- Kimchi-devel - oVirt List Archives

[PATCH] [Kimchi] Add 'Current' to CPU Number/CPUs where needed
by Socorro Stoppler 23 Mar '16

23 Mar '16

Adding 'Current' to CPU Number/CPUs per feedback in bugzilla# 133521 Socorro Stoppler (1): Change CPU Number/CPUs to Current CPU Number/Current CPUs per feedback from bugzilla# 133521 ui/pages/guest-add.html.tmpl | 2 +- ui/pages/tabs/templates.html.tmpl | 6 +++--- ui/pages/template-edit.html.tmpl | 2 +- 3 files changed, 5 insertions(+), 5 deletions(-) -- 1.9.1

3 3

[PATCH] [Wok] Implement WokMessage class
by Lucio Correia 23 Mar '16

23 Mar '16

WokMessage is a generic message implementation, which looks up and translates error messages for all plugins and Wok itself. Lookup and translation code was adapted from WokException class, which now just uses WokMessage. WokMessage is intended to be used by other modules dealing with messages, such as User Request Log and future feature Asynchronous Notifications. Signed-off-by: Lucio Correia <luciojhc(a)linux.vnet.ibm.com> --- src/wok/exception.py | 45 +++----------------------------- src/wok/message.py | 73 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 77 insertions(+), 41 deletions(-) create mode 100644 src/wok/message.py diff --git a/src/wok/exception.py b/src/wok/exception.py index 023334b..52f007e 100644 --- a/src/wok/exception.py +++ b/src/wok/exception.py @@ -20,53 +20,16 @@ # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA import cherrypy -import gettext -from wok.i18n import messages as _messages -from wok.template import get_lang, validate_language +from wok.message import WokMessage class WokException(Exception): def __init__(self, code='', args={}): self.code = code - - for key, value in args.iteritems(): - if isinstance(value, unicode): - continue - - # value is not unicode: convert it - try: - # In case the value formats itself to an ascii string. - args[key] = unicode(str(value), 'utf-8') - except UnicodeEncodeError: - # In case the value is a WokException or it formats - # itself to a unicode string. - args[key] = unicode(value) - - # First, check if it is a Wok error message, then search in plugin - # error messages list - msg = _messages.get(code, code) - if (msg == code) and (cherrypy.request.app): - msg = self._get_translation() - - msg = unicode(msg, 'utf-8') % args - pattern = "%s: %s" % (code, msg) - cherrypy.log.error_log.error(pattern) - Exception.__init__(self, pattern) - - def _get_translation(self): - lang = validate_language(get_lang()) - paths = cherrypy.request.app.root.paths - domain = cherrypy.request.app.root.domain - messages = cherrypy.request.app.root.messages - text = messages.get(self.code, self.code) - - try: - translation = gettext.translation(domain, paths.mo_dir, [lang]) - except: - translation = gettext - - return translation.gettext(text) + msg = WokMessage(code, args).get_text() + cherrypy.log.error_log.error(msg) + Exception.__init__(self, msg) class NotFoundError(WokException): diff --git a/src/wok/message.py b/src/wok/message.py new file mode 100644 index 0000000..e982fba --- /dev/null +++ b/src/wok/message.py @@ -0,0 +1,73 @@ +# +# Project Wok +# +# Copyright IBM Corp, 2015-2016 +# +# Code derived from Project Kimchi +# +# This library is free software; you can redistribute it and/or +# modify it under the terms of the GNU Lesser General Public +# License as published by the Free Software Foundation; either +# version 2.1 of the License, or (at your option) any later version. +# +# This library is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# Lesser General Public License for more details. +# +# You should have received a copy of the GNU Lesser General Public +# License along with this library; if not, write to the Free Software +# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + +import cherrypy +import gettext + +from wok.template import get_lang, validate_language + + +class WokMessage(object): + def __init__(self, code='', args={}, plugin=None): + # make all args unicode + for key, value in args.iteritems(): + if isinstance(value, unicode): + continue + + try: + # In case the value formats itself to an ascii string. + args[key] = unicode(str(value), 'utf-8') + except UnicodeEncodeError: + # In case the value is a WokException or it formats + # itself to a unicode string. + args[key] = unicode(value) + + self.code = code + self.args = args + self.plugin = plugin + + def _get_translation(self): + # get app from plugin path if specified + if self.plugin: + app = cherrypy.tree.apps[self.plugin] + # if on request, try to get app from it + elif cherrypy.request.app: + app = cherrypy.request.app + # fallback: get root app (WokRoot) + else: + app = cherrypy.tree.apps[''] + + domain = app.root.domain + paths = app.root.paths + text = app.root.messages.get(self.code, self.code) + lang = validate_language(get_lang()) + + try: + translation = gettext.translation(domain, paths.mo_dir, [lang]) + except: + translation = gettext + + return translation.gettext(text) + + def get_text(self): + msg = self._get_translation() + msg = unicode(msg, 'utf-8') % self.args + return "%s: %s" % (self.code, msg) -- 1.9.1

3 2

[PATCH][Kimchi] Fix memory value return when hotplug memory devs
by Rodrigo Trujillo 23 Mar '16

23 Mar '16

There is an issue when Kimchi returns the value of memory. It is returning the current memory value, instead of the total memory, which includes the value of the memory devices added. This patch fix this problem returning the total value from "dom.info". Signed-off-by: Rodrigo Trujillo <rodrigo.trujillo(a)linux.vnet.ibm.com> --- model/vms.py | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/model/vms.py b/model/vms.py index e3bc266..396758d 100644 --- a/model/vms.py +++ b/model/vms.py @@ -1228,7 +1228,17 @@ class VMModel(object): unit = 'KiB' memory = convert_data_size(val, unit, 'MiB') else: - memory = info[2] >> 10 + # If guest is running we need information in real time, in order + # to get real value if memory devices were added: + # + # Class Libvirt::Domain::Info + # cpu_time[R] + # max_mem[R] -> current memory plus memory devs (use it) + # memory[R] -> current memory only, does not include mem devs + # nr_virt_cpu[R] + # state[R] + # + memory = info[1] >> 10 # assure there is no zombie process left for proc in self._serial_procs[:]: -- 2.1.0

2 1

[PATCH v2][Kimchi 1/2] Issue #838: Extend memory hotplug feature test
by Rodrigo Trujillo 23 Mar '16

23 Mar '16

If QEMU version is < 2.1, qemu is not going to start guest which have maxmemory tag configured in the XML. Kimchi adds this tag by default in order to allow memory hotplug. In other words, in order to allow memory hotplug, versions must be libvirt >= 1.2.14 and qemu >= 2.1. Libvirt was already being tested by the feature test, but qemu did not. Distros as Red Hat 7 and CentOS still are in qemu 1.5. This patch extend the feature test and do not create maxmemory tag in order to avoid problems in those distros. Signed-off-by: Rodrigo Trujillo <rodrigo.trujillo(a)linux.vnet.ibm.com> --- i18n.py | 2 +- model/featuretests.py | 20 +++++++++++++------- model/vms.py | 11 ++++++++++- tests/test_model.py | 5 +++-- vmtemplate.py | 4 +++- 5 files changed, 30 insertions(+), 12 deletions(-) diff --git a/i18n.py b/i18n.py index beefeb5..6214687 100644 --- a/i18n.py +++ b/i18n.py @@ -99,7 +99,7 @@ messages = { "KCHVM0043E": _("Only increase memory is allowed in active VMs"), "KCHVM0044E": _("For live memory update, new memory value must be equal old memory value plus multiples of 1024 Mib"), "KCHVM0045E": _("There are not enough free slots of 1024 Mib in the guest."), - "KCHVM0046E": _("Host's libvirt version does not support memory devices. Libvirt must be >= 1.2.14"), + "KCHVM0046E": _("Host's libvirt or qemu version does not support memory devices and memory hotplug. Libvirt must be >= 1.2.14 and QEMU must be >= 2.1."), "KCHVM0047E": _("Error attaching memory device. Details: %(error)s"), "KCHVM0048E": _("Cannot start %(name)s. Virtual machine is already running."), "KCHVM0049E": _("Cannot power off %(name)s. Virtual machine is shut off."), diff --git a/model/featuretests.py b/model/featuretests.py index 378335a..ba728e1 100644 --- a/model/featuretests.py +++ b/model/featuretests.py @@ -72,6 +72,14 @@ MAXMEM_VM_XML = """ <type arch='%(arch)s'>hvm</type> <boot dev='hd'/> </os> + <cpu> + <numa> + <cell id='0' cpus='0' memory='10240' unit='KiB'/> + </numa> + </cpu> + <features> + <acpi/> + </features> </domain>""" DEV_MEM_XML = """ @@ -209,13 +217,9 @@ class FeatureTests(object): A memory device can be hot-plugged or hot-unplugged since libvirt version 1.2.14. ''' - # Libvirt < 1.2.14 does not support memory devices, so firstly, check - # its version, then try to attach a device. These steps avoid errors - # with Libvirt 'test' driver for KVM - version = 1000000*1 + 1000*2 + 14 - if libvirt.getVersion() < version: - return False - + # Libvirt < 1.2.14 does not support memory devices, so try to attach a + # device. Then check if QEMU (>= 2.1) supports memory hotplug, starting + # the guest These steps avoid errors with Libvirt 'test' driver for KVM with RollbackContext() as rollback: FeatureTests.disable_libvirt_error_logging() rollback.prependDefer(FeatureTests.enable_libvirt_error_logging) @@ -230,6 +234,8 @@ class FeatureTests(object): try: dom.attachDeviceFlags(DEV_MEM_XML, libvirt.VIR_DOMAIN_MEM_CONFIG) + dom.create() + rollback.prependDefer(dom.destroy) return True except libvirt.libvirtError: return False diff --git a/model/vms.py b/model/vms.py index 03ffdae..e3bc266 100644 --- a/model/vms.py +++ b/model/vms.py @@ -170,7 +170,8 @@ class VMsModel(object): stream_protocols = self.caps.libvirt_stream_protocols xml = t.to_vm_xml(name, vm_uuid, libvirt_stream_protocols=stream_protocols, - graphics=graphics) + graphics=graphics, + mem_hotplug_support = self.caps.mem_hotplug_support) cb('Defining new VM') try: @@ -809,6 +810,14 @@ class VMModel(object): return (nonascii_name if nonascii_name is not None else vm_name, dom) def _update_memory_config(self, xml, params): + # Cannot pass max memory if there is not support to memory hotplug + # Then set max memory as memory, just to continue with the update + if not self.caps.mem_hotplug_support: + if 'maxmemory' in params['memory']: + raise InvalidOperation("KCHVM0046E") + else: + params['memory']['maxmemory'] = params['memory']['current'] + root = ET.fromstring(xml) # MiB to KiB hasMem = 'current' in params['memory'] diff --git a/tests/test_model.py b/tests/test_model.py index b461172..c91ba8f 100644 --- a/tests/test_model.py +++ b/tests/test_model.py @@ -912,13 +912,14 @@ class ModelTests(unittest.TestCase): # rename and increase memory when vm is not running params = {'name': u'пeω-∨м', - 'memory': {'current': 2048, - 'maxmemory': 2048}} + 'memory': {'current': 2048}} inst.vm_update('kimchi-vm1', params) rollback.prependDefer(utils.rollback_wrapper, inst.vm_delete, u'пeω-∨м') self.assertEquals(info['uuid'], inst.vm_lookup(u'пeω-∨м')['uuid']) info = inst.vm_lookup(u'пeω-∨м') + # Max memory is returned, add to test + params['memory']['maxmemory'] = 2048 for key in params.keys(): self.assertEquals(params[key], info[key]) diff --git a/vmtemplate.py b/vmtemplate.py index 4bcf324..653ad02 100644 --- a/vmtemplate.py +++ b/vmtemplate.py @@ -376,7 +376,9 @@ class VMTemplate(object): # Rearrange memory parameters params['memory'] = self.info['memory'].get('current') params['max_memory'] = "" - if memory != maxmemory: + # if there is not support to memory hotplug in Libvirt or qemu, we + # cannot add the tag maxMemory + if memory != maxmemory and kwargs.get('mem_hotplug_support', True): maxmem_xml = "<maxMemory slots='%s' unit='MiB'>%s</maxMemory>" params['max_memory'] = maxmem_xml % (slots, maxmemory) -- 2.1.0

2 2

[PATCH][Kimchi 0/6] Use a single field to create a template
by Ramon Medeiros 23 Mar '16

23 Mar '16

Instead of specify if the media is cdrom or disk, use source_media to create a template Ramon Medeiros (6): Create a single field to pass the installation media Method to retrieve stored templates at object store Fix checking duplicate template before creating it Identify installation media while creating template Only use source_media as boot media Update tests API.json | 5 ++ i18n.py | 2 +- model/templates.py | 13 ++-- tests/test_authorization.py | 4 +- tests/test_livemigration.py | 7 +- tests/test_mockmodel.py | 14 ++-- tests/test_model.py | 69 ++++++++++++-------- tests/test_rest.py | 32 ++++------ tests/test_template.py | 39 +++++------- tests/test_vmtemplate.py | 24 +++---- utils.py | 30 +++++++++ vmtemplate.py | 152 ++++++++++++++++++++++++++++++++++++++++---- 12 files changed, 279 insertions(+), 112 deletions(-) -- 2.5.0

2 9

[PATCH] [Kimchi] Adding wok version dependency
by dhbarboza82＠gmail.com 23 Mar '16

23 Mar '16

From: Daniel Henrique Barboza <dhbarboza82(a)gmail.com> This patch adds WoK dependency >= 2.1.0 for the upcoming Kimchi 2.1.0 release packaging. Signed-off-by: Daniel Henrique Barboza <dhbarboza82(a)gmail.com> --- contrib/DEBIAN/control.in | 2 +- contrib/kimchi.spec.fedora.in | 2 +- contrib/kimchi.spec.suse.in | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/contrib/DEBIAN/control.in b/contrib/DEBIAN/control.in index b6acdaf..6cfd8ee 100644 --- a/contrib/DEBIAN/control.in +++ b/contrib/DEBIAN/control.in @@ -3,7 +3,7 @@ Version: @PACKAGE_VERSION@ Section: base Priority: optional Architecture: all -Depends: wok, +Depends: wok (>= 2.1.0), ginger-base, python-imaging, python-configobj, diff --git a/contrib/kimchi.spec.fedora.in b/contrib/kimchi.spec.fedora.in index 292ad03..29d4463 100644 --- a/contrib/kimchi.spec.fedora.in +++ b/contrib/kimchi.spec.fedora.in @@ -7,7 +7,7 @@ BuildArch: noarch Group: System Environment/Base License: LGPL/ASL2 Source0: %{name}-%{version}.tar.gz -Requires: wok +Requires: wok >= 2.1.0 Requires: ginger-base Requires: qemu-kvm Requires: gettext diff --git a/contrib/kimchi.spec.suse.in b/contrib/kimchi.spec.suse.in index 7a57538..c12bcab 100644 --- a/contrib/kimchi.spec.suse.in +++ b/contrib/kimchi.spec.suse.in @@ -7,7 +7,7 @@ BuildArch: noarch Group: System Environment/Base License: LGPL/ASL2 Source0: %{name}-%{version}.tar.gz -Requires: wok +Requires: wok >= 2.1.0 Requires: ginger-base Requires: kvm Requires: gettext-tools -- 2.5.0

2 1

[PATCH][Kimchi 1/2] Issue #838: Extend memory hotplug feature test
by Rodrigo Trujillo 23 Mar '16

23 Mar '16

If QEMU version is < 2.1, qemu is not going to start guest which have maxmemory tag configured in the XML. Kimchi adds this tag by default in order to allow memory hotplug. In other words, in order to allow memory hotplug, versions must be libvirt >= 1.2.14 and qemu >= 2.1. Libvirt was already being tested by the feature test, but qemu did not. Distros as Red Hat 7 and CentOS still are in qemu 1.5. This patch extend the feature test and do not create maxmemory tag in order to avoid problems in those distros. Signed-off-by: Rodrigo Trujillo <rodrigo.trujillo(a)linux.vnet.ibm.com> --- i18n.py | 2 +- model/featuretests.py | 20 +++++++++++++------- model/vms.py | 8 ++++++++ vmtemplate.py | 7 ++++++- 4 files changed, 28 insertions(+), 9 deletions(-) diff --git a/i18n.py b/i18n.py index beefeb5..6214687 100644 --- a/i18n.py +++ b/i18n.py @@ -99,7 +99,7 @@ messages = { "KCHVM0043E": _("Only increase memory is allowed in active VMs"), "KCHVM0044E": _("For live memory update, new memory value must be equal old memory value plus multiples of 1024 Mib"), "KCHVM0045E": _("There are not enough free slots of 1024 Mib in the guest."), - "KCHVM0046E": _("Host's libvirt version does not support memory devices. Libvirt must be >= 1.2.14"), + "KCHVM0046E": _("Host's libvirt or qemu version does not support memory devices and memory hotplug. Libvirt must be >= 1.2.14 and QEMU must be >= 2.1."), "KCHVM0047E": _("Error attaching memory device. Details: %(error)s"), "KCHVM0048E": _("Cannot start %(name)s. Virtual machine is already running."), "KCHVM0049E": _("Cannot power off %(name)s. Virtual machine is shut off."), diff --git a/model/featuretests.py b/model/featuretests.py index 378335a..ba728e1 100644 --- a/model/featuretests.py +++ b/model/featuretests.py @@ -72,6 +72,14 @@ MAXMEM_VM_XML = """ <type arch='%(arch)s'>hvm</type> <boot dev='hd'/> </os> + <cpu> + <numa> + <cell id='0' cpus='0' memory='10240' unit='KiB'/> + </numa> + </cpu> + <features> + <acpi/> + </features> </domain>""" DEV_MEM_XML = """ @@ -209,13 +217,9 @@ class FeatureTests(object): A memory device can be hot-plugged or hot-unplugged since libvirt version 1.2.14. ''' - # Libvirt < 1.2.14 does not support memory devices, so firstly, check - # its version, then try to attach a device. These steps avoid errors - # with Libvirt 'test' driver for KVM - version = 1000000*1 + 1000*2 + 14 - if libvirt.getVersion() < version: - return False - + # Libvirt < 1.2.14 does not support memory devices, so try to attach a + # device. Then check if QEMU (>= 2.1) supports memory hotplug, starting + # the guest These steps avoid errors with Libvirt 'test' driver for KVM with RollbackContext() as rollback: FeatureTests.disable_libvirt_error_logging() rollback.prependDefer(FeatureTests.enable_libvirt_error_logging) @@ -230,6 +234,8 @@ class FeatureTests(object): try: dom.attachDeviceFlags(DEV_MEM_XML, libvirt.VIR_DOMAIN_MEM_CONFIG) + dom.create() + rollback.prependDefer(dom.destroy) return True except libvirt.libvirtError: return False diff --git a/model/vms.py b/model/vms.py index 03ffdae..557d37a 100644 --- a/model/vms.py +++ b/model/vms.py @@ -809,6 +809,14 @@ class VMModel(object): return (nonascii_name if nonascii_name is not None else vm_name, dom) def _update_memory_config(self, xml, params): + # Cannot pass max memory if there is not support to memory hotplug + # Then set max memory as memory, just to continue with the update + if not self.caps.mem_hotplug_support: + if 'maxmemory' in params['memory']: + raise InvalidOperation("KCHVM0046E") + else: + params['memory']['maxmemory'] = params['memory']['current'] + root = ET.fromstring(xml) # MiB to KiB hasMem = 'current' in params['memory'] diff --git a/vmtemplate.py b/vmtemplate.py index 4bcf324..535efd7 100644 --- a/vmtemplate.py +++ b/vmtemplate.py @@ -32,6 +32,7 @@ from wok.exception import MissingParameter, OperationFailed from wok.plugins.kimchi import imageinfo from wok.plugins.kimchi import osinfo from wok.plugins.kimchi.isoinfo import IsoImage +from wok.plugins.kimchi.model.config import CapabilitiesModel from wok.plugins.kimchi.utils import check_url_path, pool_name_from_uri from wok.plugins.kimchi.xmlutils.cpu import get_cpu_xml from wok.plugins.kimchi.xmlutils.disk import get_disk_xml @@ -118,6 +119,8 @@ class VMTemplate(object): disk_info['index'] = disk_info.get('index', index) self.info['disks'][index] = disk_info + self.caps = CapabilitiesModel() + def _get_os_info(self, args, scan): distro = version = 'unknown' @@ -376,7 +379,9 @@ class VMTemplate(object): # Rearrange memory parameters params['memory'] = self.info['memory'].get('current') params['max_memory'] = "" - if memory != maxmemory: + # if there is not support to memory hotplug in Libvirt or qemu, we + # cannot add the tag maxMemory + if memory != maxmemory and self.caps.mem_hotplug_support: maxmem_xml = "<maxMemory slots='%s' unit='MiB'>%s</maxMemory>" params['max_memory'] = maxmem_xml % (slots, maxmemory) -- 2.1.0

3 5

[PATCH] [Wok] Issue #41: encode of ident should only happen if it is of type unicode.
by archus＠linux.vnet.ibm.com 23 Mar '16

23 Mar '16

From: Archana Singh <archus(a)linux.vnet.ibm.com> Added isinstance unicode check of ident before encoding it. Signed-off-by: Archana Singh <archus(a)linux.vnet.ibm.com> --- src/wok/control/base.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/wok/control/base.py b/src/wok/control/base.py index be2e35e..25f6522 100644 --- a/src/wok/control/base.py +++ b/src/wok/control/base.py @@ -1,3 +1,4 @@ +# -*- coding: utf-8 -*- # # Project Wok # @@ -69,7 +70,9 @@ class Resource(object): self.admin_methods = [] self.log_map = {} self.log_args = { - 'ident': self.ident.encode('utf-8') if self.ident else '', + 'ident': self.ident.encode('utf-8') + if isinstance(self.ident, unicode) + else self.ident if self.ident else '', } def _redirect(self, action_result, code=303): -- 2.5.0

4 3

[RFC] VEPA Network UI and multiple selection
by Samuel Henrique De Oliveira Guimaraes 21 Mar '16

21 Mar '16

Hi team, I was looking into Bootstrap-Select that is already implemented on Wok and it does allows multiple option selection, which means we can use it to select more than one interface when creating a new Vepa network. But instead of looking like a default multi-select input, it keeps the default style and puts the selected items separated by comma: https://silviomoreto.github.io/bootstrap-select/examples/#multiple-select-b… In the other hand, this is what a multiple select looks like with native HTML controls: http://www.w3schools.com/tags/tryit.asp?filename=tryhtml_select_multiple IMO the default html control looks better and it delivers a better user experience than a dropdown. We can "destroy" Bootstrap-Select style and widget if a VEPA network type is selected and customize the default html control with CSS3 to look like other form elements. However, what if there are hundreds of interfaces? In this case I propose we keep Bootstrap-Select with a search/filter on top, which would work like this jsfiddle: https://jsfiddle.net/4yvgnx9e/ Any thoughts or comments? Regards, Samuel

2 1

Gap between the scss/css file's
by Chandra Shekhar Reddy Potula 21 Mar '16

21 Mar '16

Hi All, I see a gap between scss and css always in the upstream code. For sure during the development we always re-compile and get the latest style sheets. My suggestion would be whenever there is a change in the scss file's request to upstream, will it be good idea to ask developer who want scss changes to upstream push corresponding css file changes too ie re-compiled code ? This will some times eliminate the gap between the css and scss files, which I suppose mostly taken care by the maintainer at present ? Just a thought, let me know your views on this !!! Thanks and Regards Chandra

4 3