Signed-off-by: Martin Perina <mperina@redhat.com>

We are checking maximum number of available file descriptors when
building engine for all support permutations. But in CBS this validation
doesn't work because maximum number of available file descriptors is not
reported properly. And we are also limiting number of workers for GWT
build, so we don't get near this limit.

Signed-off-by: Martin Perina <mperina@redhat.com>

Having 2 localWorkers for GWT build decreases compilation time
significantly and it doesn't exceed available resources in CBS.

Signed-off-by: Martin Perina <mperina@redhat.com>

Adding callback to RemoveDiskCommand so that it would wait for
RemoveImageCommand to finish, incl. removing the image from the DB.
This fixes an issue where RemoveDisk job was marked as FINISHED
before the image was actually removed from the DB.

Bug-Url: https://bugzilla.redhat.com/2137207

Signed-off-by: Martin Perina <mperina@redhat.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

apache-commons-jxpath is no longer needed, it was a dependency of
apache-commons-configuration, which has been already removed in
oVirt#619

Signed-off-by: Martin Perina <mperina@redhat.com>

dnf builddep needs root, let's try a passwordless sudo in case the
automation is using a different user. Sudo needs to be configred
separately.

When cloning a VM from a snapshot, isTpmEnabled parameter is not set
in VmSnapshotListModel, resulting in a loss of the TPM device and an
error if the TPM device is mandatory, i.e. for VMs with newer Windows
versions as the guest OS.  This patch fixes the omission and adds the
corresponding parameter.

Bug-Url: oVirt#702
Bug-Url: https://bugzilla.redhat.com/2136180

TPM device is not updated in
VmDeviceUtils.updateVmDevicesInParameters.  This results in omitting
the TPM device at least when exporting a VM to a data domain and an
error when the TPM device is mandatory, i.e. for VMs with a modern
Windows guest OS.  This patch fixes the omission and sets the
corresponding parameter.

Bug-Url: oVirt#702

* Create rhv-build

If getImagesList fails due to an error we should not continue with
attaching the SD assuming there no disks on the storage as this will
corrupt the OVF stores.

Instead fail the attach operation and have the user retry once the issue
is resolved.

Bug-Url: https://bugzilla.redhat.com/2126602
Signed-off-by: Benny Zlotnik <bzlotnik@redhat.com>

Previously, RemoveVm and RemoveAllVmImages didn't have a callback so they
didn't wait for all instances of RemoveImage to finish before finishing
RemoveVm and therefore the job of RemoveVm may have been completed while
the images are still locked in the database.

Now, RemoveVm is completed after RemoveImage completes and the images are
removed from the database.

Signed-off-by: Arik Hadas <ahadas@redhat.com>

GitHub builds had duplicated dates in RPM names. Also swapped date and
githash so that RPMs are time-sorted. Use githash for snapshot builds.

we run with bash -e, need to make sure all return codes are 0.

Signed-off-by: Martin Perina <mperina@redhat.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

ReduceImage is an asynchronous operation. Previously, on positive flow we
logged a successful execution message on both 'execute' and 'end-action'
phases of the command. The former is incorrect and therefore dropped (we
typically log a message that says that the operation has started at this
point but we don't plan for another translation cycle so it's not added
as part of this change).

Bug-Url: https://bugzilla.redhat.com/2141437
Signed-off-by: Arik Hadas <ahadas@redhat.com>

When importing an OVA that is not originated in oVirt the VM will diappear
after detaching its storage domain. The vm will not be seen in the import VM
tab. From now on even if the imported VM is no originated in ovirt it could
be imported after detaching the storage domain.

Bug-Url: https://bugzilla.redhat.com/2028242
Signed-off-by: Artiom Divak <adivak@redhat.com>
Signed-off-by: Arik Hadas <ahadas@redhat.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

Use ovirt-engine-build-dependencies >= 4.5.3 and < 4.5.4, which contains
latest build dependencies for ovirt-engine-4.5.3.z branch

Signed-off-by: Martin Perina <mperina@redhat.com>

…ine-setup"

This reverts commit e8eee20.

Moving it causes grafana on a separate machine to fail, because the
plugin to enable httpd_can_network_connect is now only on the engine
machine. Reverting that patch should fix that.

This does not introduce a new regression, because the need for it was a
mere optimization, but is now not needed anymore, as we removed
use/integration of ansible-runner-service.

Change-Id: I35f57983c03c2de78b16ad63d37f399db103a07b
Bug-Url: https://bugzilla.redhat.com/2126778
Signed-off-by: Yedidyah Bar David <didi@redhat.com>

When a user is asking for data via the API with some filter (name of the
required entity for example), we should call implicitly the search
engine from one hand to implement the filter and from the other hand we
are getting all the entities that the user can access according to the
permissions via a pre-defined query.
Given the search and the query results, both are intersected to given a
result that matches the given filter with the user authorization.

In AbstractBackendCollectionResource we have two signatures for getBackendCollection method:

protected List<Q> getBackendCollection(QueryType query, QueryParametersBase queryParams)
protected List<Q> getBackendCollection(QueryType query, QueryParametersBase queryParams, SearchType searchType)

The 1st just brings the query results, teh 2nd brings query and search
results and intersect them.

When a user tries to search for a specific entity and the 1st signature
is used instead of the 2nd, the result is that the first item is picked
from the resulted collection and mismatch the requested entity, this may
cause an error where an operation is attempted to be excuted with the
wrong information (for example, creating a template and giving the wrong
template cluster)

This patch fixes a wrong call to the 1st signature when the 2nd
signature should be used in the following entities:

Data Centers
Clusters
VNIC Profiles

Signed-Off-By: Eli Mesika <emesika@redhat.com>
Bug-Url: https://bugzilla.redhat.com/2144346

Signed-off-by: Eli Mesika <emesika@redhat.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

RemoveImage now attaches the created async task to itself rather than to
its parent command and its EndProcedure type is changed to COMMAND_MANAGED.
With these changes, first the execution of RemoveImage is completed and
then its parent's callback detects it and completes the execution of the
parent command.

Signed-off-by: Arik Hadas <ahadas@redhat.com>

RemoveImage may lack async tasks when:
1. The image doesn't exist
2. There's a failure during delete image

With these changes, the image is removed from the database (rather than
staying in LOCKED state) and the command would finish successfully in
the first case, and in the second case the command end with failure and
the image switches to ILLEGAL state.

Signed-off-by: Arik Hadas <ahadas@redhat.com>

Mark current directory as safe for git to be able to parse git hash when
building .src.rpm

Signed-off-by: Martin Perina <mperina@redhat.com>

Setting some of the properties returned an error [1]
This is a regression caused by https://bugzilla.redhat.com/show_bug.cgi?id=2089299

[1] https://lists.ovirt.org/archives/list/users@ovirt.org/message/OPH7BHGBGWMECQKRXY6GW3YGZR73JK2D/

Signed-off-by: dangel101 <dangel101@gmail.com>

Ansible-Runner artifacts are cleaned up periodically according to the properties
configured in engine-config:
AnsibleRunnerArtifactsLifetimeInDays: How many days artifacts will be kept before
they are cleaned up
AnsibleRunnerArtifactsCleanupCheckTimeInHours: The periodic time to execute the clean up

Bug-Url: https://bugzilla.redhat.com/show_bug.cgi?id=2151549
Signed-off-by: dangel101 <dangel101@gmail.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

Signed-off-by: Martin Perina <mperina@redhat.com>

Adding info regarding the file's creation time in days, current date in days
and the config value artifacts life time is set to, for easier debugging

Bug-Url: https://bugzilla.redhat.com/show_bug.cgi?id=2151549
Signed-off-by: dangel101 <dangel101@gmail.com>

WE have handled only IOException, but when a file cannot be removed due
to insufficient permissions then SecurityException is raised and because
of that we didn't continue with the removal of other directories.

Bug-Url: https://bugzilla.redhat.com/2151549
Signed-off-by: Martin Perina <mperina@redhat.com>

Signed-off-by: Michal Skrivanek <michal.skrivanek@redhat.com>

Signed-off-by: Michal Skrivanek <michal.skrivanek@redhat.com>

Bug-Url: http://bugzilla.redhat.com/2180230
Change-Id: Iee6fa946f52481091a431a6ef0f173d74f9e37f2
Signed-off-by: Eitan Raviv <eraviv@redhat.com>

When picking a host for transfer-image job, we filter hosts that
are UP and then process the data they reported on storage domains
without checking whether the host actually reported this data.
It can happen that hosts are UP and still have not reported data
on storage domains, in which case the data is null, resulting in
an NPE in the transfer-image flow.

This patch changes the predicate that is used to filter hosts for
transfer-image, adding a null-check to avoid that NPE.

Bug-Url: https://bugzilla.redhat.com/2203132
Signed-off-by: Arik Hadas <ahadas@redhat.com>

Signed-off-by: Laszlo Szomor <laszomor@gmail.com>

Signed-off-by: Arik Hadas <ahadas@redhat.com>