Implement VMs live snapshots

Add a new command to take live snapshots of VMs:

  clientIF.snapshot(vmId, snapDrives)

The parameter snapDrives is in the format:

  snapDrives = [
    { "domainID": "<sdUUID>",
      "imageID": "<imgUUID>",
      "baseVolumeID": "<baseVolUUID>",
      "volumeID": "<volUUID>" },
    ...
  ]

Change-Id: Id48a905938037eca2b1de966f4f83d801fdb9970

destroy may be called when the Vm is already Down

This patch fixes an error introduced in c0766ddb0852fd391d15424ae24761.

Change-Id: I50e4ca55262e64b13081cc832fba932e56aafd49
Reviewed-on: http://gerrit.ovirt.org/1321
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

add a virtio console to each VM

With this patch, vdsm-controlled VMs support `virsh console`. Note that
guest kernel should have virtio drivers and stream its console i/o to
the virtio-console device.

Change-Id: Id4e14963a5bc220f1fcc33477f19ef2eed1428ba
Reviewed-on: http://gerrit.ovirt.org/585
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Add before_vm_destroy hook

Run a hook directory when a Vm is asked to die. Note that a Vm may die
without being asked to, were the hook would not run.

This hook is currently missing from Vdsm since I was not sure what
should be its exact semantic. However in hindsight, we'd better have
something (even awkward) and improve after feedback from the real world.

Change-Id: Ide2c143921e58944df97c3dc74cd20cd85b1dae8
Reviewed-on: http://gerrit.ovirt.org/1175
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Add hotplugDisk/hotunplugDisk feature to vdsClient.

Usage: vdsClient 0 hotplugDisk|hotunplugDisk vmUUID <drivespec>
   'drivespec parameters list: r=required, o=optional',
   'r   iface:<ide|virtio> - Unique identification of the existing VM.',
   'r   index=<int> - disk index unique per interface virtio|ide',
   'r   [pool:UUID,domain:UUID,image:UUID,volume:UUID]|[GUID:guid]|[UUID:uuid]',
   'r   format: cow|raw',
   'r   readonly: True|False   - default is False',
   'r   propagateErrors: off|on   - default is off',
   'o   bootOrder: <int>  - global boot order across all bootable devices',
   'o   shared: True|False',
   'o   optional: True|False'

Change-Id: I234a2196ebfd3ee5a2a8ca00347c61adbf41f29f
Reviewed-on: http://gerrit.ovirt.org/410
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Add hotPlug/hotUnplug disk feature

Change-Id: Iecab9e28b2e77cfeedc3b9068095f7d117eb6221
Reviewed-on: http://gerrit.ovirt.org/1108
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Fix metadata file permissions for fileSD

In this patch:
* remove createmetafile (we already have other functions to create an
  empty file)
* create empty metadata files (createmetafile was using sizeStr2Int
  which is wrong)
* set the metadata files permission to 0660

Change-Id: I97ab4f724b8f061161da414c1fc3cfd493298430
Reviewed-on: http://gerrit.ovirt.org/1252
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

This patch adds some basic debian support

Change-Id: Ie021cd9f64f80fde2442b6ad7c75c0b5b1c746ff
Reviewed-on: http://gerrit.ovirt.org/1110
Tested-by: Dan Kenigsberg <danken@redhat.com>
Reviewed-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Reviewed-by: Matthias Heinz <matthias.heinz@goobernetworks.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

BZ#782348 - Fix getBridgeParams()

Blank lines are valid in /etc/sysconfig/network-scripts/ifcfg-ethX

Change-Id: Ic3f40d235fc63b727472c2f250a65755d1aa7791
Reviewed-on: http://gerrit.ovirt.org/1102
Reviewed-by: Ayal Baron <abaron@redhat.com>
Tested-by: Haim Ateya <hateya@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Set exc_info=True when self.message is used in log.error for
vds_bootstrap

Change-Id: Id5fa63d9809e5f7130e304eceede6d41a27a12a6
Reviewed-on: http://gerrit.ovirt.org/945
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#619360 enable and start ntpd

Vdsm needs to keep host time in sync, or else guest time would lap if
guest is started on a different host. Make sure ntpd is running.

Change-Id: I1504a5d86e71ae2d210ada6a89e9c55b46f99d14
Reviewed-on: http://gerrit.ovirt.org/903
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Add validate minimal keys set in dictionaries

Change-Id: I66b9d3947b676e7ed361bab1d552dfbcb4f6e342
Reviewed-on: http://gerrit.ovirt.org/540
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Saggi Mizrahi <smizrahi@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Fix address parsing during build domXml

Change-Id: Icbefee04f6716b05d55b0dd3458c08806359efbc
Reviewed-on: http://gerrit.ovirt.org/1207
Reviewed-by: Saggi Mizrahi <smizrahi@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Make getRoutes() return the default gateway of each bridge interface

Assuming `route -n` returns the followings:
Kernel IP routing table
Destination   Gateway     Genmask       Flags Metric Ref Use Iface
192.168.1.0   0.0.0.0     255.255.255.0 U     0      0   0   engine
0.0.0.0       192.168.1.1 0.0.0.0       UG    0      0   0   engine
192.168.10.0  192.168.1.2 255.255.255.0 UG    0      0   0   engine

Then the last shown gateway of engine '192.168.1.2' would be returned
by this functions since it latest occurs in /proc/net/route.
This commit makes this function always return the default gateway of
each bridge interface as ovirt-engine expects.

Change-Id: I6bceb3be2495bc80931c709e2132ff4542e6b763
Reviewed-on: http://gerrit.ovirt.org/1101
Reviewed-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Tested-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

setupNetwork: release deleted network first

We may need their resources for newly requested networks.

Change-Id: I23f5feb0005fd6853793979dbfba84f89dd9c5fb
Reviewed-on: http://gerrit.ovirt.org/893
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Shahar Havivi <shavivi@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Do not distribute vdsm-gencert.sh

... and particularly not one that has /usr/local as a prefix, and fails rpm
creation.

Change-Id: I3958c746d15698a9dea5d92423cb00a3cc07c0e1
Reviewed-on: http://gerrit.ovirt.org/1201
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>
Reviewed-by: Federico Simoncelli <fsimonce@redhat.com>

Fix race during VM recovery flow

In this patch:
 * Save conf just after build devices

Change-Id: I9253423d0f6ff849022dabf8e29409a1a5e59851
Reviewed-on: http://gerrit.ovirt.org/1183
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Ignore differences in case when comparing mac addresses

Change-Id: Id59b2228f83aa2506da1f4a1011469e63dfc1e9c
Reviewed-on: http://gerrit.ovirt.org/1059
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

vdsm/Makefile.am: fix make install

currently vdsm-logrotate.conf is not generated automatically, which break the 'make install'
[vdsmdevel@medogz vdsm]$ sudo make install
[sudo] password for vdsmdevel:

<snip>
/usr/bin/install -c -m 644 vdsm.rwtab \
/etc/rwtab.d/vdsm
/bin/mkdir -p /etc/logrotate.d
/usr/bin/install -c -m 644 ./vdsm-logrotate.conf \
/etc/logrotate.d/vdsm
/usr/bin/install: cannot stat `./vdsm-logrotate.conf': No such file or directory
make[3]: *** [install-data-logrotate] Error 1
make[3]: Leaving directory `/home/vdsmdevel/sem-patch/vdsm/vdsm'
make[2]: *** [install-am] Error 2
make[2]: Leaving directory `/home/vdsmdevel/sem-patch/vdsm/vdsm'
make[1]: *** [install-recursive] Error 1
make[1]: Leaving directory `/home/vdsmdevel/sem-patch/vdsm/vdsm'
make: *** [install-recursive] Error 1
[vdsmdevel@medogz vdsm]$

Change-Id: I6488114ad6bbdd666f5309fbb943e04c4f558f75
Reviewed-on: http://gerrit.ovirt.org/1124
Tested-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Reviewed-by: Saggi Mizrahi <smizrahi@redhat.com>
Reviewed-by: Federico Simoncelli <fsimonce@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Removing dead code.

Change-Id: I9b34293b5c7fa8393a4ac76f912ba46f9208fbba
Reviewed-on: http://gerrit.ovirt.org/1069
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Avoid to activate parent volume on cloning

The benefit is that we can now create new snapshot volumes without
looking at the parent. This is indispensable for live snapshots since
the parent volume is in use by the VM.

WARNING: if you run an SPM host with this change you need all the
HSM hosts to have: I4b5cb55a868cb9bf5a720610273af139ee858a1a (Don't
lock the image when preparing the VM path) and you won't be able use
the domain from old SPMs/HSMs anymore.

In this patch:
* Do not activate the parent volume on cloning (it could be in use
  by the VM), the only information needed is the size and it can be
  provided as additional parameter in qemu-img
* Do not change the volume permission to read-only when setting the
  parent to "internal" (it could be in use by the VM)

Change-Id: I8a7d488da5ab0c27fc2e083976d4bc44e6f810e3
Reviewed-on: http://gerrit.ovirt.org/570
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Refactor prepareVolumePath in clientIF

In this patch:
* refactoring
* set drive['volumeChain'] (required for sanlock and bz760547)
* set drive['blockDev']  (required for the hotplug feature)

Change-Id: I08f86b32a71d61ade123ceea541c51b669d84ca0
Reviewed-on: http://gerrit.ovirt.org/1141
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Remove the _preparedDrives dictionary from vm

The _preparedDrives dictionary was a fix for bz670599 introduced with
the commit 7aa57fe. Currently prepareVolumePath is not relying on the
resource manager anymore and therefore the race fix is superfluous.

In this patch:
* make prepareVolumePath and teardownVolumePath public

Change-Id: I8d310f216987b7ded89e71752464934f70c051ad
Reviewed-on: http://gerrit.ovirt.org/1123
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Do not lock the image when preparing the VM path

Since the resource manager scope is local there is no need for a VM
to take an exclusive lock on the images: in fact the SPM running on
an other machine could still manipulate such images.
Using the resource manager in general on HSM is wrong because it is
mapped on actions that are valid only on the SPM such as: modifing
the lvm permissions and the modification time in the metadata.

In this patch:
* add an activate/deactivate image method to StorageDomain
* implement the activate/deactivate image for BlockStorageDomain
* add the prepareImage/teardownImage methods in HSM

Change-Id: I4b5cb55a868cb9bf5a720610273af139ee858a1a
Reviewed-on: http://gerrit.ovirt.org/554
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#783098 - notFound should be and iterable and not an iterator.

Change-Id: If4074c6e87cff734bbae95074f5d2e4bc6ff8374
Reviewed-on: http://gerrit.ovirt.org/1139
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#782099 - Add vdsm modprobe configuration.

This is required as a hack around kernel bug 772317.

Change-Id: Iade53970c9975f732375634bd166825cefa87555
Reviewed-on: http://gerrit.ovirt.org/1106
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

getDevInfo: report partitioned devices, too

One cannot create a PV on a partitioned device, and therefor such
devices where not reported to Engine. This proved surprising to users
who wonder where their LUN disappeared.

Vdsm should report all devices, and ovirt-engine should mark partitioned
devices as unworthy of a PV. In the future, Vdsm may allow to forcefully
remove a partition table from a device, to make it usable as a PV.

Change-Id: Ia5eddfcb587aec5337adaef810bf11665a7dffde
Reviewed-on: http://gerrit.ovirt.org/944
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

hard-coding the value of se.StorageException.code is ugly

Change-Id: Icc7ea7427d3834344a89671caa526b7812a5eb2a
Reviewed-on: http://gerrit.ovirt.org/922
Tested-by: Dan Kenigsberg <danken@redhat.com>
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Douglas Schilling Landgraf <dougsland@redhat.com>

BZ#773371 Generate the VDSM certificates

VDSM is installed with ssl enabled by default, such configuration
requires the certificates to be generated when they're missing.

Change-Id: I68225e8cd58f6aecc487f570627d76bfe7060b22
Reviewed-on: http://gerrit.ovirt.org/1012
Tested-by: Haim Ateya <hateya@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Group the cluster lock code into a new SP method

Both the create and reconstructMaster methods were sharing a very
similar code for temporarly acquiring the cluster lock.

Change-Id: I981d84af7816ccd2db64eb5beb440e9f7637f734
Reviewed-on: http://gerrit.ovirt.org/603
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Federico Simoncelli <fsimonce@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Change device ownership for VM's drives.

Change-Id: I7859635b8abed67fe006886efd500eb9ac9f3b7c
Reviewed-on: http://gerrit.ovirt.org/904
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Support stable PCI address for uknown general devices

Change-Id: I664fc7326673ac0dcba487ed906c3daf652f8956
Reviewed-on: http://gerrit.ovirt.org/1040
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Support stable PCI address for controller devices

Change-Id: If69b39fd4d801043763160efbccafdc49b5eabf0
Reviewed-on: http://gerrit.ovirt.org/1039
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

BZ#770834 Handle complex interface names in nicSort

Change-Id: Ia7c6c139369b6a0a8a215662315299e5e70358a2
Reviewed-on: http://gerrit.ovirt.org/1104
Tested-by: Federico Simoncelli <fsimonce@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Move all the substitution at make time

Moving all the substitution at make time we can avoid the double pass
on the .in.in files.

Change-Id: I1f6956350971a88294559039e4ff00fce2637c78
Reviewed-on: http://gerrit.ovirt.org/1034
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Federico Simoncelli <fsimonce@redhat.com>

Support stable PCI addresses for video/sound/disk/nic devices.

- The 'getVmStats'/'getAllVmStats' verbs report the hash of the device part of domXml per VM.
- Return full vm's conf including device's addreses
- Handle new addresses parameter in vm's create flow

Change-Id: I5b3a97fa998d0f52ebe752024c62b8ec16c32c78
Reviewed-on: http://gerrit.ovirt.org/235
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>

Add 'readonly' attribute to disk's libvirt XML

Change-Id: I200cd36c7e36452019d9022789d2d3fb4303c4e2
Reviewed-on: http://gerrit.ovirt.org/1049
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>

Use drive[path] only after it is known

... and make isBlockDevice actually work. Thanks, xTs_w!

Change-Id: I2c43c0b0bc25c42c2aac4dd2e99f7931d46fabe0
Reviewed-on: http://gerrit.ovirt.org/1066
Reviewed-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#781970 kaxmlrpclib: fix plaintext transport in Python 2.7

Change-Id: I1b467cdd46d8380488ebd7225ca9a561c2c98f34
Reviewed-on: http://gerrit.ovirt.org/1065
Tested-by: Jakub Libosvar <libosvar@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

BZ#773364 cancel-migration: abort migration on dest results an exception

While starting migration, it appear in dest host vm list.
When trying to cancel migration the _dom have not set, and when trying
to call abortJob() we get AttributeError() exception.

Change-Id: I11d64e79f1a0cebde7a392af0d729ae0f192a844
Reviewed-on: http://gerrit.ovirt.org/1041
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Mark vds_bootstrap as executable in the Makefile

The bootstrap files should be always installed as executables (not only
in the rpm package).

Change-Id: I0dd77ef76174e66d39526ce3833d2470128b0e19
Reviewed-on: http://gerrit.ovirt.org/966
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#769502 Check if block device w/o invoking storage subsystem

...and fix a bug about SHAREDFS being considered as block domain.

Change-Id: Ia452c576b56f8d91fa1c6096237c2c150d746ac4
Reviewed-on: http://gerrit.ovirt.org/678
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#769502 Move limit configuration to its proper place

Currently limit configuration is managed part by vdsmd.init and part by
vdsm configuration. This is confusing an non standard. Putting a Vdsm
configuration file in limits.d will make admins life easier when looking
to change Vdsm's limits.

This may result in a surprise on upgrade - if someone has
max_open_files=nondefault on his vdsm.conf.

HOWEVER, these are very few and very technical users who could handle
(and would prefer) this sitting in limits.d

Change-Id: I26345e634e59512c47ce5a52179d65951787c5c6
Reviewed-on: http://gerrit.ovirt.org/914
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Throw key error in case on missing or already registered namespace

Change-Id: Ibb7fdaa87ea4db1d1d080ecccde43d248117f4d2
Reviewed-on: http://gerrit.ovirt.org/915
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#769502 - Make namespace registering more efficient

Change-Id: I6cc424984816bf578ff2860bb3276c7cd7fa894a
Reviewed-on: http://gerrit.ovirt.org/673
Reviewed-by: Ayal Baron <abaron@redhat.com>
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#781317 - adjust getos() to print real node type

Change-Id: Ia42fcf9570e096f7b79fed10c3a66d3fdf739cf6
Reviewed-on: http://gerrit.ovirt.org/1048
Tested-by: Haim Ateya <hateya@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

allow running old-style Vm without 'drives'

This has no effect on engine-initiated VMs which always have 'drives'
(or in the future, 'devices'). Only manual tests are affected.

Change-Id: I9f64ddfbf2596724bec624af161507bad98c2e63
Reviewed-on: http://gerrit.ovirt.org/1058
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

drop ugly dead function execAndGetOutput

Change-Id: Ice3082a0223f02bc513c7f15184bc922c00dba47
Reviewed-on: http://gerrit.ovirt.org/1050
Tested-by: Dan Kenigsberg <danken@redhat.com>
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>

Added copyright info to securable.py

Change-Id: Ied2d4963bd1843601d096144a8e08fb58b2f9aa1
Reviewed-on: http://gerrit.ovirt.org/1037
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#771329 Use a copy of the domainsToUpgrade

The original code just uses another ref that is changed while iterated
which has unexpected results. This was not apparent as the remove might
happened after the iteration because it might run in another thread.

To illustrate:

In [1]: a = [1,2,3,4,5,6,7,8,9,10]

In [2]: for i in a:
...:     a.remove(i)
...:

In [3]: a
Out[3]: [2, 4, 6, 8, 10]

In [4]: for i in a:
...:     a.remove(i)
...:

In [5]: a
Out[5]: [4, 8]

In [6]: for i in a[:]:
...:     a.remove(i)
...:

In [7]: a
Out[7]: []

Fun!

Change-Id: I3de0646af2fe9a9fcde708837cfcd83d10385026
Reviewed-on: http://gerrit.ovirt.org/900
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Raise ongoing creations count

Multi-core hosts are generally more capable to start up multiple qemu
processes efficiently. (And, the RHEL-5 kernel bug that limited
_ongoingCreations to 1 is long irrelevant)

Change-Id: I83b4f956619ee6cdfde5fdbec9d52484c355b33f
Reviewed-on: http://gerrit.ovirt.org/679
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#771686 fix MAX_VLAN_ID

Change-Id: I6ddc533d18cbc2411fd4f1b26397eb344419def4
Reviewed-on: http://gerrit.ovirt.org/1033
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#773666 - Fix HSM flows should not change rw permissions.

Changing permissions on LV/File is a clustered operation and
should never be done in HSM.
Prepare is used by HSM therefore should not change permissions.
Minimal fix needs further development.

Change-Id: Ia53f0008cf3e7b25e14c7f03798c1ae527056725
Reviewed-on: http://gerrit.ovirt.org/797
Reviewed-by: Ayal Baron <abaron@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#772556 bootstrap: ignore restorecon errors

This is particularly painful if SELinux is disabled.

Change-Id: I0af8e5f68c2d1ec52208684b95bcff2e68ba88fe
Reviewed-on: http://gerrit.ovirt.org/1022
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#772670 [rhevh] network is not set after using boot params

When using vdsm-reg we are adding/deleting network before libvirt starts.
Added a "hidden" parameter to add/delNetwork if we want to skip libvirt's
add-network.
When vdsm is up it is syncing the libvirt network with host bridge.

Change-Id: I87e425c684a187e5dd874d11646aa7acf676e0a2
Reviewed-on: http://gerrit.ovirt.org/1024
Reviewed-by: Michael Burns <mburns@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Set the drive sizes to '0' as default

In some situations (eg: empty cdrom drive) the 'reqsize' and the
'apparentsize' arguments are missing and they should be considered '0'.

Change-Id: Iba8168a40c9b2d27c60788f5f97d3f7551df2781
Reviewed-on: http://gerrit.ovirt.org/990
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#772591 Specify which user/group should be used for core dumps rotation

New 'su' directive is new to logrotate 3.8.0. This patch breaks systems
with older logrotate.

Change-Id: I38738f494e9bd241ea27e6e18b4d348fc3dfe643
Reviewed-on: http://gerrit.ovirt.org/971
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Support specParams parameter as dictionary per device

Change-Id: I0e584931d77c2f64ee5a1d613d6b73e6d0397ba7
Reviewed-on: http://gerrit.ovirt.org/979
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

All size parameters in vm's conf should be strings and not integers

Change-Id: Ief705e67908ee5a2dbe6a51959e47789d78aaa16
Reviewed-on: http://gerrit.ovirt.org/978
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Don't try to schedule SPM tasks on HSM

When the manager tries to run an SPM task on an HSM host the returned
(and logged) error is not clear:

Traceback (most recent call last):
[...]
  File "/usr/share/vdsm/storage/hsm.py", line 571, in _spmSchedule
    self.taskMng.scheduleJob("spm", pool.tasksDir, vars.task, name,
                             func, *args)
AttributeError: 'StoragePool' object has no attribute 'tasksDir'

VDSM should verify to be an SPM before trying to schedule any spm
job.

Change-Id: I7183fb2a7ef391a73cbb1a6dff9bf46238275100
Reviewed-on: http://gerrit.ovirt.org/695
Tested-by: Federico Simoncelli <fsimonce@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Don't change the name of a secured class

The variable 'name' used to name the secured classes was overridden
by the 'name' variable used to cycle over the methods.
The result was that the secured classes were renamed to their last
method name, eg: storage.sp.StoragePool => storage.sp.setMaxHostId.

Change-Id: I7d34b14391a0940e1ca73b8d736877f49b1c47f5
Reviewed-on: http://gerrit.ovirt.org/974
Tested-by: Federico Simoncelli <fsimonce@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Fix warning: File listed twice

Change-Id: I9279838367bca753190108130d10ea1d77aaf6df
Reviewed-on: http://gerrit.ovirt.org/951
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Remove unnecessary validates.

All this instances race with the "recover" state of the VDSM,
if disk info is required when the StoragePool is not reconnected
yet.

Change-Id: I55482fc5ce78d68d5af7ba62c92d253e19f45e6b
Reviewed-on: http://gerrit.ovirt.org/801
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Haim Ateya <hateya@redhat.com>

Remove unnecessary validation.

This check prevents VDSM to query info about a disk on a running
VM after VDSM restart.

Change-Id: I3b9c89402f28614a893c9c4a1d8abbb55d9eaf1f
Reviewed-on: http://gerrit.ovirt.org/800
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>

Fixing unnecessary preparePath() call on recover.

On behalf of Dan Kenisberg.

Change-Id: Ifca49f960d80176d5ddc594a2ae269cdaf13dec5
Reviewed-on: http://gerrit.ovirt.org/950
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

For BC we should to keep running VM run after vdsm upgrade.

So, because this vm doesn't have normalize conf we need to build it
in recovery flow

Change-Id: I67d555b1b75e3287d083ccde6e1f0869f09f3234
Reviewed-on: http://gerrit.ovirt.org/949
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>

Fix sendExtendMsg declaration, should be public

Change-Id: Idc060de51daf573b462c2f4ade789fd884e627a6
Reviewed-on: http://gerrit.ovirt.org/948
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>

Move domain monitor out of sp.py

Domain monitoring logic should be seperated from the core pool logic to
reduce StoragePool class complexity

Change-Id: I18f30541a3466c532b2fd6144cd54b46545aff0c
Reviewed-on: http://gerrit.ovirt.org/190
Tested-by: Saggi Mizrahi <smizrahi@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Make exporting more pythonic

This seperated the dispatcher from hsm and allows us to add propertoes
to all public function with ease

Change-Id: I4e26da2a53bae5e897cebdb41af9abe519760aca
Reviewed-on: http://gerrit.ovirt.org/208
Tested-by: Saggi Mizrahi <smizrahi@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Move reconnect file existence check

Move the check to the inside of the _saveReconnectFunction show it's
always recreated.

Change-Id: I72abecc0dab58cd008d0afe8046d2b3c8f898517
Reviewed-on: http://gerrit.ovirt.org/917
Tested-by: Saggi Mizrahi <smizrahi@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Fix getMasterDomain masterVersion handling

If getMasterDomain() gets -1 as the expected masterVersion it should
skip the masterVersion equality test and accept whatever version is in
the MD.

Change-Id: I2e5dd06800e770c098460466fa8642c5bc928e54
Reviewed-on: http://gerrit.ovirt.org/431
Tested-by: Saggi Mizrahi <smizrahi@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Fix logging formatting in resourceManager

Change-Id: I08b530e8fbf66b9b8326cf1733df6a3e4f23ef1a
Reviewed-on: http://gerrit.ovirt.org/531
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#674010 - Fix name error in vdsClient - 'commands' not found

Change-Id: I091b32ff19d59ec7036ee440ee60913c06be4c13
Reviewed-on: http://gerrit.ovirt.org/906
Tested-by: Haim Ateya <hateya@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Support new type of vmParams parameter for VM create.

- Normalize old vmParams to support new type that includes 'devices' key
- Return normalized conf including 'devices'
- Add new general Device class
- Add subclasses for each device: Drive, NetworkInterface, Sound, Video

Change-Id: I26a9bf46a23f5c6d92ea3fea163ac23dcac21df8
Reviewed-on: http://gerrit.ovirt.org/788
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

new verb: migration cancel

the verb migrateCancel will take vmid and try to stop migration,
if there is no migration in progress it will return error.

when migrateCancel stop the migration progress, the migrate verb will
return error "Migration canceled"

Change-Id: I4f503c444c4ca1469c18480d6ca29fba8f011d4c
Reviewed-on: http://gerrit.ovirt.org/480
Tested-by: Shahar Havivi <shavivi@redhat.com>
Reviewed-by: Peter V. Saveliev <peet@redhat.com>
Tested-by: Peter V. Saveliev <peet@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

vdsm_reg: rename rhevm.py to engine.py

Renamed the file rhevm to engine.
BZ#771414

Change-Id: I2a49c22a3ebfcf31fd5653a1e7fdd88d53c8f7a5
Reviewed-on: http://gerrit.ovirt.org/899
Tested-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

rhevm.py: replace RHEV-M string to oVirt-Engine

Replace RHEV-M string to oVirt Engine
BZ#771414

Change-Id: Ic52d2d508b3a5f42e1341df5532ba5974cd14d5f
Reviewed-on: http://gerrit.ovirt.org/871
Tested-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Removed useless docs from lvm.py

Man pages exist for a reason.

Keeping copies of documentation in the code is necessary cry and will only
grow stale in time.

To see the list of fields run:
$ lvm lvs -o help

Change-Id: I70975ba7e7494a6d6e62acdfed8cd298f1f6f278
Reviewed-on: http://gerrit.ovirt.org/680
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Fix floppy drives: bus should be 'fdc'.

Change-Id: I912cc1418e0df32d44bbc1a7c79c496aebd86c36
Reviewed-on: http://gerrit.ovirt.org/787
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

Compile python files using automake

In this patch:
* Use the PYTHON macro to compile python files
* Remove the redhat-rpm-config build dependency
* Move the vdsm sos plugin into its own directory to avoid the custom
  file renaming and to use the PYTHON macro
* Update the gitignore file and the copyright in the touched files
* Add the all-local target to force the substitutions in the python
  files
* Unify the macro substitution in build-aux/Makefile.subs

Change-Id: I7dce5c317605b84ec883feb04bfc83c5f64d921b
Reviewed-on: http://gerrit.ovirt.org/708
Tested-by: Federico Simoncelli <fsimonce@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

vdsm-reg.conf.in: remove vdc_authkeys_path

The auth key is defined in deployUtil.py.in
This conf value has never been used.

Change-Id: I1ed0cb70d3d77bc00c3dec8cd263968bbd37622e
Reviewed-on: http://gerrit.ovirt.org/798
Tested-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

BZ#769179 - Omit getSessionList from vdsClient and its dependent files

Change-Id: Ia448f4e8c2720b5958fef59f4f55b2f4ae63d2ac
Reviewed-on: http://gerrit.ovirt.org/659
Reviewed-by: Eduardo <ewarszaw@redhat.com>
Tested-by: Haim Ateya <hateya@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

vdsmd.8: update man page to reference oVirt

Replace any remaining reference from RHEV to oVirt

Change-Id: I7226817576b713490cdeb098356aee8894af2bfe
Reviewed-on: http://gerrit.ovirt.org/550
Tested-by: Douglas Schilling Landgraf <dougsland@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

vdsm-reg: change registration servlet name

Note that this change prohibits registration of Vdsm to RHEV-M-3.0.
It depends on ovirt-engine change I4f36a302adbafdc5f9d0bb219896c794cbad31dd
BZ# 752464.

Change-Id: I743bfc8258a7cf8fd032cde13aadd2827ed9ea29
Reviewed-on: http://gerrit.ovirt.org/310
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

replace default bridge name from brVDSM to ovirtmgmt

Change-Id: I8115192b65b431e8066313bda357f8aa86880c86
Reviewed-on: http://gerrit.ovirt.org/772
Reviewed-by: Doron Fediuck <dfediuck@redhat.com>
Tested-by: Doron Fediuck <dfediuck@redhat.com>
Reviewed-by: Dan Kenigsberg <danken@redhat.com>

deployUtil: replace rhevm cert to engine

Note that this patch breaks bootstrap for RHEV-M-3.0

Change-Id: I945838b1ad59414a382a6419875c6fdb79d4d567
Reviewed-on: http://gerrit.ovirt.org/769
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

rename rhevm bridge to brVDSM

Remove rhevm as bridge and use brVDSM
http://bugzilla.redhat.com/753879
http://bugzilla.redhat.com/753571

Change-Id: Id1c98658af80ad52a14934a962be4945c91ff48d
Reviewed-on: http://gerrit.ovirt.org/311
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Add Fedora support to vdsm-bootstrap

while (hopefully) maintaining RHEL-5 support, and leaving bridge name
games to a following patch.

Change-Id: I17e9fd3550c5db5865b88ef4cfc93953b99827b6
Reviewed-on: http://gerrit.ovirt.org/767
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

make vds_bootstrap executable again

while keeping rpmlint happy.

Change-Id: I43926966e963aee96e9a5b5ededb2974ee0196fa
Reviewed-on: http://gerrit.ovirt.org/765
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Handle EAGAIN and EINTR in ProcessPool.runExternally

Change-Id: Id1d0f35af9db6e237c87b00438b9ce3ded9faa8f
Reviewed-on: http://gerrit.ovirt.org/200
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Add NoIntr utility class

Change-Id: Ib8cd7f2c6ef8c4845fca617a0423b922fc81cf0e
Reviewed-on: http://gerrit.ovirt.org/199
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

create /rhev repository

Currently, the repository /rhev is not created automatically by
vdsm.spec.in. Without this directory, vdsm doesn't work correctly
and vdsm will throw errors messages like:
[Errno 30] Read-only file system: '/rhev' or even 'permission denied'.

Change-Id: I971a7b47e834d7b6b95874ff6437e66b1928fc82
Reviewed-on: http://gerrit.ovirt.org/548
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Douglas Schilling Landgraf <dougsland@redhat.com>

Add protocol version selection support for the NFS protocoll

Change-Id: I64537f9170d17fcbac3b357ec0bfdc0dc56e292f
Reviewed-on: http://gerrit.ovirt.org/681
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Support changing retransmission number and timeout for nfs mounts

Parameter validity is done by mount. Valid ranges should be tested for
and implemented in the UI.

Change-Id: I25e4a6f6daa5b14774e03230784b3b2dbe4630e0
Reviewed-on: http://gerrit.ovirt.org/636
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Add SharedFS Support

This adds a new connection type called sharedfs (6). This is a way for
users to try and use unsupported NAS storage as domains. This is
considered bad practice and completely unsupported.

The connection dict should be:
{"id": <ID>,
 "connection": <MOUNT SPEC> eg. server:/export
 "vfs_type": <MOUNT VFS TYPE> eg. nfs
 "mnt_options": <MOuNT OPTIONS> eg. user=0,noatime

Change-Id: I1ea9620f9179b415152d3e32509b5ebe02d8d77b
Reviewed-on: http://gerrit.ovirt.org/559
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Minor refactoring

Change-Id: I0b65e9714fb80d048e1ed244bb961ff8c2e205fd
Reviewed-on: http://gerrit.ovirt.org/558
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Rename *FileServer methods to *NFSServer to prevent confusion

Change-Id: I6a016def65a819c6478b78d569f647aeb5f05dd5
Reviewed-on: http://gerrit.ovirt.org/557
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

Rearange code in storage_connection to minimize code duplication

Change-Id: If3e28c064043f19cefccde13decf6699d2718634
Reviewed-on: http://gerrit.ovirt.org/556
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

BZ#736034 - Add metadataignore switch to pvcreate.

We would like to pre-allocate metadata space in all PVs
used for our VGs. However, we are using only the metadata
of the first PV.
This is important for an occasion where the first PV has
to be removed and the metadata it carries has to be copied
into somewhere else.

Could not have been done without Haim's help.

Change-Id: Ib040902610fd2c5b0de04de95b57f0bfbdd1e11e
Reviewed-on: http://gerrit.ovirt.org/711
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Tested-by: Dan Kenigsberg <danken@redhat.com>

storage_connection: unexpected keyword 'timeout'

remove from mount.Mount() timeout argument
added to umount() and mount() the timeout argument

vdsm log:
==========
Thread-1265:INFO::2011-12-23 17:47:54,285::storage_connection::128::Storage.ServerConnection:(validae(Request to
validate NFS storage server)
Thread-1265:ERROR:2011-12-23 18:47:54,285::storage_connection::291::StoragerServerConnection::(__validadeFileServer)
Error during storage connection validation: __init__() got an unexpected keyword argument 'timeout'
Traceback (most recent call last):
File "/usr/share/vdsm/storage/storage_connection.py", line 264, in __validateFileServer
TypeError: __init()__ got an unexpected keyword argument 'timeout'

Change-Id: Ic64bf1bd366d1e9cb4c80c50b85a57ae022374cd
Reviewed-on: http://gerrit.ovirt.org/734
Reviewed-by: Dan Kenigsberg <danken@redhat.com>
Reviewed-by: Igor Lvovsky <ilvovsky@redhat.com>
Tested-by: Igor Lvovsky <ilvovsky@redhat.com>