

Change-Id: Ic2eb1380273c811bf659ac067e6ebb65cbf31b99
Bug-Url: https://bugzilla.redhat.com/1380356
Signed-off-by: Dominik Holler <dholler@redhat.com>
(cherry picked from commit 5f7c657)

Show a warning about recommendation of adding
new volumes to scale out an HC cluster

Change-Id: I1534ccbaab95dc386ba373877a7a0c6fb9de0f05
Bug-Url: https://bugzilla.redhat.com/1417582
Signed-off-by: Sahina Bose <sabose@redhat.com>

While creating a arbiter volume or adding bricks to arbiter
volume, every third brick should be marked as arbiter brick

Change-Id: I7161ebca6e262a7b2606817a940b399461fd8460
Bug-Url: https://bugzilla.redhat.com/1417101
Signed-off-by: Ramesh Nachimuthu <rnachimu@redhat.com>
(cherry picked from commit e8a1110)

Renaming ImageProxyURL ConfigValue to ImageProxyAddress.
This is mandatory for upgrading from 3.6 as ImageProxyURL
has been removed in 4.0 and replaced with ImageProxyAddress.
See: https://gerrit.ovirt.org/#/c/56231/

Change-Id: Ifee377bf1e2122e6265d250575c011ed7de269e6
Bug-Url: https://bugzilla.redhat.com/1426064
Signed-off-by: Daniel Erez <derez@redhat.com>

Modify the NewDiskModel of Direct LUN 'Read-Only' of value, the value
should come from View, but not false.

Change-Id: Ic68a1d3dcc91b1d186d4894dca0a34c34687fa16
Bug-Url: https://bugzilla.redhat.com/1425347
Signed-off-by: Tony Wang <Sline798@gmail.com>

SetupNetworksBondPopupView.updateBondOptions() updates the value of
customBondEditor manually via the view. Setting the value this way
doesn't fire any event. Therefore, the model is not updated during
initialization.

Moving updateBondOptions() and its registration to
SetupNetworksBondModel fixes this behavior.

Change-Id: I6a16a729633c3e1bd90a47689dbaffa9d5e813d8
Bug-Url: https://bugzilla.redhat.com/1424813
Signed-off-by: Dominik Holler <dholler@redhat.com>
(cherry picked from commit d33523d)

The dependency of HostDevice feature on the `dedicated_vm_for_vds_list` was not
reified in the respecitve scheduling filter - thus allowing errorneous attempts
at Run or Migrate VM (that were bound to fail).

Change-Id: I2653769f94d6c9209fd98c6fd524c5377fc590fb
Bug-Url: https://bugzilla.redhat.com/show_bug.cgi?id=1414970
Signed-off-by: Martin Betak <mbetak@redhat.com>

Debug packages requires same exact version of the runtime
packages in order to properly work.
Added explicit require ensuring that the version match.

Concern has been raised related to eventual yum update
error messages related to version locking:
debug packages are not meant to be installed on any production
environment other than for debugging purpose.
Once the debug is finished they need to be removed.

Change-Id: I77f1e509c7e0d8ed903ad33c154098c974e08c39
Bug-Url: https://bugzilla.redhat.com/1401963
Signed-off-by: Sandro Bonazzola <sbonazzo@redhat.com>
(cherry picked from commit 1406ba1)

- Fixed a few more labels that were not properly
  associated with their associated radio/checkboxes

Change-Id: I1e27cfdc4c3cade5dc056c6e539e3aebbc82747a
Bug-Url: https://bugzilla.redhat.com/1414418
Signed-off-by: Alexander Wels <awels@redhat.com>

- Fixed issue with ComboBox not working as before

Change-Id: I5216d4de7bcb753d5715aac3833f663ad665b009
Bug-Url: https://bugzilla.redhat.com/1422779
Signed-off-by: Alexander Wels <awels@redhat.com>

Change-Id: I9b8a187c7805ece0796d7988d9b6a15c59be04d6
Bug-Url: https://bugzilla.redhat.com/1419967
Signed-off-by: Shahar Havivi <shaharh@redhat.com>

Change-Id: Ib8487e7cc5271d1ea17e6109b49bd6c6c51687aa
Signed-off-by: Marcin Mirecki <mmirecki@redhat.com>
Bug-Url: https://bugzilla.redhat.com/1422447

Required for bare metal to HE migration procedure.
Also allow to quickly map VDSM's host_id to host
in the engine.

Change-Id: I07e8efdb31dc92bfa2958b7261fcde15e14c0bb4
Bug-Url: https://bugzilla.redhat.com/1417518
Signed-off-by: Denis Chaplygin <dchaplyg@redhat.com>

Supplied script asks user in which datacenter
hosted engine will be installed and puts
hosts with spm_id=1 in that DC to the Maintenance mode.

Later, during database restore procedure,
engine-backup script should be run with
--he-remove-hosts options, which will
replace spm_id=1 with another value,
thus freeing it for the hosted engine.

Change-Id: I48e11a863b28b0c9a32f5a19b72392afbe717fe9
Bug-Url: https://bugzilla.redhat.com/1417518
Signed-off-by: Denis Chaplygin <dchaplyg@redhat.com>

- Moved text that needed to be translated from Templates to
  Constants.

Change-Id: I0c5a923530844916ada4deb5369abfbb0756e96c
Bug-Url: https://bugzilla.redhat.com/1422455
Signed-off-by: Alexander Wels <awels@redhat.com>

To allow collecting data and sending to a central store.

Change-Id: Ie497bae2d80c7bf27ff4bd401e9c83dc49a33694
Bug-Url: https://bugzilla.redhat.com/1406005
Signed-off-by: Yedidyah Bar David <didi@redhat.com>
(cherry picked from commit eb5e384)

Refreshed translations from oVirt's zanata project, version master,
following the completion of oVirt 4.1 translation cycle 2.

The following locales were updated:
    de_DE,
    es_ES,
    fr_FR,
    it_IT,
    ja_JP,
    ko_KR,
    pt_BR,
    zh_CN

Change-Id: I2ae07cf97291059c1370114b4dd7a8fa2dba1734
Bug-Url: https://bugzilla.redhat.com/1415806
Signed-off-by: Scott J Dickerson <sdickers@redhat.com>

When attempting to remove a template's disk from the Storage
sub-tab the operation fails with a ClassCastException.

Fixed by using a list of DiskModel instead of a StorageDomainModel
since we are selecting a disk and operating on it.

Another ClassCastException occurred on line 175 since
result.getState() returns a TemplateStorageListModel and not
a ConfirmationModel.

Change-Id: I81f8d145d4e0ec79f635b1e84589a53fda91f538
Bug-Url: https://bugzilla.redhat.com/1420265
Signed-off-by: Benny Zlotnik <bzlotnik@redhat.com>

Currently a selected table row font color is set to white but in cases
where the table row contains a select element the white font on the
selected row background will cause it to look disabled, this patch changes
the color to still be black but lighter

Change-Id: Idba9502f71ae69171497aca6cd01d5cf8956ccc1
Signed-off-by: Tal Nisan <tnisan@redhat.com>
Bug-Url: https://bugzilla.redhat.com/1408134

When copying a template's disk to a storage domain, the
StorageDomainValidator#hasSpaceForDiskWithSnapshots calculates
the disk's size with all other disk's copies as well which may results
in the wrong failure of the space requirement validation

This is fixed by only the image we are copying to the image
snapshots list, because it's the only one relevant to the
calculation of the free space.

Change-Id: Ibd5a925197e94a12e43a2c97d9350b913d209bf5
Bug-Url: https://bugzilla.redhat.com/1414320
Signed-off-by: Benny Zlotnik <bzlotnik@redhat.com>

…ndow

The elements did not fit in on the panel when dns entries were added,
resulting in elements being visible outside the popup window.
Added a scroll panel.
Also enlarged the window, so that the scrollbars fit in without breaking
the panel layout.
The dns size on the subnet panel was made smaller, so that the +/- buttons
fit in. The other use of ExternalSubnetWidget in the AbstractNetworkPopupView
also looks good with the new size.
Screenshots of affected dialogs:
http://imgur.com/a/mGqaM

Change-Id: I66c8fe243299a46b743370627489ae184952af6b
Signed-off-by: Marcin Mirecki <mmirecki@redhat.com>
Bug-Url: https://bugzilla.redhat.com/1425156

getEntitiesFromStorageOvfDisk() will enter infinte loop if the
ovfStoreDiskImages list contains only not updated ovf store disks.
This patch changes it to break in that case, avoiding the inifinte loop.
The issue existed since the relevant code was introduced, it was exposed
by https://bugzilla.redhat.com/1412547 which made it  occur frequently.

This patch makes the minimal change to solve the issue as it's going
to be backported to the stable branch, more refactors will be made on
the master branch.

Change-Id: I8669fc6a14879456116fb95ea23737f1ff6672a3
Bug-Url: https://bugzilla.redhat.com/1424755
Signed-off-by: Liron Aravot <laravot@redhat.com>

Enable setting the overflow in the disks table inside DisksAllocationView
and set it to auto in DisksAllocationPopupView which is used for move VM
disks dialog, this will avoid the disks list running over the dialog
buttons in case we are moving more than 8 disks

Change-Id: Ib5425ab4b156972a5db345f0754ef4758b9eaeb0
Bug-Url: https://bugzilla.redhat.com/1419352
Signed-off-by: Tal Nisan <tnisan@redhat.com>

BaseDisk entity and all it's subclasses are now nameable to allow running
actions that require naming on them, the name will refer to the disk alias

Change-Id: I57e208e6763a06fc11669a7fb8a49b3f7b349386
Related-To: https://bugzilla.redhat.com/1317490
Signed-off-by: Tal Nisan <tnisan@redhat.com>

The disk order sent in the Libvirt XML should be ordered lexonumerically
to avoid situations where "Disk10" will come before "Disk2"

Change-Id: I9522b4e893d42e0a44aa26557ff1d19a5ff10433
Bug-Url: https://bugzilla.redhat.com/1317490
Signed-off-by: Tal Nisan <tnisan@redhat.com>

PostDeleteActionHandler::fixParameters gets a parameters object of type
<T extends StorageDomainIdParametersBase & PostDeleteAction>.
While this is a correct generic type, it can be more specific since all
the parameters that are sent to this handler extend
StoragePoolDomainAndGroupIdBaseVDSCommandParameters, which is a subtype
of StorageDomainIdParametersBase.

Change-Id: Iff6988365db5e237dc736f333711047762dc4967
Bug-Url: https://bugzilla.redhat.com/1427543
Signed-off-by: Idan Shaby <ishaby@redhat.com>

Up until now, DAD (Discard After Delete) [1] was not related to Pass
Discard [2], i.e a disk was first discarded before it was deleted only
if its block storage domain's DAD value was true.

That created a situation where a vm disk that had its Pass Discard
enabled was not discarded before it was deleted if its storage domain's
DAD value was disabled. This situation does not make sense because if
the user specifies that he wants "live" discarding (Pass Discard), he
will most likely want to discard the whole disk before it's deleted.

This patch fixes the DAD logic so that a disk will be discarded before
it's deleted if DAD is enabled for its storage domain, or if the same
disk is attached to at least one vm with its Pass Discard enabled.

[1] http://www.ovirt.org/develop/release-management/features/storage/discard-after-delete
[2] http://www.ovirt.org/develop/release-management/features/storage/pass-discard-from-guest-to-underlying-storage/

Change-Id: I06345cd0cebed0ee4ad9ccc930e978807ab47f11
Bug-Url: https://bugzilla.redhat.com/1427543
Signed-off-by: Idan Shaby <ishaby@redhat.com>

This patch removes the support for TLSv1, as well as makes sure the
older SSLv3 protocol is also disabled.

This leaves only the TLSv1.1 and TLSv1.2, which are currently
considered as safe.

Change-Id: Icf8d0cc990ea6c2051f6275435fc9fe309e0fa37
Signed-off-by: Lev Veyde <lveyde@redhat.com>
Bug-Url: https://bugzilla.redhat.com/1388456
(cherry picked from commit 7f43bdf)

When running a live dc upgrade, active storage domains don't get updated
with the latest luns information from vdsm.
Since on new dc versions, it is possible that new information about the
storge domains' luns can be returned from vdsm, this patch synchronizes
these luns as a part of the upgrade process.

Change-Id: I81f9b247c73cffa688e7c2776b0da6fc937710ea
Bug-Url: https://bugzilla.redhat.com/1419327
Signed-off-by: Idan Shaby <ishaby@redhat.com>

The API supports the 'filter' parameter to indicate if results should be
filtered according to the permissions of the user. Due to the way this
is implemented, non admin users need to set this parameter for almost
all operations, as the default value is 'false'. To simplify things for
non admin users, this patch changes the default value to 'true', but
only for non admin users. If the value is explicitly given in a request
it will be honored.

This is a backwards compatibility breaking change, as clients that used
non admin users and did *not* provide explicitly the 'filter' parameter
will start to behave differently. However, this is unlikely, as calls
from non admin users without the 'filter=true' is almost useless. For
those unlikely cases where this may be a problem, the patch also
introduces a new 'ENGINE_API_FILTER_BY_DEFAUL' configuration parameter:

  #
  # This flags indicates if 'filtering' should be enabled by default for
  # users that aren't administartors.
  #
  ENGINE_API_FILTER_BY_DEFAULT="true"

If it is necessary to revert to the behaviour of previous versions of the
engine, it can be achieved by changing this parameter in a configuration
file inside the '/etc/ovirt-engine/engine.conf.d' directory. For
example:

  # echo 'ENGINE_API_FILTER_BY_DEFAULT="true"' > \
  /etc/ovirt-engine/engine.conf.d/99-filter-by-default.conf

  # systemctl restart ovirt-engine

Change-Id: I14d2e1101c32727f880ca8c82d48eb65fca9b8c2
Related-To: https://bugzilla.redhat.com/1420310
Bug-Url: https://bugzilla.redhat.com/1427987
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>
(cherry picked from commit 233e85a)

The previous patch enabled filtering by default for non-admin users,
like in the master branch. But we don't want to do that in the 4.1
branch, as it is a backwards compatibility breaking change. To avoid
that this patch flips the default value from 'true' to 'false'.

Change-Id: I851d04fa0f1f03b326e40f5434896418d24b7f56
Bug-Url: https://bugzilla.redhat.com/1427987
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>

Removing keepaliveInterval for download/upload image through the API.
The UploadImageUiInactivityTimeoutInSeconds ConfigValue is irrelevant
for the API as chunk_size is being explicitly specified by the script.
I.e. when the chunk_size is relatively large, a keepalive timeout
might occur, and hence, the download/upload would be paused.

Change-Id: Ie0ce6d9fbecc8d50ca410e678150e717fc166e3d
Bug-Url: https://bugzilla.redhat.com/1414484
Signed-off-by: Daniel Erez <derez@redhat.com>

And require instead ovirt-engine-metrics, which got its functionality

Change-Id: I9b95eba077b73cbcc8f09fa22d25f1846762ae51
Bug-Url: https://bugzilla.redhat.com/1422505
Signed-off-by: Yedidyah Bar David <didi@redhat.com>
(cherry picked from commit 3350190)

Two more NUMA node validations are added:
- memory consumed by all NUMA nodes of a VM
  should not be more than the total memory of a VM

- one CPU should not be pinned to two or more NUMA nodes

Change-Id: I7e91c23cb1dbc438f65613662e7e03c8c22ec1a5
Bug-Url: https://bugzilla.redhat.com/1291064
Signed-off-by: Andrej Krejcir <akrejcir@redhat.com>

This patch adds a LongValidation class to provide validation of Long
values.

Previously, the EditQuotaClusterModel and EditQuotaStorageModel classes
used the IntegerValidation class to validate some Long type values,
which resulted in mistakes such as those detailed in
https://bugzilla.redhat.com/1423468.

Change-Id: Ieddd09396be5bdb113bec430afa71913715130db
Bug-Url: https://bugzilla.redhat.com/1423468
Signed-off-by: Tony Wang <Sline798@gmail.com>
Signed-off-by: Phillip Bailey <phbailey@redhat.com>

The more relevant changes in this new version of the model are the
following:

* Add `auto_storage_select` attribute to the `VmPool` type.

* Move `vnic_profile_mappings` and `reassign_bad_macs` from
  the `import` operation to the `register` operation.

Change-Id: I9ce11fe08ce8e539d7d8cf74d0a284bda7de2d2b
Related-To: https://bugzilla.redhat.com/1425731
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>

This patch updates the CommandCallbacksPoller to set the command
correlation id before executing the needed callback method in order to
be able to track the flow easily in the logs.

Change-Id: Ib934bf4421a7fbf890b5c006263ff559b9d88462
Bug-Url: https://bugzilla.redhat.com/1428346
Signed-off-by: Liron Aravot <laravot@redhat.com>

When command is executed as a child command the correlation id should be
derived from the parent parameters to ease tracking the flow in the
logs.

Change-Id: Iaea59a769f6a44051340a9b9247f27fc5fc76be3
Bug-Url: https://bugzilla.redhat.com/1428346
Signed-off-by: Liron Aravot <laravot@redhat.com>

This patch changes the MoveImageGroup command to pass the correlation id
to the executed RemoveImage to ease tracking the entire flow in the log.

Change-Id: I31141ac43903ce723c3549efe4f58aa99f62e2a1
Bug-Url: https://bugzilla.redhat.com/1428346
Signed-off-by: Liron Aravot <laravot@redhat.com>

Before this fix vm devices hash was stored in an internal cache
for all vms that were reported by vdsm no matter if they were
present in the engine db at the time.

This caused problems with hosted engine vm that was imported
after the device hash was saved in the cache. Since device monitoring
relies on this cache and there was no change in the hash reported
from vdsm and the hash in the cache the devices weren't imported.

This patch allows monitoring caching info only for vms that already
exist in the db and so solves this issue.

Change-Id: Ib011ac5c7f28de7891b0b6e3c09b107679c1ff29
Bug-Url: https://bugzilla.redhat.com/1425108
Signed-off-by: Jenny Tokar <jtokar@redhat.com>

When importing vm devices the db generation column was
not updated.
The ovf generation process relies on this column to know
when a new ovf needs to be generated. Since the column
wasn't updated the imported devices were not included in
the ovf.

Change-Id: I0fe8c7a350e417db1ce0ba273a7d470a89afbe69
Bug-Url: https://bugzilla.redhat.com/1416846
Signed-off-by: Jenny Tokar <jtokar@redhat.com>

Pause image upload upon detecting file read error
(e.g. file was deleted during upload).

Change-Id: Iefb9cbe60b8b71ae10cfe272093f9eec8f88be69
Bug-Url: https://bugzilla.redhat.com/1365237
Signed-off-by: Daniel Erez <derez@redhat.com>

determineSourceImageGroup() is used to determine the source image group
(which may differ between the first volume and the rest of the volume
 chain in case the disk is based on a template).
The method had two issues:
1. It returned the image id instead of the disk id.
2. It used image.getImageTemplateId() to determine if the disk is based
on template. When disk snapshots are created their image template id
field is copied from the parent volume, which means that all the chain
has the same value in tht field - only the first volume needs to be
created as based on the template disk. This patch changes the check to
use the template disk id only when the template volume is the parent of
the volume by checking that getParentId() equals to
getImageTemplateId().

Change-Id: Ibf5f6cac89846865bb36e07507390e1a9b21e99a
Bug-Url: https://bugzilla.redhat.com/1429184
Bug-Url: https://bugzilla.redhat.com/1429170
Signed-off-by: Liron Aravot <laravot@redhat.com>

when setting a VM / Template / Pool / Instance Type as headless, the
following fields are now greyed out (were not till now):
'Monitors' label
'Single Sign On method' label

In addition, the following fields:
USB Support
Monitors
Enable SPICE file transfer
Enable SPICE clipboard copy and paste
are all left greyed out for headless even when the OS is changed to "Linux"
or "Red Hat Enterprise Linux 6.x x64 (were changed to non greyed out
till this patch).

Change-Id: Ia7f93a77cced74529b3fe194ba8746750d6ef50b
Bug-Url: https://bugzilla.redhat.com/1425124
Signed-off-by: Sharon Gratch <sgratch@redhat.com>

When setDiskProfilesList() is executed, the disk profile set on the disk
is considered when updating the disk profile list.
However, for disk models that doesn't represnt disk which were already 
actually created the profile set in the model isn't the actual disk
profile - therefore if the domain is changed we can update it to the 
newly selected domain profile and not consider it as set profile.

Change-Id: I24d357cdd324817834b8ee2902ec7d3c9add99b6
Bug-Url: https://bugzilla.redhat.com/1360355
Signed-off-by: Liron Aravot <laravot@redhat.com>

Bug-Url: https://bugzilla.redhat.com/1425725
Change-Id: Ib23e3ecf9ca4aefcc4fd2bef2bdf692e01e3a40e
Signed-off-by: Piotr Kliczewski <piotr.kliczewski@gmail.com>

'auto storage select' field indicate automatic and
balanced disk distribution for multiple storage.

Change-Id: I5f7407bd7883092ddb599c283ba3cc1a52055879
Bug-Url: https://bugzilla.redhat.com/1425493
Signed-off-by: Shahar Havivi <shaharh@redhat.com>

Consume newer ovirt-host-deploy 1.6.3.

Change-Id: Ic1820141c81d6eeb40ccf9f97b9d6d152f354aca
Signed-off-by: Sandro Bonazzola <sbonazzo@redhat.com>

Direct LUNs should only be sent with device="lun" when using virtio-scsi
when they are connected with virtio interface they should be sent with
device="disk" as all the other disks

Change-Id: I96e9604b69136e05b771aa54616810f7d1aee8db
Bug-Url: https://bugzilla.redhat.com/1410506
Signed-off-by: Tal Nisan <tnisan@redhat.com>

Added a warning for unplugged disk on remove snapshot (live) dialog.

Change-Id: I7f4b2549ceaf752c531acd128de43aecee99c002
Bug-Url: https://bugzilla.redhat.com/1411572
Signed-off-by: Daniel Erez <derez@redhat.com>

When the user requests a shutdown of the engine with the 'systemctl stop
ovirt-engine' command, systemd sends a TERM signal to the service
start/stop script, which in turn sends it to the application server. The
application server then performs its shutdown process. This includes
undeployment of the data sources. When the data source used by the jobs
scheduler is undeployed and there are still jobs pending, they are
retried for ever, and they always fail because there is no longer data
source that they can use. This blocks the application server shutdown
process. The application server is later killed using the KILL signal,
by the start/stop script, after waiting 10 seconds. The net result is
that the engine is stopped anyhow, but systemd reports it as failed, and
some shutdown tasks may have been skipped, because the process was
blocked. To avoid that this patch changes the scheduler implementation
so that it doesn't wait for jobs to finish during shutdown.

Change-Id: I6fc474324d1dc8551933761c3b729d13243b323a
Bug-Url: https://bugzilla.redhat.com/1394617
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>
(cherry picked from commit f475d6d)

The more relevant changes in the new version of the model are the
following:

* Replace the `Seal` operation of the service that manages a template
  with a new `seal` parameter in the operation that adds a template.

Change-Id: I6f6b36b57cc6c7cb7d6ad1372de07f2da7a179b7
Related-To: https://bugzilla.redhat.com/1335642
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>

Change-Id: I84fb574bd671345bf293a41effd4417e30005daa
Bug-Url: https://bugzilla.redhat.com/1424821
Signed-off-by: Tal Nisan <tnisan@redhat.com>

The bonding options are normalized by placing the option "mode"
in the first place and the option "miimon" in the second place
in the options string.

Previously, if the option "miimon" was missing, a space was added
the first place of the options string instead of the present option.

If the option "miimon" is missing, the option "mode" is still
positioned in the first place.

Change-Id: Ie3417fe2dc5bffb1b98c6a664da1c0be87fc54be
Bug-Url: https://bugzilla.redhat.com/1429810
Signed-off-by: Dominik Holler <dholler@redhat.com>
(cherry picked from commit a07cfab)

Currently when the API doesn't isn't able to calculate the URL path for
a class, it sends an ERROR message to the log, something like this:

  ERROR Can't find relative path for class "o.o.e.a.r.VmDisksResource", will return null

That isn't really an error, rather a hint for developers that something
may be missing in the 'LinkHelper' class. But the ERROR serverity alarms
users. This patch changes the severity to WARNING, which is closer to
the real meaning.

Change-Id: Ib65c904360787d50d22d21fe50747b6ed7453a01
Related-To: https://bugzilla.redhat.com/1410038
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>
(cherry picked from commit da7e8b2)

…lable

Currently we allow memory over commitment even if nothing can
support it on a host (no ksm and no ballooning).
This patch adds verification for that issue so an error is raised
if the configuration is wrong.

Change-Id: I912285e8d8cd1d55cf58b14bf04f7d2e8bb84e75
Bug-Url: https://bugzilla.redhat.com/1402838
Signed-off-by: Jenny Tokar <jtokar@redhat.com>

service.py needs it.

Change-Id: Id9443fbd02770355e1ea3e50a36d10c28997e10d
Bug-Url: https://bugzilla.redhat.com/1430817
Signed-off-by: Yedidyah Bar David <didi@redhat.com>

Allow getting total amount of memory

Make setup code use that

Allow parsing values intended for java -Xm{s,x}value

Change-Id: I14bfa4b8165c1b899b82d623e2e0dc3fbea2b1bc
Bug-Url: https://bugzilla.redhat.com/1329119
Signed-off-by: Yedidyah Bar David <didi@redhat.com>
(cherry picked from commit 60233fb)

When the VM is down, the vm_dynamic-dependent `deriveGraphicsType()` returns
null which caused NPE when forwarding to a subresource of a given console type.

Fixed by forwarding the null graphics type to the backend where the down VM
case is already handled.

In order to do this the ticket invocation code was extracted to the Ticketer
helper class that encapsulates the single operation and is used both by the old
BackendVmResource endpoint and the newer BackendVmGraphicsConsoleResource
subresource endpoints.

Change-Id: I58250c34a5624f8be39b84e7eed1956dc589d995
Bug-Url: https://bugzilla.redhat.com/show_bug.cgi?id=1428863
Signed-off-by: Martin Betak <mbetak@redhat.com>

AttachDiskModel is a sub class of AbstractDiskModel, and thus when
events like storageDomain_SelectedItemChanged are raised in
AttachDiskModel, updatePassDiscardAvailability and
updatePassDiscardChangeability are called although Pass Discard is
irrelevant for AttachDiskModel.
This patch adds an empty override of these methods to AttachDiskModel to
solve this problem.

Change-Id: I451b58a0618098690c89dce9aae5929e30d64b54
Bug-Url: https://bugzilla.redhat.com/1430754
Signed-off-by: Idan Shaby <ishaby@redhat.com>

Audit log event should be added on
user login.

Change-Id: I31dfc3ecc4e9cc89ca9de2f602690c33b181b2cf
Bug-Url: https://bugzilla.redhat.com/1426715
Signed-off-by: Ravi Nori <rnori@redhat.com>

Polpuplate macPoolId on DC in the following queries:
* GetAllStoragePools
* Search

Change-Id: Ia76a8ca63c2a3aca6c81e5e0c361d780dbf8a4d1
Bug-Url: https://bugzilla.redhat.com/1421434
Signed-off-by: Yevgeny Zaspitsky <yzaspits@redhat.com>
(cherry picked from commit fc6b190)

Since every IO Thread has to have it's own virtio scsi controller and the
current code supported only one controller, this patch adds the support for
having more than one.

If the num of io threads of a VM is lowered, the controller Id is re-generated
for this disks.

If the num of io threads is enlarged, the existing disks which already have
addresses are not re-generated, only for the new disks will be assigned to
new controller.

Therefore, in order to enforce the rearrangement of the disks to io threads
which has already been given some address, the disks have to be de-activated
and activated again. The reason is to not to touch the assignement of the disks
to controllers since it influences the disk arrangement inside of the guest.

Change-Id: I5bc8d58237ae28901bf866a50f30e87ee852c980
Bug-Url: https://bugzilla.redhat.com/1422450
Signed-off-by: Tomas Jelinek <tjelinek@redhat.com>

The VmDeviceUtils#updateUsbSlots method was not being called in the flows in
UpdateVmTemplateCommand and AddVmCommand, which resulted in creation of
templates (and consequently also VMs) with incorrect number of usb controllers
and thus preventing the VM to be run in some cases.

Change-Id: Ic8fbae5d863307bc423c3e04cfd56c370a850863
Bug-Url: https://bugzilla.redhat.com/show_bug.cgi?id=1430009
Signed-off-by: Martin Betak <mbetak@redhat.com>

This reverts commit 0af00ce.

Change-Id: Iaa2a5ebb0c1f96d05693731e3cc5d241cd5aa159
Signed-off-by: Maor Lipchuk <mlipchuk@redhat.com>

Libvirt expects the IO Threads to be indexed from 1, not from 0.

For virtio blk it was already implemented correctly in

VmInfoBuilderImpl.pinToIoThreads()

but the case of virtio-scsi have indexed the threads from 0.

Change-Id: Ifa818d31d56cbd8f098620d6bc6598ce6ca5d4ea
Bug-Url: https://bugzilla.redhat.com/1366507
Signed-off-by: Tomas Jelinek <tjelinek@redhat.com>

The more relevant changes in this new version of the model are the
following:

* Add `unregistered` parameter for the operations to list disks,
  virtual machines, and templates in storage domains.

* Fix the direction of the `statistic` parameter of the `Statistic`
  service, must be out only.

* Revert the addition of the operation to update disks.

Change-Id: Ic88d7e257d24ddcf1782b968e7d2b27b0a125240
Related-To: https://bugzilla.redhat.com/1428159.
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>

Fix a bug of not setting the domain name value in case of enabling the
Cloud-Init/Sysprep option for VM (new/edit) and Run-Once dialogs.

Change-Id: I8f7934347e05497cca1d4a6b0c8950cbb85f7cf7
Bug-Url: https://bugzilla.redhat.com/1425122
Signed-off-by: Sharon Gratch <sgratch@redhat.com>

Problem: Since engine 4.1 VM-like entities have new property - max
memory. If one tried to create a new VM using script created for engine
4.0 specifying custom memory size larger template max memory and didn't
specifying max memory explicitly then creation of VM failed. Max memory
copied from template was smaller than memory.

Solution: If max memory is not specified and memory is specified, then
max memory is automatically set to value
VmCommonUtils#getMaxMemorySizeDefault(int memory) (currently 4*memory).
UI uses same method.

Change-Id: I6e1d0847c68a774f6028824e4533df772c19dfea
Bug-Url: https://bugzilla.redhat.com/1430795
Signed-off-by: Jakub Niedermertl <jniederm@redhat.com>

When failing on creation of desktop/server vm based on a template
currently there is no rollback of the created disks in case of failure
to complete the entire operation.
For server vms the rollback was never implemented while for desktop vms
the rollback isn't perform since the flow was updated to use the coco
framework (as it was performed using the SPMRevertTask).
As the endWithFailure() method is shared between the flows, this patch
adds the rollback for both.

Note that is this patch is targeted to the 4.1 branch the solution isn't
yet complete - the flow should be rewritten so that the disk is added in
transaction and then the copy flow is initiated.
In case of failure to complete the creation/copy the engine should
attempt
to remove the disk, if the removal fails the disk should remain in
ILLEGAL status so that the user will be able to attempt to delete it to
free the space used by it.

Change-Id: I530f522cd9fe064bb2844ed89f8d4877b25af0f5
Bug-Url: https://bugzilla.redhat.com/1426136
Signed-off-by: Liron Aravot <laravot@redhat.com>

Change-Id: I8b3b0c7ebda2e91c90478a8735efd401ec16fd44
Bug-Url: https://bugzilla.redhat.com/1426265
Signed-off-by: Tal Nisan <tnisan@redhat.com>

Currently when version 3 of the API is used to request the disks of a
virtual machine, the API uses the GetAllDisksByVmId query, which in turn
uses the GetDisksbyVmGuid stored procedure to retrive the disks. In
addition this query, for each disk, calls the GetSnapshotByLeafGuid
stored procedure, to retrieve snapshot details, and the
GetDiskVmElementVyDiskVmElementId, to get attachment details. The
GetSnapshotByLeafGuid stored procedure is complex, and consumes a
important amount of database time. Version 4 of the API uses lighter
stored procedures. This patch changes the V3 compatibility layer of the
API so that it calls the V4 disk attachments and disk resources instead
of using the GetAllDisksByVmId query. In an otherwise idle enviroment
with approxy 4000 virtual machines this reduces the total use of CPU
(85% database + 15% engine) from 57 minutes to 14 minutes.

This can be reproduced using the following Ruby script, which uses the
'ovirt' gem to retrieve the virtual machines and their disks:

  require 'active_support/all'
  require 'ovirt'

  service = Ovirt::Service.new(
    server: 'rhvm41.local',
    port: 443,
    domain: 'internal',
    username: 'admin',
    password: '...',
    verify_ssl: false,
    timeout: 1.hour
  )

  vms = Ovirt::Vm.all(service)

  vms.each do |vm|
    vm.disks.each do |disk|
      puts "#{vm[:name]}: #{disk[:name]}"
    end
  end

  service.disconnect

Change-Id: I4429aadb4c33271c6ba10752cd3502ceb1d2329e
Bug-Url: https://bugzilla.redhat.com/1431578
Signed-off-by: Juan Hernandez <juan.hernandez@redhat.com>
(cherry picked from commit 85779a0)

• renamed parameters
• created new parameter: macToMigrate, which was formerly initialized
inside of method

Change-Id: I5e74e218895d6c0b6fbc015d29a424fffab033ed
Bug-Url: https://bugzilla.redhat.com/1410440
Signed-off-by: Martin Mucha <mmucha@redhat.com>

MoveMacsOfUpdatedCluster.java -> MoveMacs.java
Class was renamed, because macs will be moved not only because of
updating cluster.

Change-Id: I524abafd06ab5e57e8e2fc36f62c31fd58ec993e
Bug-Url: https://bugzilla.redhat.com/1410440
Signed-off-by: Martin Mucha <mmucha@redhat.com>

to allow adding and 'force-adding' list of Macs.

Change-Id: I18e60646da2bd82952dd2eae3d4849d1fb39d311
Bug-Url: https://bugzilla.redhat.com/1410440
Signed-off-by: Martin Mucha <mmucha@redhat.com>

Adding ID to MacPool to be able to identify different MacPools in the
logs. The DB ID (mac_pools.id) will be used for this.

Note:
ID is not a 'required' property of in-memory object (with an exception
in compensation flow), and might bind itself to DB, it *is* a property
of MacPool, and a good candidate to identify pool. A pool is related
to db record, and has ID. It's very unlikely, that any future
implementation of pools won't be able to have ID, thus proceeding
without pool having ID is just complicating work for ourselves.

Change-Id: Icb688c24e9105850e70291a67d7fcc129773a9ca
Bug-Url: https://bugzilla.redhat.com/1395462
Signed-off-by: Martin Mucha <mmucha@redhat.com>

when moving VM to another cluster, and those clusters has set
different Mac Pools set, then macs needs to be updated. All macs of
such VM needs to be removed from Mac Pool of source Cluster and
added to Mac Pool of target cluster.

If target Mac Pool disallows duplicates and there's clash, operation
will be suspended.

Change-Id: Ic57ff7ac249946fa0eafc6034bada1ee86d31b7f
Bug-Url: https://bugzilla.redhat.com/1410440
Signed-off-by: Martin Mucha <mmucha@redhat.com>

Failure during RestoreAllSnapshotsCommand execution unlocks the
'In Preview' snapshot while keeping the 'Active VM before the
preview' snapshot. This allows previewing the snapshot again,
thus, making another 'preview' snapshot and a lot of mess.

The suggested solution is to simply avoid unlocking the snapshot
on endWithFailure.

Since the current logic exists from 2013 and seems invalid now,
I'm open for any other suggestions :)

Change-Id: I94c0b30b7621b8831f13815b3fdd8cdbce0117b7
Bug-Url: https://bugzilla.redhat.com/1427104
Signed-off-by: Daniel Erez <derez@redhat.com>

... since in that case users often want to perform a memory hot plug and
prefer max memory to remain untouched rather than being adjusted for the
next run.

Change-Id: I64bc7941c37cca667586c56e32ef3000c6c6efe6
Bug-Url: https://bugzilla.redhat.com/1430451
Signed-off-by: Jakub Niedermertl <jniederm@redhat.com>