diff options
| author | Todd Willey <todd@ansolabs.com> | 2011-05-26 11:30:07 -0400 |
|---|---|---|
| committer | Todd Willey <todd@ansolabs.com> | 2011-05-26 11:30:07 -0400 |
| commit | eb90339f094da74eafcc6555907e247e25a8902b (patch) | |
| tree | e5ac40e24ead1273497bca3dba6006f51f5a8d73 /nova/virt | |
| parent | 2278f2886d369af285f914a7b5883d3a7a5727cc (diff) | |
| parent | db18a792414240cbdb1221d0e79e8a63313f103e (diff) | |
Merge trunk.
Diffstat (limited to 'nova/virt')
| -rw-r--r-- | nova/virt/connection.py | 2 | ||||
| -rw-r--r-- | nova/virt/disk.py | 50 | ||||
| -rw-r--r-- | nova/virt/fake.py | 5 | ||||
| -rw-r--r-- | nova/virt/hyperv.py | 21 | ||||
| -rw-r--r-- | nova/virt/images.py | 69 | ||||
| -rw-r--r-- | nova/virt/libvirt/__init__.py | 0 | ||||
| -rw-r--r-- | nova/virt/libvirt/connection.py (renamed from nova/virt/libvirt_conn.py) | 1227 | ||||
| -rw-r--r-- | nova/virt/libvirt/firewall.py | 798 | ||||
| -rw-r--r-- | nova/virt/libvirt/netutils.py | 97 | ||||
| -rw-r--r-- | nova/virt/vmwareapi/fake.py | 9 | ||||
| -rw-r--r-- | nova/virt/vmwareapi/vim.py | 1 | ||||
| -rw-r--r-- | nova/virt/vmwareapi/vmops.py | 43 | ||||
| -rw-r--r-- | nova/virt/vmwareapi_conn.py | 4 | ||||
| -rw-r--r-- | nova/virt/xenapi/fake.py | 2 | ||||
| -rw-r--r-- | nova/virt/xenapi/vm_utils.py | 53 | ||||
| -rw-r--r-- | nova/virt/xenapi/vmops.py | 123 | ||||
| -rw-r--r-- | nova/virt/xenapi/volume_utils.py | 43 | ||||
| -rw-r--r-- | nova/virt/xenapi/volumeops.py | 6 | ||||
| -rw-r--r-- | nova/virt/xenapi_conn.py | 181 |
19 files changed, 1495 insertions, 1239 deletions
diff --git a/nova/virt/connection.py b/nova/virt/connection.py index 99a8849f1..aeec17c98 100644 --- a/nova/virt/connection.py +++ b/nova/virt/connection.py @@ -27,9 +27,9 @@ from nova import utils from nova.virt import driver from nova.virt import fake from nova.virt import hyperv -from nova.virt import libvirt_conn from nova.virt import vmwareapi_conn from nova.virt import xenapi_conn +from nova.virt.libvirt import connection as libvirt_conn LOG = logging.getLogger("nova.virt.connection") diff --git a/nova/virt/disk.py b/nova/virt/disk.py index ddea1a1f7..f8aea1f34 100644 --- a/nova/virt/disk.py +++ b/nova/virt/disk.py @@ -81,34 +81,36 @@ def inject_data(image, key=None, net=None, partition=None, nbd=False): else: mapped_device = device - # We can only loopback mount raw images. If the device isn't there, - # it's normally because it's a .vmdk or a .vdi etc - if not os.path.exists(mapped_device): - raise exception.Error('Mapped device was not found (we can' - ' only inject raw disk images): %s' % - mapped_device) - - # Configure ext2fs so that it doesn't auto-check every N boots - out, err = utils.execute('sudo', 'tune2fs', - '-c', 0, '-i', 0, mapped_device) - - tmpdir = tempfile.mkdtemp() try: - # mount loopback to dir - out, err = utils.execute( - 'sudo', 'mount', mapped_device, tmpdir) - if err: - raise exception.Error(_('Failed to mount filesystem: %s') - % err) - + # We can only loopback mount raw images. If the device isn't there, + # it's normally because it's a .vmdk or a .vdi etc + if not os.path.exists(mapped_device): + raise exception.Error('Mapped device was not found (we can' + ' only inject raw disk images): %s' % + mapped_device) + + # Configure ext2fs so that it doesn't auto-check every N boots + out, err = utils.execute('sudo', 'tune2fs', + '-c', 0, '-i', 0, mapped_device) + + tmpdir = tempfile.mkdtemp() try: - inject_data_into_fs(tmpdir, key, net, utils.execute) + # mount loopback to dir + out, err = utils.execute( + 'sudo', 'mount', mapped_device, tmpdir) + if err: + raise exception.Error(_('Failed to mount filesystem: %s') + % err) + + try: + inject_data_into_fs(tmpdir, key, net, utils.execute) + finally: + # unmount device + utils.execute('sudo', 'umount', mapped_device) finally: - # unmount device - utils.execute('sudo', 'umount', mapped_device) + # remove temporary directory + utils.execute('rmdir', tmpdir) finally: - # remove temporary directory - utils.execute('rmdir', tmpdir) if not partition is None: # remove partitions utils.execute('sudo', 'kpartx', '-d', device) diff --git a/nova/virt/fake.py b/nova/virt/fake.py index 3f030833e..970434c6c 100644 --- a/nova/virt/fake.py +++ b/nova/virt/fake.py @@ -288,8 +288,7 @@ class FakeConnection(driver.ComputeDriver): knowledge of the instance """ if instance_name not in self.instances: - raise exception.NotFound(_("Instance %s Not Found") - % instance_name) + raise exception.InstanceNotFound(instance_id=instance_name) i = self.instances[instance_name] return {'state': i.state, 'max_mem': 0, @@ -368,7 +367,7 @@ class FakeConnection(driver.ComputeDriver): return [0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L] def get_console_output(self, instance): - return 'FAKE CONSOLE OUTPUT' + return 'FAKE CONSOLE\xffOUTPUT' def get_ajax_console(self, instance): return {'token': 'FAKETOKEN', diff --git a/nova/virt/hyperv.py b/nova/virt/hyperv.py index 13f403a66..1142e97a4 100644 --- a/nova/virt/hyperv.py +++ b/nova/virt/hyperv.py @@ -143,8 +143,7 @@ class HyperVConnection(driver.ComputeDriver): """ Create a new VM and start it.""" vm = self._lookup(instance.name) if vm is not None: - raise exception.Duplicate(_('Attempt to create duplicate vm %s') % - instance.name) + raise exception.InstanceExists(name=instance.name) user = manager.AuthManager().get_user(instance['user_id']) project = manager.AuthManager().get_project(instance['project_id']) @@ -368,7 +367,7 @@ class HyperVConnection(driver.ComputeDriver): """Reboot the specified instance.""" vm = self._lookup(instance.name) if vm is None: - raise exception.NotFound('instance not present %s' % instance.name) + raise exception.InstanceNotFound(instance_id=instance.id) self._set_vm_state(instance.name, 'Reboot') def destroy(self, instance): @@ -412,7 +411,7 @@ class HyperVConnection(driver.ComputeDriver): """Get information about the VM""" vm = self._lookup(instance_id) if vm is None: - raise exception.NotFound('instance not present %s' % instance_id) + raise exception.InstanceNotFound(instance_id=instance_id) vm = self._conn.Msvm_ComputerSystem(ElementName=instance_id)[0] vs_man_svc = self._conn.Msvm_VirtualSystemManagementService()[0] vmsettings = vm.associators( @@ -474,14 +473,12 @@ class HyperVConnection(driver.ComputeDriver): def attach_volume(self, instance_name, device_path, mountpoint): vm = self._lookup(instance_name) if vm is None: - raise exception.NotFound('Cannot attach volume to missing %s vm' - % instance_name) + raise exception.InstanceNotFound(instance_id=instance_name) def detach_volume(self, instance_name, mountpoint): vm = self._lookup(instance_name) if vm is None: - raise exception.NotFound('Cannot detach volume from missing %s ' - % instance_name) + raise exception.InstanceNotFound(instance_id=instance_name) def poll_rescued_instances(self, timeout): pass @@ -489,3 +486,11 @@ class HyperVConnection(driver.ComputeDriver): def update_available_resource(self, ctxt, host): """This method is supported only by libvirt.""" return + + def update_host_status(self): + """See xenapi_conn.py implementation.""" + pass + + def get_host_stats(self, refresh=False): + """See xenapi_conn.py implementation.""" + pass diff --git a/nova/virt/images.py b/nova/virt/images.py index 2e3f2ee4d..02c898fda 100644 --- a/nova/virt/images.py +++ b/nova/virt/images.py @@ -21,19 +21,10 @@ Handling of VM disk images. """ -import os.path -import shutil -import sys -import time -import urllib2 -import urlparse - from nova import context from nova import flags from nova import log as logging from nova import utils -from nova.auth import manager -from nova.auth import signer FLAGS = flags.FLAGS @@ -52,66 +43,6 @@ def fetch(image_id, path, _user, _project): return metadata -# NOTE(vish): The methods below should be unnecessary, but I'm leaving -# them in case the glance client does not work on windows. -def _fetch_image_no_curl(url, path, headers): - request = urllib2.Request(url) - for (k, v) in headers.iteritems(): - request.add_header(k, v) - - def urlretrieve(urlfile, fpath): - chunk = 1 * 1024 * 1024 - f = open(fpath, "wb") - while 1: - data = urlfile.read(chunk) - if not data: - break - f.write(data) - - urlopened = urllib2.urlopen(request) - urlretrieve(urlopened, path) - LOG.debug(_("Finished retreving %(url)s -- placed in %(path)s") % locals()) - - -def _fetch_s3_image(image, path, user, project): - url = image_url(image) - - # This should probably move somewhere else, like e.g. a download_as - # method on User objects and at the same time get rewritten to use - # a web client. - headers = {} - headers['Date'] = time.strftime("%a, %d %b %Y %H:%M:%S GMT", time.gmtime()) - - (_, _, url_path, _, _, _) = urlparse.urlparse(url) - access = manager.AuthManager().get_access_key(user, project) - signature = signer.Signer(user.secret.encode()).s3_authorization(headers, - 'GET', - url_path) - headers['Authorization'] = 'AWS %s:%s' % (access, signature) - - if sys.platform.startswith('win'): - return _fetch_image_no_curl(url, path, headers) - else: - cmd = ['/usr/bin/curl', '--fail', '--silent', url] - for (k, v) in headers.iteritems(): - cmd += ['-H', '\'%s: %s\'' % (k, v)] - - cmd += ['-o', path] - return utils.execute(*cmd) - - -def _fetch_local_image(image, path, user, project): - source = _image_path(os.path.join(image, 'image')) - if sys.platform.startswith('win'): - return shutil.copy(source, path) - else: - return utils.execute('cp', source, path) - - -def _image_path(path): - return os.path.join(FLAGS.images_path, path) - - # TODO(vish): xenapi should use the glance client code directly instead # of retrieving the image using this method. def image_url(image): diff --git a/nova/virt/libvirt/__init__.py b/nova/virt/libvirt/__init__.py new file mode 100644 index 000000000..e69de29bb --- /dev/null +++ b/nova/virt/libvirt/__init__.py diff --git a/nova/virt/libvirt_conn.py b/nova/virt/libvirt/connection.py index ae9e84120..dc2dd1219 100644 --- a/nova/virt/libvirt_conn.py +++ b/nova/virt/libvirt/connection.py @@ -58,7 +58,6 @@ from nova import db from nova import exception from nova import flags from nova import log as logging -#from nova import test from nova import utils from nova import vnc from nova.auth import manager @@ -67,20 +66,23 @@ from nova.compute import power_state from nova.virt import disk from nova.virt import driver from nova.virt import images +from nova.virt.libvirt import netutils + libvirt = None libxml2 = None Template = None + LOG = logging.getLogger('nova.virt.libvirt_conn') + FLAGS = flags.FLAGS flags.DECLARE('live_migration_retry_count', 'nova.compute.manager') # TODO(vish): These flags should probably go into a shared location flags.DEFINE_string('rescue_image_id', 'ami-rescue', 'Rescue ami image') flags.DEFINE_string('rescue_kernel_id', 'aki-rescue', 'Rescue aki image') flags.DEFINE_string('rescue_ramdisk_id', 'ari-rescue', 'Rescue ari image') - flags.DEFINE_string('libvirt_xml_template', utils.abspath('virt/libvirt.xml.template'), 'Libvirt XML Template') @@ -102,7 +104,7 @@ flags.DEFINE_string('ajaxterm_portrange', '10000-12000', 'Range of ports that ajaxterm should randomly try to bind') flags.DEFINE_string('firewall_driver', - 'nova.virt.libvirt_conn.IptablesFirewallDriver', + 'nova.virt.libvirt.firewall.IptablesFirewallDriver', 'Firewall driver (defaults to iptables)') flags.DEFINE_string('cpuinfo_xml_template', utils.abspath('virt/cpuinfo.xml.template'), @@ -117,6 +119,8 @@ flags.DEFINE_integer('live_migration_bandwidth', 0, 'Define live migration behavior') flags.DEFINE_string('qemu_img', 'qemu-img', 'binary to use for qemu-img commands') +flags.DEFINE_bool('start_guests_on_host_boot', False, + 'Whether to restart guests when the host reboots') def get_connection(read_only): @@ -142,66 +146,6 @@ def _late_load_cheetah(): Template = t.Template -def _get_net_and_mask(cidr): - net = IPy.IP(cidr) - return str(net.net()), str(net.netmask()) - - -def _get_net_and_prefixlen(cidr): - net = IPy.IP(cidr) - return str(net.net()), str(net.prefixlen()) - - -def _get_ip_version(cidr): - net = IPy.IP(cidr) - return int(net.version()) - - -def _get_network_info(instance): - # TODO(adiantum) If we will keep this function - # we should cache network_info - admin_context = context.get_admin_context() - - ip_addresses = db.fixed_ip_get_all_by_instance(admin_context, - instance['id']) - - networks = db.network_get_all_by_instance(admin_context, - instance['id']) - network_info = [] - - def ip_dict(ip): - return { - "ip": ip.address, - "netmask": network["netmask"], - "enabled": "1"} - - def ip6_dict(ip6): - prefix = ip6.network.cidr_v6 - mac = instance.mac_address - return { - "ip": utils.to_global_ipv6(prefix, mac), - "netmask": ip6.network.netmask_v6, - "gateway": ip6.network.gateway_v6, - "enabled": "1"} - - for network in networks: - network_ips = [ip for ip in ip_addresses - if ip.network_id == network.id] - - mapping = { - 'label': network['label'], - 'gateway': network['gateway'], - 'mac': instance.mac_address, - 'dns': [network['dns']], - 'ips': [ip_dict(ip) for ip in network_ips]} - - if FLAGS.use_ipv6: - mapping['ip6s'] = [ip6_dict(ip) for ip in network_ips] - - network_info.append((network, mapping)) - return network_info - - class LibvirtConnection(driver.ComputeDriver): def __init__(self, read_only): @@ -209,7 +153,6 @@ class LibvirtConnection(driver.ComputeDriver): self.libvirt_uri = self.get_uri() self.libvirt_xml = open(FLAGS.libvirt_xml_template).read() - self.interfaces_xml = open(FLAGS.injected_network_template).read() self.cpuinfo_xml = open(FLAGS.cpuinfo_xml_template).read() self._wrapped_conn = None self.read_only = read_only @@ -231,12 +174,8 @@ class LibvirtConnection(driver.ComputeDriver): {'name': instance['name'], 'state': state}) db.instance_set_state(ctxt, instance['id'], state) - if state == power_state.SHUTOFF: - # TODO(soren): This is what the compute manager does when you - # terminate # an instance. At some point I figure we'll have a - # "terminated" state and some sort of cleanup job that runs - # occasionally, cleaning them out. - db.instance_destroy(ctxt, instance['id']) + # NOTE(justinsb): We no longer delete SHUTOFF instances, + # the user may want to power them back on if state != power_state.RUNNING: continue @@ -311,29 +250,62 @@ class LibvirtConnection(driver.ComputeDriver): return infos def destroy(self, instance, cleanup=True): + instance_name = instance['name'] + try: - virt_dom = self._conn.lookupByName(instance['name']) - virt_dom.destroy() - except Exception as _err: - pass - # If the instance is already terminated, we're still happy - - # We'll save this for when we do shutdown, - # instead of destroy - but destroy returns immediately - timer = utils.LoopingCall(f=None) + virt_dom = self._lookup_by_name(instance_name) + except exception.NotFound: + virt_dom = None - while True: + # If the instance is already terminated, we're still happy + # Otherwise, destroy it + if virt_dom is not None: try: - state = self.get_info(instance['name'])['state'] - db.instance_set_state(context.get_admin_context(), - instance['id'], state) - if state == power_state.SHUTDOWN: - break - except Exception: - db.instance_set_state(context.get_admin_context(), - instance['id'], - power_state.SHUTDOWN) - break + virt_dom.destroy() + except libvirt.libvirtError as e: + is_okay = False + errcode = e.get_error_code() + if errcode == libvirt.VIR_ERR_OPERATION_INVALID: + # If the instance if already shut off, we get this: + # Code=55 Error=Requested operation is not valid: + # domain is not running + (state, _max_mem, _mem, _cpus, _t) = virt_dom.info() + if state == power_state.SHUTOFF: + is_okay = True + + if not is_okay: + LOG.warning(_("Error from libvirt during destroy of " + "%(instance_name)s. Code=%(errcode)s " + "Error=%(e)s") % + locals()) + raise + + try: + # NOTE(justinsb): We remove the domain definition. We probably + # would do better to keep it if cleanup=False (e.g. volumes?) + # (e.g. #2 - not losing machines on failure) + virt_dom.undefine() + except libvirt.libvirtError as e: + errcode = e.get_error_code() + LOG.warning(_("Error from libvirt during undefine of " + "%(instance_name)s. Code=%(errcode)s " + "Error=%(e)s") % + locals()) + raise + + def _wait_for_destroy(): + """Called at an interval until the VM is gone.""" + instance_name = instance['name'] + + try: + state = self.get_info(instance_name)['state'] + except exception.NotFound: + msg = _("Instance %s destroyed successfully.") % instance_name + LOG.info(msg) + raise utils.LoopingCallDone + + timer = utils.LoopingCall(_wait_for_destroy) + timer.start(interval=0.5, now=True) self.firewall_driver.unfilter_instance(instance) @@ -354,7 +326,7 @@ class LibvirtConnection(driver.ComputeDriver): @exception.wrap_exception def attach_volume(self, instance_name, device_path, mountpoint): - virt_dom = self._conn.lookupByName(instance_name) + virt_dom = self._lookup_by_name(instance_name) mount_device = mountpoint.rpartition("/")[2] if device_path.startswith('/dev/'): xml = """<disk type='block'> @@ -372,7 +344,7 @@ class LibvirtConnection(driver.ComputeDriver): name, mount_device) else: - raise exception.Invalid(_("Invalid device path %s") % device_path) + raise exception.InvalidDevicePath(path=device_path) virt_dom.attachDevice(xml) @@ -391,18 +363,18 @@ class LibvirtConnection(driver.ComputeDriver): if child.prop('dev') == device: return str(node) finally: - if ctx != None: + if ctx is not None: ctx.xpathFreeContext() - if doc != None: + if doc is not None: doc.freeDoc() @exception.wrap_exception def detach_volume(self, instance_name, mountpoint): - virt_dom = self._conn.lookupByName(instance_name) + virt_dom = self._lookup_by_name(instance_name) mount_device = mountpoint.rpartition("/")[2] xml = self._get_disk_xml(virt_dom.XMLDesc(0), mount_device) if not xml: - raise exception.NotFound(_("No disk at %s") % mount_device) + raise exception.DiskNotFound(location=mount_device) virt_dom.detachDevice(xml) @exception.wrap_exception @@ -415,7 +387,7 @@ class LibvirtConnection(driver.ComputeDriver): """ image_service = utils.import_object(FLAGS.image_service) - virt_dom = self._conn.lookupByName(instance['name']) + virt_dom = self._lookup_by_name(instance['name']) elevated = context.get_admin_context() base = image_service.show(elevated, instance['image_id']) @@ -423,9 +395,8 @@ class LibvirtConnection(driver.ComputeDriver): metadata = {'disk_format': base['disk_format'], 'container_format': base['container_format'], 'is_public': False, + 'name': '%s.%s' % (base['name'], image_id), 'properties': {'architecture': base['architecture'], - 'type': base['type'], - 'name': '%s.%s' % (base['name'], image_id), 'kernel_id': instance['kernel_id'], 'image_location': 'snapshot', 'image_state': 'available', @@ -452,12 +423,17 @@ class LibvirtConnection(driver.ComputeDriver): # Export the snapshot to a raw image temp_dir = tempfile.mkdtemp() out_path = os.path.join(temp_dir, snapshot_name) - qemu_img_cmd = '%s convert -f qcow2 -O raw -s %s %s %s' % ( - FLAGS.qemu_img, - snapshot_name, - disk_path, - out_path) - utils.execute(qemu_img_cmd) + qemu_img_cmd = (FLAGS.qemu_img, + 'convert', + '-f', + 'qcow2', + '-O', + 'raw', + '-s', + snapshot_name, + disk_path, + out_path) + utils.execute(*qemu_img_cmd) # Upload that image to the image service with open(out_path) as image_file: @@ -471,31 +447,43 @@ class LibvirtConnection(driver.ComputeDriver): @exception.wrap_exception def reboot(self, instance): + """Reboot a virtual machine, given an instance reference. + + This method actually destroys and re-creates the domain to ensure the + reboot happens, as the guest OS cannot ignore this action. + + """ + virt_dom = self._conn.lookupByName(instance['name']) + # NOTE(itoumsn): Use XML delived from the running instance + # instead of using to_xml(instance). This is almost the ultimate + # stupid workaround. + xml = virt_dom.XMLDesc(0) + # NOTE(itoumsn): self.shutdown() and wait instead of self.destroy() is + # better because we cannot ensure flushing dirty buffers + # in the guest OS. But, in case of KVM, shutdown() does not work... self.destroy(instance, False) - xml = self.to_xml(instance) self.firewall_driver.setup_basic_filtering(instance) self.firewall_driver.prepare_instance_filter(instance) - self._conn.createXML(xml, 0) + self._create_new_domain(xml) self.firewall_driver.apply_instance_filter(instance) - timer = utils.LoopingCall(f=None) - def _wait_for_reboot(): + """Called at an interval until the VM is running again.""" + instance_name = instance['name'] + try: - state = self.get_info(instance['name'])['state'] - db.instance_set_state(context.get_admin_context(), - instance['id'], state) - if state == power_state.RUNNING: - LOG.debug(_('instance %s: rebooted'), instance['name']) - timer.stop() - except Exception, exn: - LOG.exception(_('_wait_for_reboot failed: %s'), exn) - db.instance_set_state(context.get_admin_context(), - instance['id'], - power_state.SHUTDOWN) - timer.stop() + state = self.get_info(instance_name)['state'] + except exception.NotFound: + msg = _("During reboot, %s disappeared.") % instance_name + LOG.error(msg) + raise utils.LoopingCallDone + + if state == power_state.RUNNING: + msg = _("Instance %s rebooted successfully.") % instance_name + LOG.info(msg) + raise utils.LoopingCallDone - timer.f = _wait_for_reboot + timer = utils.LoopingCall(_wait_for_reboot) return timer.start(interval=0.5, now=True) @exception.wrap_exception @@ -515,7 +503,15 @@ class LibvirtConnection(driver.ComputeDriver): raise exception.ApiError("resume not supported for libvirt") @exception.wrap_exception - def rescue(self, instance, callback=None): + def rescue(self, instance): + """Loads a VM using rescue images. + + A rescue is normally performed when something goes wrong with the + primary images and data needs to be corrected/recovered. Rescuing + should not edit or over-ride the original image, only allow for + data recovery. + + """ self.destroy(instance, False) xml = self.to_xml(instance, rescue=True) @@ -523,31 +519,35 @@ class LibvirtConnection(driver.ComputeDriver): 'kernel_id': FLAGS.rescue_kernel_id, 'ramdisk_id': FLAGS.rescue_ramdisk_id} self._create_image(instance, xml, '.rescue', rescue_images) - self._conn.createXML(xml, 0) - - timer = utils.LoopingCall(f=None) + self._create_new_domain(xml) def _wait_for_rescue(): + """Called at an interval until the VM is running again.""" + instance_name = instance['name'] + try: - state = self.get_info(instance['name'])['state'] - db.instance_set_state(None, instance['id'], state) - if state == power_state.RUNNING: - LOG.debug(_('instance %s: rescued'), instance['name']) - timer.stop() - except Exception, exn: - LOG.exception(_('_wait_for_rescue failed: %s'), exn) - db.instance_set_state(None, - instance['id'], - power_state.SHUTDOWN) - timer.stop() + state = self.get_info(instance_name)['state'] + except exception.NotFound: + msg = _("During reboot, %s disappeared.") % instance_name + LOG.error(msg) + raise utils.LoopingCallDone - timer.f = _wait_for_rescue + if state == power_state.RUNNING: + msg = _("Instance %s rescued successfully.") % instance_name + LOG.info(msg) + raise utils.LoopingCallDone + + timer = utils.LoopingCall(_wait_for_rescue) return timer.start(interval=0.5, now=True) @exception.wrap_exception - def unrescue(self, instance, callback=None): - # NOTE(vish): Because reboot destroys and recreates an instance using - # the normal xml file, we can just call reboot here + def unrescue(self, instance): + """Reboot the VM which is being rescued back into primary images. + + Because reboot destroys and re-creates instances, unresue should + simply call reboot. + + """ self.reboot(instance) @exception.wrap_exception @@ -558,37 +558,36 @@ class LibvirtConnection(driver.ComputeDriver): # for xenapi(tr3buchet) @exception.wrap_exception def spawn(self, instance, network_info=None): - xml = self.to_xml(instance, network_info) - db.instance_set_state(context.get_admin_context(), - instance['id'], - power_state.NOSTATE, - 'launching') + xml = self.to_xml(instance, False, network_info) self.firewall_driver.setup_basic_filtering(instance, network_info) self.firewall_driver.prepare_instance_filter(instance, network_info) - self._create_image(instance, xml, network_info) - self._conn.createXML(xml, 0) + self._create_image(instance, xml, network_info=network_info) + domain = self._create_new_domain(xml) LOG.debug(_("instance %s: is running"), instance['name']) self.firewall_driver.apply_instance_filter(instance) - timer = utils.LoopingCall(f=None) + if FLAGS.start_guests_on_host_boot: + LOG.debug(_("instance %s: setting autostart ON") % + instance['name']) + domain.setAutostart(1) def _wait_for_boot(): + """Called at an interval until the VM is running.""" + instance_name = instance['name'] + try: - state = self.get_info(instance['name'])['state'] - db.instance_set_state(context.get_admin_context(), - instance['id'], state) - if state == power_state.RUNNING: - LOG.debug(_('instance %s: booted'), instance['name']) - timer.stop() - except: - LOG.exception(_('instance %s: failed to boot'), - instance['name']) - db.instance_set_state(context.get_admin_context(), - instance['id'], - power_state.SHUTDOWN) - timer.stop() + state = self.get_info(instance_name)['state'] + except exception.NotFound: + msg = _("During reboot, %s disappeared.") % instance_name + LOG.error(msg) + raise utils.LoopingCallDone + + if state == power_state.RUNNING: + msg = _("Instance %s spawned successfully.") % instance_name + LOG.info(msg) + raise utils.LoopingCallDone - timer.f = _wait_for_boot + timer = utils.LoopingCall(_wait_for_boot) return timer.start(interval=0.5, now=True) def _flush_xen_console(self, virsh_output): @@ -654,7 +653,7 @@ class LibvirtConnection(driver.ComputeDriver): raise Exception(_('Unable to find an open port')) def get_pty_for_instance(instance_name): - virt_dom = self._conn.lookupByName(instance_name) + virt_dom = self._lookup_by_name(instance_name) xml = virt_dom.XMLDesc(0) dom = minidom.parseString(xml) @@ -676,10 +675,13 @@ class LibvirtConnection(driver.ComputeDriver): subprocess.Popen(cmd, shell=True) return {'token': token, 'host': host, 'port': port} + def get_host_ip_addr(self): + return FLAGS.my_ip + @exception.wrap_exception def get_vnc_console(self, instance): def get_vnc_port_for_instance(instance_name): - virt_dom = self._conn.lookupByName(instance_name) + virt_dom = self._lookup_by_name(instance_name) xml = virt_dom.XMLDesc(0) # TODO: use etree instead of minidom dom = minidom.parseString(xml) @@ -743,7 +745,7 @@ class LibvirtConnection(driver.ComputeDriver): def _create_image(self, inst, libvirt_xml, suffix='', disk_images=None, network_info=None): if not network_info: - network_info = _get_network_info(inst) + network_info = netutils.get_network_info(inst) if not suffix: suffix = '' @@ -797,7 +799,10 @@ class LibvirtConnection(driver.ComputeDriver): root_fname = '%08x' % int(disk_images['image_id']) size = FLAGS.minimum_root_size - if inst['instance_type'] == 'm1.tiny' or suffix == '.rescue': + + inst_type_id = inst['instance_type_id'] + inst_type = instance_types.get_instance_type(inst_type_id) + if inst_type['name'] == 'm1.tiny' or suffix == '.rescue': size = None root_fname += "_sm" @@ -809,14 +814,13 @@ class LibvirtConnection(driver.ComputeDriver): user=user, project=project, size=size) - type_data = instance_types.get_instance_type(inst['instance_type']) - if type_data['local_gb']: + if inst_type['local_gb']: self._cache_image(fn=self._create_local, target=basepath('disk.local'), - fname="local_%s" % type_data['local_gb'], + fname="local_%s" % inst_type['local_gb'], cow=FLAGS.use_cow_images, - local_gb=type_data['local_gb']) + local_gb=inst_type['local_gb']) # For now, we assume that if we're not using a kernel, we're using a # partitioned disk image where the target partition is the first @@ -899,26 +903,16 @@ class LibvirtConnection(driver.ComputeDriver): mac_id = mapping['mac'].replace(':', '') if FLAGS.allow_project_net_traffic: + template = "<parameter name=\"%s\"value=\"%s\" />\n" + net, mask = netutils.get_net_and_mask(network['cidr']) + values = [("PROJNET", net), ("PROJMASK", mask)] if FLAGS.use_ipv6: - net, mask = _get_net_and_mask(network['cidr']) - net_v6, prefixlen_v6 = _get_net_and_prefixlen( + net_v6, prefixlen_v6 = netutils.get_net_and_prefixlen( network['cidr_v6']) - extra_params = ("<parameter name=\"PROJNET\" " - "value=\"%s\" />\n" - "<parameter name=\"PROJMASK\" " - "value=\"%s\" />\n" - "<parameter name=\"PROJNETV6\" " - "value=\"%s\" />\n" - "<parameter name=\"PROJMASKV6\" " - "value=\"%s\" />\n") % \ - (net, mask, net_v6, prefixlen_v6) - else: - net, mask = _get_net_and_mask(network['cidr']) - extra_params = ("<parameter name=\"PROJNET\" " - "value=\"%s\" />\n" - "<parameter name=\"PROJMASK\" " - "value=\"%s\" />\n") % \ - (net, mask) + values.extend([("PROJNETV6", net_v6), + ("PROJMASKV6", prefixlen_v6)]) + + extra_params = "".join([template % value for value in values]) else: extra_params = "\n" @@ -936,22 +930,18 @@ class LibvirtConnection(driver.ComputeDriver): return result - def to_xml(self, instance, rescue=False, network_info=None): - # TODO(termie): cache? - LOG.debug(_('instance %s: starting toXML method'), instance['name']) - + def _prepare_xml_info(self, instance, rescue=False, network_info=None): # TODO(adiantum) remove network_info creation code # when multinics will be completed if not network_info: - network_info = _get_network_info(instance) + network_info = netutils.get_network_info(instance) nics = [] for (network, mapping) in network_info: - nics.append(self._get_nic_for_xml(network, - mapping)) + nics.append(self._get_nic_for_xml(network, mapping)) # FIXME(vish): stick this in db - instance_type_name = instance['instance_type'] - instance_type = instance_types.get_instance_type(instance_type_name) + inst_type_id = instance['instance_type_id'] + inst_type = instance_types.get_instance_type(inst_type_id) if FLAGS.use_cow_images: driver_type = 'qcow2' @@ -962,10 +952,10 @@ class LibvirtConnection(driver.ComputeDriver): 'name': instance['name'], 'basepath': os.path.join(FLAGS.instances_path, instance['name']), - 'memory_kb': instance_type['memory_mb'] * 1024, - 'vcpus': instance_type['vcpus'], + 'memory_kb': inst_type['memory_mb'] * 1024, + 'vcpus': inst_type['vcpus'], 'rescue': rescue, - 'local': instance_type['local_gb'], + 'local': inst_type['local_gb'], 'driver_type': driver_type, 'nics': nics} @@ -980,18 +970,43 @@ class LibvirtConnection(driver.ComputeDriver): xml_info['ramdisk'] = xml_info['basepath'] + "/ramdisk" xml_info['disk'] = xml_info['basepath'] + "/disk" + return xml_info + def to_xml(self, instance, rescue=False, network_info=None): + # TODO(termie): cache? + LOG.debug(_('instance %s: starting toXML method'), instance['name']) + xml_info = self._prepare_xml_info(instance, rescue, network_info) xml = str(Template(self.libvirt_xml, searchList=[xml_info])) - LOG.debug(_('instance %s: finished toXML method'), - instance['name']) + LOG.debug(_('instance %s: finished toXML method'), instance['name']) return xml - def get_info(self, instance_name): + def _lookup_by_name(self, instance_name): + """Retrieve libvirt domain object given an instance name. + + All libvirt error handling should be handled in this method and + relevant nova exceptions should be raised in response. + + """ try: - virt_dom = self._conn.lookupByName(instance_name) - except: - raise exception.NotFound(_("Instance %s not found") - % instance_name) + return self._conn.lookupByName(instance_name) + except libvirt.libvirtError as ex: + error_code = ex.get_error_code() + if error_code == libvirt.VIR_ERR_NO_DOMAIN: + raise exception.InstanceNotFound(instance_id=instance_name) + + msg = _("Error from libvirt while looking up %(instance_name)s: " + "[Error Code %(error_code)s] %(ex)s") % locals() + raise exception.Error(msg) + + def get_info(self, instance_name): + """Retrieve information from libvirt for a specific instance name. + + If a libvirt error is encountered during lookup, we might raise a + NotFound exception or Error exception depending on how severe the + libvirt error is. + + """ + virt_dom = self._lookup_by_name(instance_name) (state, max_mem, mem, num_cpu, cpu_time) = virt_dom.info() return {'state': state, 'max_mem': max_mem, @@ -999,6 +1014,24 @@ class LibvirtConnection(driver.ComputeDriver): 'num_cpu': num_cpu, 'cpu_time': cpu_time} + def _create_new_domain(self, xml, persistent=True, launch_flags=0): + # NOTE(justinsb): libvirt has two types of domain: + # * a transient domain disappears when the guest is shutdown + # or the host is rebooted. + # * a permanent domain is not automatically deleted + # NOTE(justinsb): Even for ephemeral instances, transient seems risky + + if persistent: + # To create a persistent domain, first define it, then launch it. + domain = self._conn.defineXML(xml) + + domain.createWithFlags(launch_flags) + else: + # createXML call creates a transient domain + domain = self._conn.createXML(xml, launch_flags) + + return domain + def get_diagnostics(self, instance_name): raise exception.ApiError(_("diagnostics are not supported " "for libvirt")) @@ -1010,7 +1043,7 @@ class LibvirtConnection(driver.ComputeDriver): Returns a list of all block devices for this domain. """ - domain = self._conn.lookupByName(instance_name) + domain = self._lookup_by_name(instance_name) # TODO(devcamcar): Replace libxml2 with etree. xml = domain.XMLDesc(0) doc = None @@ -1033,14 +1066,14 @@ class LibvirtConnection(driver.ComputeDriver): if child.name == 'target': devdst = child.prop('dev') - if devdst == None: + if devdst is None: continue disks.append(devdst) finally: - if ctx != None: + if ctx is not None: ctx.xpathFreeContext() - if doc != None: + if doc is not None: doc.freeDoc() return disks @@ -1052,7 +1085,7 @@ class LibvirtConnection(driver.ComputeDriver): Returns a list of all network interfaces for this instance. """ - domain = self._conn.lookupByName(instance_name) + domain = self._lookup_by_name(instance_name) # TODO(devcamcar): Replace libxml2 with etree. xml = domain.XMLDesc(0) doc = None @@ -1075,14 +1108,14 @@ class LibvirtConnection(driver.ComputeDriver): if child.name == 'target': devdst = child.prop('dev') - if devdst == None: + if devdst is None: continue interfaces.append(devdst) finally: - if ctx != None: + if ctx is not None: ctx.xpathFreeContext() - if doc != None: + if doc is not None: doc.freeDoc() return interfaces @@ -1218,9 +1251,9 @@ class LibvirtConnection(driver.ComputeDriver): xml = libxml2.parseDoc(xml) nodes = xml.xpathEval('//host/cpu') if len(nodes) != 1: - raise exception.Invalid(_("Invalid xml. '<cpu>' must be 1," - "but %d\n") % len(nodes) - + xml.serialize()) + reason = _("'<cpu>' must be 1, but %d\n") % len(nodes) + reason += xml.serialize() + raise exception.InvalidCPUInfo(reason=reason) cpu_info = dict() @@ -1249,9 +1282,8 @@ class LibvirtConnection(driver.ComputeDriver): tkeys = topology.keys() if set(tkeys) != set(keys): ks = ', '.join(keys) - raise exception.Invalid(_("Invalid xml: topology" - "(%(topology)s) must have " - "%(ks)s") % locals()) + reason = _("topology (%(topology)s) must have %(ks)s") + raise exception.InvalidCPUInfo(reason=reason % locals()) feature_nodes = xml.xpathEval('//host/cpu/feature') features = list() @@ -1267,7 +1299,7 @@ class LibvirtConnection(driver.ComputeDriver): Note that this function takes an instance name, not an Instance, so that it can be called by monitor. """ - domain = self._conn.lookupByName(instance_name) + domain = self._lookup_by_name(instance_name) return domain.blockStats(disk) def interface_stats(self, instance_name, interface): @@ -1275,7 +1307,7 @@ class LibvirtConnection(driver.ComputeDriver): Note that this function takes an instance name, not an Instance, so that it can be called by monitor. """ - domain = self._conn.lookupByName(instance_name) + domain = self._lookup_by_name(instance_name) return domain.interfaceStats(interface) def get_console_pool_info(self, console_type): @@ -1309,9 +1341,7 @@ class LibvirtConnection(driver.ComputeDriver): try: service_ref = db.service_get_all_compute_by_host(ctxt, host)[0] except exception.NotFound: - raise exception.Invalid(_("Cannot update compute manager " - "specific info, because no service " - "record was found.")) + raise exception.ComputeServiceUnavailable(host=host) # Updating host information dic = {'vcpus': self.get_vcpu_total(), @@ -1364,7 +1394,7 @@ class LibvirtConnection(driver.ComputeDriver): raise if ret <= 0: - raise exception.Invalid(m % locals()) + raise exception.InvalidCPUInfo(reason=m % locals()) return @@ -1405,18 +1435,13 @@ class LibvirtConnection(driver.ComputeDriver): # wait for completion timeout_count = range(FLAGS.live_migration_retry_count) while timeout_count: - try: - filter_name = 'nova-instance-%s' % instance_ref.name - self._conn.nwfilterLookupByName(filter_name) + if self.firewall_driver.instance_filter_exists(instance_ref): break - except libvirt.libvirtError: - timeout_count.pop() - if len(timeout_count) == 0: - ec2_id = instance_ref['hostname'] - iname = instance_ref.name - msg = _('Timeout migrating for %(ec2_id)s(%(iname)s)') - raise exception.Error(msg % locals()) - time.sleep(1) + timeout_count.pop() + if len(timeout_count) == 0: + msg = _('Timeout migrating for %s. nwfilter not found.') + raise exception.Error(msg % instance_ref.name) + time.sleep(1) def live_migration(self, ctxt, instance_ref, dest, post_method, recover_method): @@ -1479,7 +1504,7 @@ class LibvirtConnection(driver.ComputeDriver): FLAGS.live_migration_bandwidth) except Exception: - recover_method(ctxt, instance_ref) + recover_method(ctxt, instance_ref, dest=dest) raise # Waiting for completion of live_migration. @@ -1500,725 +1525,11 @@ class LibvirtConnection(driver.ComputeDriver): """See comments of same method in firewall_driver.""" self.firewall_driver.unfilter_instance(instance_ref) - -class FirewallDriver(object): - def prepare_instance_filter(self, instance, network_info=None): - """Prepare filters for the instance. - - At this point, the instance isn't running yet.""" - raise NotImplementedError() - - def unfilter_instance(self, instance): - """Stop filtering instance""" - raise NotImplementedError() - - def apply_instance_filter(self, instance): - """Apply instance filter. - - Once this method returns, the instance should be firewalled - appropriately. This method should as far as possible be a - no-op. It's vastly preferred to get everything set up in - prepare_instance_filter. - """ - raise NotImplementedError() - - def refresh_security_group_rules(self, security_group_id): - """Refresh security group rules from data store - - Gets called when a rule has been added to or removed from - the security group.""" - raise NotImplementedError() - - def refresh_security_group_members(self, security_group_id): - """Refresh security group members from data store - - Gets called when an instance gets added to or removed from - the security group.""" - raise NotImplementedError() - - def refresh_provider_fw_rules(self): - """Refresh common rules for all hosts/instances from data store. - - Gets called when a rule has been added to or removed from - the list of rules (via admin api). - - """ - raise NotImplementedError() - - def setup_basic_filtering(self, instance, network_info=None): - """Create rules to block spoofing and allow dhcp. - - This gets called when spawning an instance, before - :method:`prepare_instance_filter`. - - """ - raise NotImplementedError() - - -class NWFilterFirewall(FirewallDriver): - """ - This class implements a network filtering mechanism versatile - enough for EC2 style Security Group filtering by leveraging - libvirt's nwfilter. - - First, all instances get a filter ("nova-base-filter") applied. - This filter provides some basic security such as protection against - MAC spoofing, IP spoofing, and ARP spoofing. - - This filter drops all incoming ipv4 and ipv6 connections. - Outgoing connections are never blocked. - - Second, every security group maps to a nwfilter filter(*). - NWFilters can be updated at runtime and changes are applied - immediately, so changes to security groups can be applied at - runtime (as mandated by the spec). - - Security group rules are named "nova-secgroup-<id>" where <id> - is the internal id of the security group. They're applied only on - hosts that have instances in the security group in question. - - Updates to security groups are done by updating the data model - (in response to API calls) followed by a request sent to all - the nodes with instances in the security group to refresh the - security group. - - Each instance has its own NWFilter, which references the above - mentioned security group NWFilters. This was done because - interfaces can only reference one filter while filters can - reference multiple other filters. This has the added benefit of - actually being able to add and remove security groups from an - instance at run time. This functionality is not exposed anywhere, - though. - - Outstanding questions: - - The name is unique, so would there be any good reason to sync - the uuid across the nodes (by assigning it from the datamodel)? - - - (*) This sentence brought to you by the redundancy department of - redundancy. - - """ - - def __init__(self, get_connection, **kwargs): - self._libvirt_get_connection = get_connection - self.static_filters_configured = False - self.handle_security_groups = False - - def apply_instance_filter(self, instance): - """No-op. Everything is done in prepare_instance_filter""" + def update_host_status(self): + """See xenapi_conn.py implementation.""" pass - def _get_connection(self): - return self._libvirt_get_connection() - _conn = property(_get_connection) - - def nova_dhcp_filter(self): - """The standard allow-dhcp-server filter is an <ip> one, so it uses - ebtables to allow traffic through. Without a corresponding rule in - iptables, it'll get blocked anyway.""" - - return '''<filter name='nova-allow-dhcp-server' chain='ipv4'> - <uuid>891e4787-e5c0-d59b-cbd6-41bc3c6b36fc</uuid> - <rule action='accept' direction='out' - priority='100'> - <udp srcipaddr='0.0.0.0' - dstipaddr='255.255.255.255' - srcportstart='68' - dstportstart='67'/> - </rule> - <rule action='accept' direction='in' - priority='100'> - <udp srcipaddr='$DHCPSERVER' - srcportstart='67' - dstportstart='68'/> - </rule> - </filter>''' - - def nova_ra_filter(self): - return '''<filter name='nova-allow-ra-server' chain='root'> - <uuid>d707fa71-4fb5-4b27-9ab7-ba5ca19c8804</uuid> - <rule action='accept' direction='inout' - priority='100'> - <icmpv6 srcipaddr='$RASERVER'/> - </rule> - </filter>''' - - def setup_basic_filtering(self, instance, network_info=None): - """Set up basic filtering (MAC, IP, and ARP spoofing protection)""" - logging.info('called setup_basic_filtering in nwfilter') - - if not network_info: - network_info = _get_network_info(instance) - - if self.handle_security_groups: - # No point in setting up a filter set that we'll be overriding - # anyway. - return - - logging.info('ensuring static filters') - self._ensure_static_filters() - - for (network, mapping) in network_info: - nic_id = mapping['mac'].replace(':', '') - instance_filter_name = self._instance_filter_name(instance, nic_id) - self._define_filter(self._filter_container(instance_filter_name, - ['nova-base'])) - - def _ensure_static_filters(self): - """Static filters are filters that have no need to be IP aware. - - There is no configuration or tuneability of these filters, so they - can be set up once and forgotten about. - - """ - - if self.static_filters_configured: - return - - self._define_filter(self._filter_container('nova-base', - ['no-mac-spoofing', - 'no-ip-spoofing', - 'no-arp-spoofing', - 'allow-dhcp-server'])) - self._define_filter(self.nova_base_ipv4_filter) - self._define_filter(self.nova_base_ipv6_filter) - self._define_filter(self.nova_dhcp_filter) - self._define_filter(self.nova_ra_filter) - self._define_filter(self.nova_vpn_filter) - if FLAGS.allow_project_net_traffic: - self._define_filter(self.nova_project_filter) - if FLAGS.use_ipv6: - self._define_filter(self.nova_project_filter_v6) - - self.static_filters_configured = True - - def _filter_container(self, name, filters): - xml = '''<filter name='%s' chain='root'>%s</filter>''' % ( - name, - ''.join(["<filterref filter='%s'/>" % (f,) for f in filters])) - return xml - - nova_vpn_filter = '''<filter name='nova-vpn' chain='root'> - <uuid>2086015e-cf03-11df-8c5d-080027c27973</uuid> - <filterref filter='allow-dhcp-server'/> - <filterref filter='nova-allow-dhcp-server'/> - <filterref filter='nova-base-ipv4'/> - <filterref filter='nova-base-ipv6'/> - </filter>''' - - def nova_base_ipv4_filter(self): - retval = "<filter name='nova-base-ipv4' chain='ipv4'>" - for protocol in ['tcp', 'udp', 'icmp']: - for direction, action, priority in [('out', 'accept', 399), - ('in', 'drop', 400)]: - retval += """<rule action='%s' direction='%s' priority='%d'> - <%s /> - </rule>""" % (action, direction, - priority, protocol) - retval += '</filter>' - return retval - - def nova_base_ipv6_filter(self): - retval = "<filter name='nova-base-ipv6' chain='ipv6'>" - for protocol in ['tcp-ipv6', 'udp-ipv6', 'icmpv6']: - for direction, action, priority in [('out', 'accept', 399), - ('in', 'drop', 400)]: - retval += """<rule action='%s' direction='%s' priority='%d'> - <%s /> - </rule>""" % (action, direction, - priority, protocol) - retval += '</filter>' - return retval - - def nova_project_filter(self): - retval = "<filter name='nova-project' chain='ipv4'>" - for protocol in ['tcp', 'udp', 'icmp']: - retval += """<rule action='accept' direction='in' priority='200'> - <%s srcipaddr='$PROJNET' srcipmask='$PROJMASK' /> - </rule>""" % protocol - retval += '</filter>' - return retval - - def nova_project_filter_v6(self): - retval = "<filter name='nova-project-v6' chain='ipv6'>" - for protocol in ['tcp-ipv6', 'udp-ipv6', 'icmpv6']: - retval += """<rule action='accept' direction='inout' - priority='200'> - <%s srcipaddr='$PROJNETV6' - srcipmask='$PROJMASKV6' /> - </rule>""" % (protocol) - retval += '</filter>' - return retval - - def _define_filter(self, xml): - if callable(xml): - xml = xml() - # execute in a native thread and block current greenthread until done - tpool.execute(self._conn.nwfilterDefineXML, xml) - - def unfilter_instance(self, instance): - # Nothing to do + def get_host_stats(self, refresh=False): + """See xenapi_conn.py implementation.""" pass - def prepare_instance_filter(self, instance, network_info=None): - """Creates an NWFilter for the given instance. - - In the process, it makes sure the filters for the provider blocks, - security groups, and base filter are all in place. - - """ - if not network_info: - network_info = _get_network_info(instance) - if instance['image_id'] == FLAGS.vpn_image_id: - base_filter = 'nova-vpn' - else: - base_filter = 'nova-base' - - self.refresh_provider_fw_rules() - - ctxt = context.get_admin_context() - - instance_secgroup_filter_name = \ - '%s-secgroup' % (self._instance_filter_name(instance)) - - instance_secgroup_filter_children = ['nova-base-ipv4', - 'nova-base-ipv6', - 'nova-allow-dhcp-server'] - - for security_group in \ - db.security_group_get_by_instance(ctxt, instance['id']): - - self.refresh_security_group_rules(security_group['id']) - - instance_secgroup_filter_children += [('nova-secgroup-%s' % - security_group['id'])] - - self._define_filter( - self._filter_container(instance_secgroup_filter_name, - instance_secgroup_filter_children)) - - for (network, mapping) in network_info: - nic_id = mapping['mac'].replace(':', '') - instance_filter_name = self._instance_filter_name(instance, nic_id) - instance_filter_children = [base_filter, 'nova-provider-rules', - instance_secgroup_filter_name] - - if FLAGS.use_ipv6: - gateway_v6 = network['gateway_v6'] - - if gateway_v6: - instance_secgroup_filter_children += \ - ['nova-allow-ra-server'] - - if FLAGS.allow_project_net_traffic: - instance_filter_children += ['nova-project'] - if FLAGS.use_ipv6: - instance_filter_children += ['nova-project-v6'] - - self._define_filter( - self._filter_container(instance_filter_name, - instance_filter_children)) - - return - - def refresh_security_group_rules(self, security_group_id): - return self._define_filter( - self.security_group_to_nwfilter_xml(security_group_id)) - - def refresh_provider_fw_rules(self): - """Update rules for all instances. - - This is part of the FirewallDriver API and is called when the - provider firewall rules change in the database. In the - `prepare_instance_filter` we add a reference to the - 'nova-provider-rules' filter for each instance's firewall, and - by changing that filter we update them all. - - """ - xml = self.provider_fw_to_nwfilter_xml() - return self._define_filter(xml) - - def security_group_to_nwfilter_xml(self, security_group_id): - security_group = db.security_group_get(context.get_admin_context(), - security_group_id) - rule_xml = "" - v6protocol = {'tcp': 'tcp-ipv6', 'udp': 'udp-ipv6', 'icmp': 'icmpv6'} - for rule in security_group.rules: - rule_xml += "<rule action='accept' direction='in' priority='300'>" - if rule.cidr: - version = _get_ip_version(rule.cidr) - if(FLAGS.use_ipv6 and version == 6): - net, prefixlen = _get_net_and_prefixlen(rule.cidr) - rule_xml += "<%s srcipaddr='%s' srcipmask='%s' " % \ - (v6protocol[rule.protocol], net, prefixlen) - else: - net, mask = _get_net_and_mask(rule.cidr) - rule_xml += "<%s srcipaddr='%s' srcipmask='%s' " % \ - (rule.protocol, net, mask) - if rule.protocol in ['tcp', 'udp']: - rule_xml += "dstportstart='%s' dstportend='%s' " % \ - (rule.from_port, rule.to_port) - elif rule.protocol == 'icmp': - LOG.info('rule.protocol: %r, rule.from_port: %r, ' - 'rule.to_port: %r', rule.protocol, - rule.from_port, rule.to_port) - if rule.from_port != -1: - rule_xml += "type='%s' " % rule.from_port - if rule.to_port != -1: - rule_xml += "code='%s' " % rule.to_port - - rule_xml += '/>\n' - rule_xml += "</rule>\n" - xml = "<filter name='nova-secgroup-%s' " % security_group_id - if(FLAGS.use_ipv6): - xml += "chain='root'>%s</filter>" % rule_xml - else: - xml += "chain='ipv4'>%s</filter>" % rule_xml - return xml - - def provider_fw_to_nwfilter_xml(self): - """Compose a filter of drop rules from specified cidrs.""" - rule_xml = "" - v6protocol = {'tcp': 'tcp-ipv6', 'udp': 'udp-ipv6', 'icmp': 'icmpv6'} - rules = db.provider_fw_rule_get_all(context.get_admin_context()) - for rule in rules: - rule_xml += "<rule action='block' direction='in' priority='150'>" - version = _get_ip_version(rule.cidr) - if(FLAGS.use_ipv6 and version == 6): - net, prefixlen = _get_net_and_prefixlen(rule.cidr) - rule_xml += "<%s srcipaddr='%s' srcipmask='%s' " % \ - (v6protocol[rule.protocol], net, prefixlen) - else: - net, mask = _get_net_and_mask(rule.cidr) - rule_xml += "<%s srcipaddr='%s' srcipmask='%s' " % \ - (rule.protocol, net, mask) - if rule.protocol in ['tcp', 'udp']: - rule_xml += "dstportstart='%s' dstportend='%s' " % \ - (rule.from_port, rule.to_port) - elif rule.protocol == 'icmp': - LOG.info('rule.protocol: %r, rule.from_port: %r, ' - 'rule.to_port: %r', rule.protocol, - rule.from_port, rule.to_port) - if rule.from_port != -1: - rule_xml += "type='%s' " % rule.from_port - if rule.to_port != -1: - rule_xml += "code='%s' " % rule.to_port - - rule_xml += '/>\n' - rule_xml += "</rule>\n" - xml = "<filter name='nova-provider-rules' " - if(FLAGS.use_ipv6): - xml += "chain='root'>%s</filter>" % rule_xml - else: - xml += "chain='ipv4'>%s</filter>" % rule_xml - return xml - - def _instance_filter_name(self, instance, nic_id=None): - if not nic_id: - return 'nova-instance-%s' % (instance['name']) - return 'nova-instance-%s-%s' % (instance['name'], nic_id) - - -class IptablesFirewallDriver(FirewallDriver): - def __init__(self, execute=None, **kwargs): - from nova.network import linux_net - self.iptables = linux_net.iptables_manager - self.instances = {} - self.nwfilter = NWFilterFirewall(kwargs['get_connection']) - self.basicly_filtered = False - - self.iptables.ipv4['filter'].add_chain('sg-fallback') - self.iptables.ipv4['filter'].add_rule('sg-fallback', '-j DROP') - self.iptables.ipv6['filter'].add_chain('sg-fallback') - self.iptables.ipv6['filter'].add_rule('sg-fallback', '-j DROP') - - def setup_basic_filtering(self, instance, network_info=None): - """Set up provider rules and basic NWFilter.""" - if not network_info: - network_info = _get_network_info(instance) - self.nwfilter.setup_basic_filtering(instance, network_info) - if not self.basicly_filtered: - LOG.debug("Setup Basic Filtering") - self.refresh_provider_fw_rules() - self.basicly_filtered = True - - def apply_instance_filter(self, instance): - """No-op. Everything is done in prepare_instance_filter""" - pass - - def unfilter_instance(self, instance): - if self.instances.pop(instance['id'], None): - self.remove_filters_for_instance(instance) - self.iptables.apply() - else: - LOG.info(_('Attempted to unfilter instance %s which is not ' - 'filtered'), instance['id']) - - def prepare_instance_filter(self, instance, network_info=None): - if not network_info: - network_info = _get_network_info(instance) - self.instances[instance['id']] = instance - self.add_filters_for_instance(instance, network_info) - self.iptables.apply() - - def add_filters_for_instance(self, instance, network_info=None): - if not network_info: - network_info = _get_network_info(instance) - chain_name = self._instance_chain_name(instance) - - self.iptables.ipv4['filter'].add_chain(chain_name) - self.iptables.ipv4['filter'].empty_chain(chain_name) - - ips_v4 = [ip['ip'] for (_, mapping) in network_info - for ip in mapping['ips']] - - for ipv4_address in ips_v4: - self.iptables.ipv4['filter'].add_rule('local', - '-d %s -j $%s' % - (ipv4_address, chain_name)) - - if FLAGS.use_ipv6: - self.iptables.ipv6['filter'].add_chain(chain_name) - self.iptables.ipv6['filter'].empty_chain(chain_name) - ips_v6 = [ip['ip'] for (_, mapping) in network_info - for ip in mapping['ip6s']] - - for ipv6_address in ips_v6: - self.iptables.ipv6['filter'].add_rule('local', - '-d %s -j $%s' % - (ipv6_address, - chain_name)) - - ipv4_rules, ipv6_rules = self.instance_rules(instance, network_info) - for rule in ipv4_rules: - self.iptables.ipv4['filter'].add_rule(chain_name, rule) - - if FLAGS.use_ipv6: - for rule in ipv6_rules: - self.iptables.ipv6['filter'].add_rule(chain_name, rule) - - def remove_filters_for_instance(self, instance): - chain_name = self._instance_chain_name(instance) - - self.iptables.ipv4['filter'].remove_chain(chain_name) - if FLAGS.use_ipv6: - self.iptables.ipv6['filter'].remove_chain(chain_name) - - def instance_rules(self, instance, network_info=None): - if not network_info: - network_info = _get_network_info(instance) - ctxt = context.get_admin_context() - - ipv4_rules = [] - ipv6_rules = [] - - # Always drop invalid packets - ipv4_rules += ['-m state --state ' 'INVALID -j DROP'] - ipv6_rules += ['-m state --state ' 'INVALID -j DROP'] - - # Pass through provider-wide drops - ipv4_rules += ['-j $provider'] - ipv6_rules += ['-j $provider'] - - # Allow established connections - ipv4_rules += ['-m state --state ESTABLISHED,RELATED -j ACCEPT'] - ipv6_rules += ['-m state --state ESTABLISHED,RELATED -j ACCEPT'] - - dhcp_servers = [network['gateway'] for (network, _m) in network_info] - - for dhcp_server in dhcp_servers: - ipv4_rules.append('-s %s -p udp --sport 67 --dport 68 ' - '-j ACCEPT' % (dhcp_server,)) - - #Allow project network traffic - if FLAGS.allow_project_net_traffic: - cidrs = [network['cidr'] for (network, _m) in network_info] - for cidr in cidrs: - ipv4_rules.append('-s %s -j ACCEPT' % (cidr,)) - - # We wrap these in FLAGS.use_ipv6 because they might cause - # a DB lookup. The other ones are just list operations, so - # they're not worth the clutter. - if FLAGS.use_ipv6: - # Allow RA responses - gateways_v6 = [network['gateway_v6'] for (network, _m) in - network_info] - for gateway_v6 in gateways_v6: - ipv6_rules.append( - '-s %s/128 -p icmpv6 -j ACCEPT' % (gateway_v6,)) - - #Allow project network traffic - if FLAGS.allow_project_net_traffic: - cidrv6s = [network['cidr_v6'] for (network, _m) - in network_info] - - for cidrv6 in cidrv6s: - ipv6_rules.append('-s %s -j ACCEPT' % (cidrv6,)) - - security_groups = db.security_group_get_by_instance(ctxt, - instance['id']) - - # then, security group chains and rules - for security_group in security_groups: - rules = db.security_group_rule_get_by_security_group(ctxt, - security_group['id']) - - for rule in rules: - LOG.debug(_("Adding security group rule: %r"), rule) - - if not rule.cidr: - # Eventually, a mechanism to grant access for security - # groups will turn up here. It'll use ipsets. - continue - - version = _get_ip_version(rule.cidr) - if version == 4: - fw_rules = ipv4_rules - else: - fw_rules = ipv6_rules - - protocol = rule.protocol - if version == 6 and rule.protocol == 'icmp': - protocol = 'icmpv6' - - args = ['-p', protocol, '-s', rule.cidr] - - if rule.protocol in ['udp', 'tcp']: - if rule.from_port == rule.to_port: - args += ['--dport', '%s' % (rule.from_port,)] - else: - args += ['-m', 'multiport', - '--dports', '%s:%s' % (rule.from_port, - rule.to_port)] - elif rule.protocol == 'icmp': - icmp_type = rule.from_port - icmp_code = rule.to_port - - if icmp_type == -1: - icmp_type_arg = None - else: - icmp_type_arg = '%s' % icmp_type - if not icmp_code == -1: - icmp_type_arg += '/%s' % icmp_code - - if icmp_type_arg: - if version == 4: - args += ['-m', 'icmp', '--icmp-type', - icmp_type_arg] - elif version == 6: - args += ['-m', 'icmp6', '--icmpv6-type', - icmp_type_arg] - - args += ['-j ACCEPT'] - fw_rules += [' '.join(args)] - - ipv4_rules += ['-j $sg-fallback'] - ipv6_rules += ['-j $sg-fallback'] - - return ipv4_rules, ipv6_rules - - def refresh_security_group_members(self, security_group): - pass - - def refresh_security_group_rules(self, security_group): - self.do_refresh_security_group_rules(security_group) - self.iptables.apply() - - @utils.synchronized('iptables', external=True) - def do_refresh_security_group_rules(self, security_group): - for instance in self.instances.values(): - self.remove_filters_for_instance(instance) - self.add_filters_for_instance(instance) - - def refresh_provider_fw_rules(self): - """See class:FirewallDriver: docs.""" - self._do_refresh_provider_fw_rules() - self.iptables.apply() - - @utils.synchronized('iptables', external=True) - def _do_refresh_provider_fw_rules(self): - """Internal, synchronized version of refresh_provider_fw_rules.""" - self._purge_provider_fw_rules() - self._build_provider_fw_rules() - - def _purge_provider_fw_rules(self): - """Remove all rules from the provider chains.""" - self.iptables.ipv4['filter'].empty_chain('provider') - if FLAGS.use_ipv6: - self.iptables.ipv6['filter'].empty_chain('provider') - - def _build_provider_fw_rules(self): - """Create all rules for the provider IP DROPs.""" - self.iptables.ipv4['filter'].add_chain('provider') - if FLAGS.use_ipv6: - self.iptables.ipv6['filter'].add_chain('provider') - ipv4_rules, ipv6_rules = self._provider_rules() - for rule in ipv4_rules: - self.iptables.ipv4['filter'].add_rule('provider', rule) - - if FLAGS.use_ipv6: - for rule in ipv6_rules: - self.iptables.ipv6['filter'].add_rule('provider', rule) - - def _provider_rules(self): - """Generate a list of rules from provider for IP4 & IP6.""" - ctxt = context.get_admin_context() - ipv4_rules = [] - ipv6_rules = [] - rules = db.provider_fw_rule_get_all(ctxt) - for rule in rules: - LOG.debug(_('Adding prvider rule: %r'), rule) - - if not rule.cidr: - # Eventually, a mechanism to grant access for security - # groups will turn up here. It'll use ipsets. - continue - - version = _get_ip_version(rule.cidr) - if version == 4: - fw_rules = ipv4_rules - else: - fw_rules = ipv6_rules - - protocol = rule.protocol - if version == 6 and protocol == 'icmp': - protocol = 'icmpv6' - - args = ['-p', protocol, '-s', rule.cidr] - - if protocol in ['udp', 'tcp']: - if rule.from_port == rule.to_port: - args += ['--dport', '%s' % (rule.from_port,)] - else: - args += ['-m', 'multiport', - '--dports', '%s:%s' % (rule.from_port, - rule.to_port)] - elif protocol == 'icmp': - icmp_type = rule.from_port - icmp_code = rule.to_port - - if icmp_type == -1: - icmp_type_arg = None - else: - icmp_type_arg = '%s' % icmp_type - if not icmp_code == -1: - icmp_type_arg += '/%s' % icmp_code - - if icmp_type_arg: - if version == 4: - args += ['-m', 'icmp', '--icmp-type', - icmp_type_arg] - elif version == 6: - args += ['-m', 'icmp6', '--icmpv6-type', - icmp_type_arg] - args += ['-j DROP'] - fw_rules += [' '.join(args)] - return ipv4_rules, ipv6_rules - - def _security_group_chain_name(self, security_group_id): - return 'nova-sg-%s' % (security_group_id,) - - def _instance_chain_name(self, instance): - return 'inst-%s' % (instance['id'],) diff --git a/nova/virt/libvirt/firewall.py b/nova/virt/libvirt/firewall.py new file mode 100644 index 000000000..fd4c120a6 --- /dev/null +++ b/nova/virt/libvirt/firewall.py @@ -0,0 +1,798 @@ +# vim: tabstop=4 shiftwidth=4 softtabstop=4 + +# Copyright 2010 United States Government as represented by the +# Administrator of the National Aeronautics and Space Administration. +# All Rights Reserved. +# Copyright (c) 2010 Citrix Systems, Inc. +# +# Licensed under the Apache License, Version 2.0 (the "License"); you may +# not use this file except in compliance with the License. You may obtain +# a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT +# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the +# License for the specific language governing permissions and limitations +# under the License. + + +from eventlet import tpool + +from nova import context +from nova import db +from nova import flags +from nova import log as logging +from nova import utils +from nova.virt.libvirt import netutils + + +LOG = logging.getLogger("nova.virt.libvirt.firewall") +FLAGS = flags.FLAGS + + +try: + import libvirt +except ImportError: + LOG.warn(_("Libvirt module could not be loaded. NWFilterFirewall will " + "not work correctly.")) + + +class FirewallDriver(object): + def prepare_instance_filter(self, instance, network_info=None): + """Prepare filters for the instance. + + At this point, the instance isn't running yet.""" + raise NotImplementedError() + + def unfilter_instance(self, instance): + """Stop filtering instance""" + raise NotImplementedError() + + def apply_instance_filter(self, instance): + """Apply instance filter. + + Once this method returns, the instance should be firewalled + appropriately. This method should as far as possible be a + no-op. It's vastly preferred to get everything set up in + prepare_instance_filter. + """ + raise NotImplementedError() + + def refresh_security_group_rules(self, + security_group_id, + network_info=None): + """Refresh security group rules from data store + + Gets called when a rule has been added to or removed from + the security group.""" + raise NotImplementedError() + + def refresh_security_group_members(self, security_group_id): + """Refresh security group members from data store + + Gets called when an instance gets added to or removed from + the security group.""" + raise NotImplementedError() + + def refresh_provider_fw_rules(self): + """Refresh common rules for all hosts/instances from data store. + + Gets called when a rule has been added to or removed from + the list of rules (via admin api). + + """ + raise NotImplementedError() + + def setup_basic_filtering(self, instance, network_info=None): + """Create rules to block spoofing and allow dhcp. + + This gets called when spawning an instance, before + :method:`prepare_instance_filter`. + + """ + raise NotImplementedError() + + def instance_filter_exists(self, instance): + """Check nova-instance-instance-xxx exists""" + raise NotImplementedError() + + +class NWFilterFirewall(FirewallDriver): + """ + This class implements a network filtering mechanism versatile + enough for EC2 style Security Group filtering by leveraging + libvirt's nwfilter. + + First, all instances get a filter ("nova-base-filter") applied. + This filter provides some basic security such as protection against + MAC spoofing, IP spoofing, and ARP spoofing. + + This filter drops all incoming ipv4 and ipv6 connections. + Outgoing connections are never blocked. + + Second, every security group maps to a nwfilter filter(*). + NWFilters can be updated at runtime and changes are applied + immediately, so changes to security groups can be applied at + runtime (as mandated by the spec). + + Security group rules are named "nova-secgroup-<id>" where <id> + is the internal id of the security group. They're applied only on + hosts that have instances in the security group in question. + + Updates to security groups are done by updating the data model + (in response to API calls) followed by a request sent to all + the nodes with instances in the security group to refresh the + security group. + + Each instance has its own NWFilter, which references the above + mentioned security group NWFilters. This was done because + interfaces can only reference one filter while filters can + reference multiple other filters. This has the added benefit of + actually being able to add and remove security groups from an + instance at run time. This functionality is not exposed anywhere, + though. + + Outstanding questions: + + The name is unique, so would there be any good reason to sync + the uuid across the nodes (by assigning it from the datamodel)? + + + (*) This sentence brought to you by the redundancy department of + redundancy. + + """ + + def __init__(self, get_connection, **kwargs): + self._libvirt_get_connection = get_connection + self.static_filters_configured = False + self.handle_security_groups = False + + def apply_instance_filter(self, instance): + """No-op. Everything is done in prepare_instance_filter""" + pass + + def _get_connection(self): + return self._libvirt_get_connection() + _conn = property(_get_connection) + + def nova_dhcp_filter(self): + """The standard allow-dhcp-server filter is an <ip> one, so it uses + ebtables to allow traffic through. Without a corresponding rule in + iptables, it'll get blocked anyway.""" + + return '''<filter name='nova-allow-dhcp-server' chain='ipv4'> + <uuid>891e4787-e5c0-d59b-cbd6-41bc3c6b36fc</uuid> + <rule action='accept' direction='out' + priority='100'> + <udp srcipaddr='0.0.0.0' + dstipaddr='255.255.255.255' + srcportstart='68' + dstportstart='67'/> + </rule> + <rule action='accept' direction='in' + priority='100'> + <udp srcipaddr='$DHCPSERVER' + srcportstart='67' + dstportstart='68'/> + </rule> + </filter>''' + + def nova_ra_filter(self): + return '''<filter name='nova-allow-ra-server' chain='root'> + <uuid>d707fa71-4fb5-4b27-9ab7-ba5ca19c8804</uuid> + <rule action='accept' direction='inout' + priority='100'> + <icmpv6 srcipaddr='$RASERVER'/> + </rule> + </filter>''' + + def setup_basic_filtering(self, instance, network_info=None): + """Set up basic filtering (MAC, IP, and ARP spoofing protection)""" + logging.info('called setup_basic_filtering in nwfilter') + + if not network_info: + network_info = netutils.get_network_info(instance) + + if self.handle_security_groups: + # No point in setting up a filter set that we'll be overriding + # anyway. + return + + logging.info('ensuring static filters') + self._ensure_static_filters() + + if instance['image_id'] == str(FLAGS.vpn_image_id): + base_filter = 'nova-vpn' + else: + base_filter = 'nova-base' + + for (network, mapping) in network_info: + nic_id = mapping['mac'].replace(':', '') + instance_filter_name = self._instance_filter_name(instance, nic_id) + self._define_filter(self._filter_container(instance_filter_name, + [base_filter])) + + def _ensure_static_filters(self): + """Static filters are filters that have no need to be IP aware. + + There is no configuration or tuneability of these filters, so they + can be set up once and forgotten about. + + """ + + if self.static_filters_configured: + return + + self._define_filter(self._filter_container('nova-base', + ['no-mac-spoofing', + 'no-ip-spoofing', + 'no-arp-spoofing', + 'allow-dhcp-server'])) + self._define_filter(self._filter_container('nova-vpn', + ['allow-dhcp-server'])) + self._define_filter(self.nova_base_ipv4_filter) + self._define_filter(self.nova_base_ipv6_filter) + self._define_filter(self.nova_dhcp_filter) + self._define_filter(self.nova_ra_filter) + if FLAGS.allow_project_net_traffic: + self._define_filter(self.nova_project_filter) + if FLAGS.use_ipv6: + self._define_filter(self.nova_project_filter_v6) + + self.static_filters_configured = True + + def _filter_container(self, name, filters): + xml = '''<filter name='%s' chain='root'>%s</filter>''' % ( + name, + ''.join(["<filterref filter='%s'/>" % (f,) for f in filters])) + return xml + + def nova_base_ipv4_filter(self): + retval = "<filter name='nova-base-ipv4' chain='ipv4'>" + for protocol in ['tcp', 'udp', 'icmp']: + for direction, action, priority in [('out', 'accept', 399), + ('in', 'drop', 400)]: + retval += """<rule action='%s' direction='%s' priority='%d'> + <%s /> + </rule>""" % (action, direction, + priority, protocol) + retval += '</filter>' + return retval + + def nova_base_ipv6_filter(self): + retval = "<filter name='nova-base-ipv6' chain='ipv6'>" + for protocol in ['tcp-ipv6', 'udp-ipv6', 'icmpv6']: + for direction, action, priority in [('out', 'accept', 399), + ('in', 'drop', 400)]: + retval += """<rule action='%s' direction='%s' priority='%d'> + <%s /> + </rule>""" % (action, direction, + priority, protocol) + retval += '</filter>' + return retval + + def nova_project_filter(self): + retval = "<filter name='nova-project' chain='ipv4'>" + for protocol in ['tcp', 'udp', 'icmp']: + retval += """<rule action='accept' direction='in' priority='200'> + <%s srcipaddr='$PROJNET' srcipmask='$PROJMASK' /> + </rule>""" % protocol + retval += '</filter>' + return retval + + def nova_project_filter_v6(self): + retval = "<filter name='nova-project-v6' chain='ipv6'>" + for protocol in ['tcp-ipv6', 'udp-ipv6', 'icmpv6']: + retval += """<rule action='accept' direction='inout' + priority='200'> + <%s srcipaddr='$PROJNETV6' + srcipmask='$PROJMASKV6' /> + </rule>""" % (protocol) + retval += '</filter>' + return retval + + def _define_filter(self, xml): + if callable(xml): + xml = xml() + # execute in a native thread and block current greenthread until done + tpool.execute(self._conn.nwfilterDefineXML, xml) + + def unfilter_instance(self, instance): + # Nothing to do + pass + + def prepare_instance_filter(self, instance, network_info=None): + """Creates an NWFilter for the given instance. + + In the process, it makes sure the filters for the provider blocks, + security groups, and base filter are all in place. + + """ + if not network_info: + network_info = netutils.get_network_info(instance) + + self.refresh_provider_fw_rules() + + ctxt = context.get_admin_context() + + instance_secgroup_filter_name = \ + '%s-secgroup' % (self._instance_filter_name(instance)) + + instance_secgroup_filter_children = ['nova-base-ipv4', + 'nova-base-ipv6', + 'nova-allow-dhcp-server'] + + if FLAGS.use_ipv6: + networks = [network for (network, _m) in network_info if + network['gateway_v6']] + + if networks: + instance_secgroup_filter_children.\ + append('nova-allow-ra-server') + + for security_group in \ + db.security_group_get_by_instance(ctxt, instance['id']): + + self.refresh_security_group_rules(security_group['id']) + + instance_secgroup_filter_children.append('nova-secgroup-%s' % + security_group['id']) + + self._define_filter( + self._filter_container(instance_secgroup_filter_name, + instance_secgroup_filter_children)) + + network_filters = self.\ + _create_network_filters(instance, network_info, + instance_secgroup_filter_name) + + for (name, children) in network_filters: + self._define_filters(name, children) + + def _create_network_filters(self, instance, network_info, + instance_secgroup_filter_name): + if instance['image_id'] == str(FLAGS.vpn_image_id): + base_filter = 'nova-vpn' + else: + base_filter = 'nova-base' + + result = [] + for (_n, mapping) in network_info: + nic_id = mapping['mac'].replace(':', '') + instance_filter_name = self._instance_filter_name(instance, nic_id) + instance_filter_children = [base_filter, 'nova-provider-rules', + instance_secgroup_filter_name] + + if FLAGS.allow_project_net_traffic: + instance_filter_children.append('nova-project') + if FLAGS.use_ipv6: + instance_filter_children.append('nova-project-v6') + + result.append((instance_filter_name, instance_filter_children)) + + return result + + def _define_filters(self, filter_name, filter_children): + self._define_fitler(self._filter_container(filter_name, + filter_children)) + + def refresh_security_group_rules(self, + security_group_id, + network_info=None): + return self._define_filter( + self.security_group_to_nwfilter_xml(security_group_id)) + + def refresh_provider_fw_rules(self): + """Update rules for all instances. + + This is part of the FirewallDriver API and is called when the + provider firewall rules change in the database. In the + `prepare_instance_filter` we add a reference to the + 'nova-provider-rules' filter for each instance's firewall, and + by changing that filter we update them all. + + """ + xml = self.provider_fw_to_nwfilter_xml() + return self._define_filter(xml) + + def security_group_to_nwfilter_xml(self, security_group_id): + security_group = db.security_group_get(context.get_admin_context(), + security_group_id) + rule_xml = "" + v6protocol = {'tcp': 'tcp-ipv6', 'udp': 'udp-ipv6', 'icmp': 'icmpv6'} + for rule in security_group.rules: + rule_xml += "<rule action='accept' direction='in' priority='300'>" + if rule.cidr: + version = netutils.get_ip_version(rule.cidr) + if(FLAGS.use_ipv6 and version == 6): + net, prefixlen = netutils.get_net_and_prefixlen(rule.cidr) + rule_xml += "<%s srcipaddr='%s' srcipmask='%s' " % \ + (v6protocol[rule.protocol], net, prefixlen) + else: + net, mask = netutils.get_net_and_mask(rule.cidr) + rule_xml += "<%s srcipaddr='%s' srcipmask='%s' " % \ + (rule.protocol, net, mask) + if rule.protocol in ['tcp', 'udp']: + rule_xml += "dstportstart='%s' dstportend='%s' " % \ + (rule.from_port, rule.to_port) + elif rule.protocol == 'icmp': + LOG.info('rule.protocol: %r, rule.from_port: %r, ' + 'rule.to_port: %r', rule.protocol, + rule.from_port, rule.to_port) + if rule.from_port != -1: + rule_xml += "type='%s' " % rule.from_port + if rule.to_port != -1: + rule_xml += "code='%s' " % rule.to_port + + rule_xml += '/>\n' + rule_xml += "</rule>\n" + xml = "<filter name='nova-secgroup-%s' " % security_group_id + if(FLAGS.use_ipv6): + xml += "chain='root'>%s</filter>" % rule_xml + else: + xml += "chain='ipv4'>%s</filter>" % rule_xml + return xml + + def provider_fw_to_nwfilter_xml(self): + """Compose a filter of drop rules from specified cidrs.""" + rule_xml = "" + v6protocol = {'tcp': 'tcp-ipv6', 'udp': 'udp-ipv6', 'icmp': 'icmpv6'} + rules = db.provider_fw_rule_get_all(context.get_admin_context()) + for rule in rules: + rule_xml += "<rule action='block' direction='in' priority='150'>" + version = netutils.get_ip_version(rule.cidr) + if(FLAGS.use_ipv6 and version == 6): + net, prefixlen = netutils.get_net_and_prefixlen(rule.cidr) + rule_xml += "<%s srcipaddr='%s' srcipmask='%s' " % \ + (v6protocol[rule.protocol], net, prefixlen) + else: + net, mask = netutils.get_net_and_mask(rule.cidr) + rule_xml += "<%s srcipaddr='%s' srcipmask='%s' " % \ + (rule.protocol, net, mask) + if rule.protocol in ['tcp', 'udp']: + rule_xml += "dstportstart='%s' dstportend='%s' " % \ + (rule.from_port, rule.to_port) + elif rule.protocol == 'icmp': + LOG.info('rule.protocol: %r, rule.from_port: %r, ' + 'rule.to_port: %r', rule.protocol, + rule.from_port, rule.to_port) + if rule.from_port != -1: + rule_xml += "type='%s' " % rule.from_port + if rule.to_port != -1: + rule_xml += "code='%s' " % rule.to_port + + rule_xml += '/>\n' + rule_xml += "</rule>\n" + xml = "<filter name='nova-provider-rules' " + if(FLAGS.use_ipv6): + xml += "chain='root'>%s</filter>" % rule_xml + else: + xml += "chain='ipv4'>%s</filter>" % rule_xml + return xml + + def _instance_filter_name(self, instance, nic_id=None): + if not nic_id: + return 'nova-instance-%s' % (instance['name']) + return 'nova-instance-%s-%s' % (instance['name'], nic_id) + + def instance_filter_exists(self, instance): + """Check nova-instance-instance-xxx exists""" + network_info = netutils.get_network_info(instance) + for (network, mapping) in network_info: + nic_id = mapping['mac'].replace(':', '') + instance_filter_name = self._instance_filter_name(instance, nic_id) + try: + self._conn.nwfilterLookupByName(instance_filter_name) + except libvirt.libvirtError: + name = instance.name + LOG.debug(_('The nwfilter(%(instance_filter_name)s) for' + '%(name)s is not found.') % locals()) + return False + return True + + +class IptablesFirewallDriver(FirewallDriver): + def __init__(self, execute=None, **kwargs): + from nova.network import linux_net + self.iptables = linux_net.iptables_manager + self.instances = {} + self.nwfilter = NWFilterFirewall(kwargs['get_connection']) + self.basicly_filtered = False + + self.iptables.ipv4['filter'].add_chain('sg-fallback') + self.iptables.ipv4['filter'].add_rule('sg-fallback', '-j DROP') + self.iptables.ipv6['filter'].add_chain('sg-fallback') + self.iptables.ipv6['filter'].add_rule('sg-fallback', '-j DROP') + + def setup_basic_filtering(self, instance, network_info=None): + """Set up provider rules and basic NWFilter.""" + if not network_info: + network_info = netutils.get_network_info(instance) + self.nwfilter.setup_basic_filtering(instance, network_info) + if not self.basicly_filtered: + LOG.debug(_("iptables firewall: Setup Basic Filtering")) + self.refresh_provider_fw_rules() + self.basicly_filtered = True + + def apply_instance_filter(self, instance): + """No-op. Everything is done in prepare_instance_filter""" + pass + + def unfilter_instance(self, instance): + if self.instances.pop(instance['id'], None): + self.remove_filters_for_instance(instance) + self.iptables.apply() + else: + LOG.info(_('Attempted to unfilter instance %s which is not ' + 'filtered'), instance['id']) + + def prepare_instance_filter(self, instance, network_info=None): + if not network_info: + network_info = netutils.get_network_info(instance) + self.instances[instance['id']] = instance + self.add_filters_for_instance(instance, network_info) + self.iptables.apply() + + def _create_filter(self, ips, chain_name): + return ['-d %s -j $%s' % (ip, chain_name) for ip in ips] + + def _filters_for_instance(self, chain_name, network_info): + ips_v4 = [ip['ip'] for (_n, mapping) in network_info + for ip in mapping['ips']] + ipv4_rules = self._create_filter(ips_v4, chain_name) + + ipv6_rules = [] + if FLAGS.use_ipv6: + ips_v6 = [ip['ip'] for (_n, mapping) in network_info + for ip in mapping['ip6s']] + ipv6_rules = self._create_filter(ips_v6, chain_name) + + return ipv4_rules, ipv6_rules + + def _add_filters(self, chain_name, ipv4_rules, ipv6_rules): + for rule in ipv4_rules: + self.iptables.ipv4['filter'].add_rule(chain_name, rule) + + if FLAGS.use_ipv6: + for rule in ipv6_rules: + self.iptables.ipv6['filter'].add_rule(chain_name, rule) + + def add_filters_for_instance(self, instance, network_info=None): + chain_name = self._instance_chain_name(instance) + if FLAGS.use_ipv6: + self.iptables.ipv6['filter'].add_chain(chain_name) + self.iptables.ipv4['filter'].add_chain(chain_name) + ipv4_rules, ipv6_rules = self._filters_for_instance(chain_name, + network_info) + self._add_filters('local', ipv4_rules, ipv6_rules) + ipv4_rules, ipv6_rules = self.instance_rules(instance, network_info) + self._add_filters(chain_name, ipv4_rules, ipv6_rules) + + def remove_filters_for_instance(self, instance): + chain_name = self._instance_chain_name(instance) + + self.iptables.ipv4['filter'].remove_chain(chain_name) + if FLAGS.use_ipv6: + self.iptables.ipv6['filter'].remove_chain(chain_name) + + def instance_rules(self, instance, network_info=None): + if not network_info: + network_info = netutils.get_network_info(instance) + ctxt = context.get_admin_context() + + ipv4_rules = [] + ipv6_rules = [] + + # Always drop invalid packets + ipv4_rules += ['-m state --state ' 'INVALID -j DROP'] + ipv6_rules += ['-m state --state ' 'INVALID -j DROP'] + + # Pass through provider-wide drops + ipv4_rules += ['-j $provider'] + ipv6_rules += ['-j $provider'] + + # Allow established connections + ipv4_rules += ['-m state --state ESTABLISHED,RELATED -j ACCEPT'] + ipv6_rules += ['-m state --state ESTABLISHED,RELATED -j ACCEPT'] + + dhcp_servers = [network['gateway'] for (network, _m) in network_info] + + for dhcp_server in dhcp_servers: + ipv4_rules.append('-s %s -p udp --sport 67 --dport 68 ' + '-j ACCEPT' % (dhcp_server,)) + + #Allow project network traffic + if FLAGS.allow_project_net_traffic: + cidrs = [network['cidr'] for (network, _m) in network_info] + for cidr in cidrs: + ipv4_rules.append('-s %s -j ACCEPT' % (cidr,)) + + # We wrap these in FLAGS.use_ipv6 because they might cause + # a DB lookup. The other ones are just list operations, so + # they're not worth the clutter. + if FLAGS.use_ipv6: + # Allow RA responses + gateways_v6 = [network['gateway_v6'] for (network, _m) in + network_info] + for gateway_v6 in gateways_v6: + ipv6_rules.append( + '-s %s/128 -p icmpv6 -j ACCEPT' % (gateway_v6,)) + + #Allow project network traffic + if FLAGS.allow_project_net_traffic: + cidrv6s = [network['cidr_v6'] for (network, _m) + in network_info] + + for cidrv6 in cidrv6s: + ipv6_rules.append('-s %s -j ACCEPT' % (cidrv6,)) + + security_groups = db.security_group_get_by_instance(ctxt, + instance['id']) + + # then, security group chains and rules + for security_group in security_groups: + rules = db.security_group_rule_get_by_security_group(ctxt, + security_group['id']) + + for rule in rules: + LOG.debug(_("Adding security group rule: %r"), rule) + + if not rule.cidr: + # Eventually, a mechanism to grant access for security + # groups will turn up here. It'll use ipsets. + continue + + version = netutils.get_ip_version(rule.cidr) + if version == 4: + fw_rules = ipv4_rules + else: + fw_rules = ipv6_rules + + protocol = rule.protocol + if version == 6 and rule.protocol == 'icmp': + protocol = 'icmpv6' + + args = ['-p', protocol, '-s', rule.cidr] + + if rule.protocol in ['udp', 'tcp']: + if rule.from_port == rule.to_port: + args += ['--dport', '%s' % (rule.from_port,)] + else: + args += ['-m', 'multiport', + '--dports', '%s:%s' % (rule.from_port, + rule.to_port)] + elif rule.protocol == 'icmp': + icmp_type = rule.from_port + icmp_code = rule.to_port + + if icmp_type == -1: + icmp_type_arg = None + else: + icmp_type_arg = '%s' % icmp_type + if not icmp_code == -1: + icmp_type_arg += '/%s' % icmp_code + + if icmp_type_arg: + if version == 4: + args += ['-m', 'icmp', '--icmp-type', + icmp_type_arg] + elif version == 6: + args += ['-m', 'icmp6', '--icmpv6-type', + icmp_type_arg] + + args += ['-j ACCEPT'] + fw_rules += [' '.join(args)] + + ipv4_rules += ['-j $sg-fallback'] + ipv6_rules += ['-j $sg-fallback'] + + return ipv4_rules, ipv6_rules + + def instance_filter_exists(self, instance): + """Check nova-instance-instance-xxx exists""" + return self.nwfilter.instance_filter_exists(instance) + + def refresh_security_group_members(self, security_group): + pass + + def refresh_security_group_rules(self, security_group, network_info=None): + self.do_refresh_security_group_rules(security_group, network_info) + self.iptables.apply() + + @utils.synchronized('iptables', external=True) + def do_refresh_security_group_rules(self, + security_group, + network_info=None): + for instance in self.instances.values(): + self.remove_filters_for_instance(instance) + if not network_info: + network_info = netutils.get_network_info(instance) + self.add_filters_for_instance(instance, network_info) + + def refresh_provider_fw_rules(self): + """See class:FirewallDriver: docs.""" + self._do_refresh_provider_fw_rules() + self.iptables.apply() + + @utils.synchronized('iptables', external=True) + def _do_refresh_provider_fw_rules(self): + """Internal, synchronized version of refresh_provider_fw_rules.""" + self._purge_provider_fw_rules() + self._build_provider_fw_rules() + + def _purge_provider_fw_rules(self): + """Remove all rules from the provider chains.""" + self.iptables.ipv4['filter'].empty_chain('provider') + if FLAGS.use_ipv6: + self.iptables.ipv6['filter'].empty_chain('provider') + + def _build_provider_fw_rules(self): + """Create all rules for the provider IP DROPs.""" + self.iptables.ipv4['filter'].add_chain('provider') + if FLAGS.use_ipv6: + self.iptables.ipv6['filter'].add_chain('provider') + ipv4_rules, ipv6_rules = self._provider_rules() + for rule in ipv4_rules: + self.iptables.ipv4['filter'].add_rule('provider', rule) + + if FLAGS.use_ipv6: + for rule in ipv6_rules: + self.iptables.ipv6['filter'].add_rule('provider', rule) + + def _provider_rules(self): + """Generate a list of rules from provider for IP4 & IP6.""" + ctxt = context.get_admin_context() + ipv4_rules = [] + ipv6_rules = [] + rules = db.provider_fw_rule_get_all(ctxt) + for rule in rules: + LOG.debug(_('Adding prvider rule: %r'), rule) + version = netutils.get_ip_version(rule.cidr) + if version == 4: + fw_rules = ipv4_rules + else: + fw_rules = ipv6_rules + + protocol = rule.protocol + if version == 6 and protocol == 'icmp': + protocol = 'icmpv6' + + args = ['-p', protocol, '-s', rule.cidr] + + if protocol in ['udp', 'tcp']: + if rule.from_port == rule.to_port: + args += ['--dport', '%s' % (rule.from_port,)] + else: + args += ['-m', 'multiport', + '--dports', '%s:%s' % (rule.from_port, + rule.to_port)] + elif protocol == 'icmp': + icmp_type = rule.from_port + icmp_code = rule.to_port + + if icmp_type == -1: + icmp_type_arg = None + else: + icmp_type_arg = '%s' % icmp_type + if not icmp_code == -1: + icmp_type_arg += '/%s' % icmp_code + + if icmp_type_arg: + if version == 4: + args += ['-m', 'icmp', '--icmp-type', + icmp_type_arg] + elif version == 6: + args += ['-m', 'icmp6', '--icmpv6-type', + icmp_type_arg] + args += ['-j DROP'] + fw_rules += [' '.join(args)] + return ipv4_rules, ipv6_rules + + def _security_group_chain_name(self, security_group_id): + return 'nova-sg-%s' % (security_group_id,) + + def _instance_chain_name(self, instance): + return 'inst-%s' % (instance['id'],) diff --git a/nova/virt/libvirt/netutils.py b/nova/virt/libvirt/netutils.py new file mode 100644 index 000000000..4d596078a --- /dev/null +++ b/nova/virt/libvirt/netutils.py @@ -0,0 +1,97 @@ +# vim: tabstop=4 shiftwidth=4 softtabstop=4 + +# Copyright 2010 United States Government as represented by the +# Administrator of the National Aeronautics and Space Administration. +# All Rights Reserved. +# Copyright (c) 2010 Citrix Systems, Inc. +# +# Licensed under the Apache License, Version 2.0 (the "License"); you may +# not use this file except in compliance with the License. You may obtain +# a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT +# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the +# License for the specific language governing permissions and limitations +# under the License. + + +"""Network-releated utilities for supporting libvirt connection code.""" + + +import IPy + +from nova import context +from nova import db +from nova import flags +from nova import ipv6 +from nova import utils + + +FLAGS = flags.FLAGS + + +def get_net_and_mask(cidr): + net = IPy.IP(cidr) + return str(net.net()), str(net.netmask()) + + +def get_net_and_prefixlen(cidr): + net = IPy.IP(cidr) + return str(net.net()), str(net.prefixlen()) + + +def get_ip_version(cidr): + net = IPy.IP(cidr) + return int(net.version()) + + +def get_network_info(instance): + # TODO(adiantum) If we will keep this function + # we should cache network_info + admin_context = context.get_admin_context() + + ip_addresses = db.fixed_ip_get_all_by_instance(admin_context, + instance['id']) + networks = db.network_get_all_by_instance(admin_context, + instance['id']) + flavor = db.instance_type_get_by_id(admin_context, + instance['instance_type_id']) + network_info = [] + + for network in networks: + network_ips = [ip for ip in ip_addresses + if ip['network_id'] == network['id']] + + def ip_dict(ip): + return { + 'ip': ip['address'], + 'netmask': network['netmask'], + 'enabled': '1'} + + def ip6_dict(): + prefix = network['cidr_v6'] + mac = instance['mac_address'] + project_id = instance['project_id'] + return { + 'ip': ipv6.to_global(prefix, mac, project_id), + 'netmask': network['netmask_v6'], + 'enabled': '1'} + + mapping = { + 'label': network['label'], + 'gateway': network['gateway'], + 'broadcast': network['broadcast'], + 'mac': instance['mac_address'], + 'rxtx_cap': flavor['rxtx_cap'], + 'dns': [network['dns']], + 'ips': [ip_dict(ip) for ip in network_ips]} + + if FLAGS.use_ipv6: + mapping['ip6s'] = [ip6_dict()] + mapping['gateway6'] = network['gateway_v6'] + + network_info.append((network, mapping)) + return network_info diff --git a/nova/virt/vmwareapi/fake.py b/nova/virt/vmwareapi/fake.py index 4bb467fa9..7370684bd 100644 --- a/nova/virt/vmwareapi/fake.py +++ b/nova/virt/vmwareapi/fake.py @@ -387,12 +387,11 @@ def _add_file(file_path): def _remove_file(file_path):
"""Removes a file reference from the db."""
if _db_content.get("files") is None:
- raise exception.NotFound(_("No files have been added yet"))
+ raise exception.NoFilesFound()
# Check if the remove is for a single file object or for a folder
if file_path.find(".vmdk") != -1:
if file_path not in _db_content.get("files"):
- raise exception.NotFound(_("File- '%s' is not there in the "
- "datastore") % file_path)
+ raise exception.FileNotFound(file_path=file_path)
_db_content.get("files").remove(file_path)
else:
# Removes the files in the folder and the folder too from the db
@@ -579,7 +578,7 @@ class FakeVim(object): """Searches the datastore for a file."""
ds_path = kwargs.get("datastorePath")
if _db_content.get("files", None) is None:
- raise exception.NotFound(_("No files have been added yet"))
+ raise exception.NoFilesFound()
for file in _db_content.get("files"):
if file.find(ds_path) != -1:
task_mdo = create_task(method, "success")
@@ -591,7 +590,7 @@ class FakeVim(object): """Creates a directory in the datastore."""
ds_path = kwargs.get("name")
if _db_content.get("files", None) is None:
- raise exception.NotFound(_("No files have been added yet"))
+ raise exception.NoFilesFound()
_db_content["files"].append(ds_path)
def _set_power_state(self, method, vm_ref, pwr_state="poweredOn"):
diff --git a/nova/virt/vmwareapi/vim.py b/nova/virt/vmwareapi/vim.py index 159e16a80..0cbdba363 100644 --- a/nova/virt/vmwareapi/vim.py +++ b/nova/virt/vmwareapi/vim.py @@ -43,6 +43,7 @@ flags.DEFINE_string('vmwareapi_wsdl_loc', if suds:
+
class VIMMessagePlugin(suds.plugin.MessagePlugin):
def addAttributeForValue(self, node):
diff --git a/nova/virt/vmwareapi/vmops.py b/nova/virt/vmwareapi/vmops.py index cf6c88bbd..c3e79a92f 100644 --- a/nova/virt/vmwareapi/vmops.py +++ b/nova/virt/vmwareapi/vmops.py @@ -100,8 +100,7 @@ class VMWareVMOps(object): """
vm_ref = self._get_vm_ref_from_the_name(instance.name)
if vm_ref:
- raise exception.Duplicate(_("Attempted to create a VM with a name"
- " %s, but that already exists on the host") % instance.name)
+ raise exception.InstanceExists(name=instance.name)
client_factory = self._session._get_vim().client.factory
service_content = self._session._get_vim().get_service_content()
@@ -116,8 +115,7 @@ class VMWareVMOps(object): network_utils.get_network_with_the_name(self._session,
net_name)
if network_ref is None:
- raise exception.NotFound(_("Network with the name '%s' doesn't"
- " exist on the ESX host") % net_name)
+ raise exception.NetworkNotFoundForBridge(bridge=net_name)
_check_if_network_bridge_exists()
@@ -337,8 +335,7 @@ class VMWareVMOps(object): """
vm_ref = self._get_vm_ref_from_the_name(instance.name)
if vm_ref is None:
- raise exception.NotFound(_("instance - %s not present") %
- instance.name)
+ raise exception.InstanceNotFound(instance_id=instance.id)
client_factory = self._session._get_vim().client.factory
service_content = self._session._get_vim().get_service_content()
@@ -388,8 +385,7 @@ class VMWareVMOps(object): "VirtualMachine",
"datastore")
if not ds_ref_ret:
- raise exception.NotFound(_("Failed to get the datastore "
- "reference(s) which the VM uses"))
+ raise exception.DatastoreNotFound()
ds_ref = ds_ref_ret.ManagedObjectReference[0]
ds_browser = vim_util.get_dynamic_property(
self._session._get_vim(),
@@ -480,8 +476,7 @@ class VMWareVMOps(object): """Reboot a VM instance."""
vm_ref = self._get_vm_ref_from_the_name(instance.name)
if vm_ref is None:
- raise exception.NotFound(_("instance - %s not present") %
- instance.name)
+ raise exception.InstanceNotFound(instance_id=instance.id)
lst_properties = ["summary.guest.toolsStatus", "runtime.powerState",
"summary.guest.toolsRunningStatus"]
props = self._session._call_method(vim_util, "get_object_properties",
@@ -501,8 +496,8 @@ class VMWareVMOps(object): # Raise an exception if the VM is not powered On.
if pwr_state not in ["poweredOn"]:
- raise exception.Invalid(_("instance - %s not poweredOn. So can't "
- "be rebooted.") % instance.name)
+ reason = _("instance is not powered on")
+ raise exception.InstanceRebootFailure(reason=reason)
# If latest vmware tools are installed in the VM, and that the tools
# are running, then only do a guest reboot. Otherwise do a hard reset.
@@ -605,8 +600,7 @@ class VMWareVMOps(object): """Suspend the specified instance."""
vm_ref = self._get_vm_ref_from_the_name(instance.name)
if vm_ref is None:
- raise exception.NotFound(_("instance - %s not present") %
- instance.name)
+ raise exception.InstanceNotFound(instance_id=instance.id)
pwr_state = self._session._call_method(vim_util,
"get_dynamic_property", vm_ref,
@@ -620,8 +614,9 @@ class VMWareVMOps(object): LOG.debug(_("Suspended the VM %s ") % instance.name)
# Raise Exception if VM is poweredOff
elif pwr_state == "poweredOff":
- raise exception.Invalid(_("instance - %s is poweredOff and hence "
- " can't be suspended.") % instance.name)
+ reason = _("instance is powered off and can not be suspended.")
+ raise exception.InstanceSuspendFailure(reason=reason)
+
LOG.debug(_("VM %s was already in suspended state. So returning "
"without doing anything") % instance.name)
@@ -629,8 +624,7 @@ class VMWareVMOps(object): """Resume the specified instance."""
vm_ref = self._get_vm_ref_from_the_name(instance.name)
if vm_ref is None:
- raise exception.NotFound(_("instance - %s not present") %
- instance.name)
+ raise exception.InstanceNotFound(instance_id=instance.id)
pwr_state = self._session._call_method(vim_util,
"get_dynamic_property", vm_ref,
@@ -643,15 +637,14 @@ class VMWareVMOps(object): self._wait_with_callback(instance.id, suspend_task, callback)
LOG.debug(_("Resumed the VM %s ") % instance.name)
else:
- raise exception.Invalid(_("instance - %s not in Suspended state "
- "and hence can't be Resumed.") % instance.name)
+ reason = _("instance is not in a suspended state")
+ raise exception.InstanceResumeFailure(reason=reason)
def get_info(self, instance_name):
"""Return data about the VM instance."""
vm_ref = self._get_vm_ref_from_the_name(instance_name)
if vm_ref is None:
- raise exception.NotFound(_("instance - %s not present") %
- instance_name)
+ raise exception.InstanceNotFound(instance_id=instance_name)
lst_properties = ["summary.config.numCpu",
"summary.config.memorySizeMB",
@@ -687,8 +680,7 @@ class VMWareVMOps(object): """Return snapshot of console."""
vm_ref = self._get_vm_ref_from_the_name(instance.name)
if vm_ref is None:
- raise exception.NotFound(_("instance - %s not present") %
- instance.name)
+ raise exception.InstanceNotFound(instance_id=instance.id)
param_list = {"id": str(vm_ref)}
base_url = "%s://%s/screen?%s" % (self._session._scheme,
self._session._host_ip,
@@ -716,8 +708,7 @@ class VMWareVMOps(object): """
vm_ref = self._get_vm_ref_from_the_name(instance.name)
if vm_ref is None:
- raise exception.NotFound(_("instance - %s not present") %
- instance.name)
+ raise exception.InstanceNotFound(instance_id=instance.id)
network = db.network_get_by_instance(context.get_admin_context(),
instance['id'])
mac_addr = instance.mac_address
diff --git a/nova/virt/vmwareapi_conn.py b/nova/virt/vmwareapi_conn.py index 20c1b2b45..1c6d2572d 100644 --- a/nova/virt/vmwareapi_conn.py +++ b/nova/virt/vmwareapi_conn.py @@ -42,6 +42,7 @@ from nova import exception from nova import flags
from nova import log as logging
from nova import utils
+from nova.virt import driver
from nova.virt.vmwareapi import error_util
from nova.virt.vmwareapi import vim
from nova.virt.vmwareapi import vim_util
@@ -104,11 +105,12 @@ def get_connection(_): api_retry_count)
-class VMWareESXConnection(object):
+class VMWareESXConnection(driver.ComputeDriver):
"""The ESX host connection object."""
def __init__(self, host_ip, host_username, host_password,
api_retry_count, scheme="https"):
+ super(VMWareESXConnection, self).__init__()
session = VMWareAPISession(host_ip, host_username, host_password,
api_retry_count, scheme=scheme)
self._vmops = VMWareVMOps(session)
diff --git a/nova/virt/xenapi/fake.py b/nova/virt/xenapi/fake.py index 4434dbf0b..e36ef3288 100644 --- a/nova/virt/xenapi/fake.py +++ b/nova/virt/xenapi/fake.py @@ -294,7 +294,7 @@ class Failure(Exception): def __str__(self): try: return str(self.details) - except Exception, exc: + except Exception: return "XenAPI Fake Failure: %s" % str(self.details) def _details_map(self): diff --git a/nova/virt/xenapi/vm_utils.py b/nova/virt/xenapi/vm_utils.py index d07d60800..9f6cd608c 100644 --- a/nova/virt/xenapi/vm_utils.py +++ b/nova/virt/xenapi/vm_utils.py @@ -28,10 +28,7 @@ import urllib import uuid from xml.dom import minidom -from eventlet import event import glance.client -from nova import context -from nova import db from nova import exception from nova import flags from nova import log as logging @@ -49,6 +46,10 @@ LOG = logging.getLogger("nova.virt.xenapi.vm_utils") FLAGS = flags.FLAGS flags.DEFINE_string('default_os_type', 'linux', 'Default OS type') +flags.DEFINE_integer('block_device_creation_timeout', 10, + 'time to wait for a block device to be created') +flags.DEFINE_integer('max_kernel_ramdisk_size', 16 * 1024 * 1024, + 'maximum size in bytes of kernel or ramdisk images') XENAPI_POWER_STATE = { 'Halted': power_state.SHUTDOWN, @@ -101,8 +102,8 @@ class VMHelper(HelperBase): 3. Using hardware virtualization """ - instance_type = instance_types.\ - get_instance_type(instance.instance_type) + inst_type_id = instance.instance_type_id + instance_type = instance_types.get_instance_type(inst_type_id) mem = str(long(instance_type['memory_mb']) * 1024 * 1024) vcpus = str(instance_type['vcpus']) rec = { @@ -169,8 +170,8 @@ class VMHelper(HelperBase): @classmethod def ensure_free_mem(cls, session, instance): - instance_type = instance_types.get_instance_type( - instance.instance_type) + inst_type_id = instance.instance_type_id + instance_type = instance_types.get_instance_type(inst_type_id) mem = long(instance_type['memory_mb']) * 1024 * 1024 #get free memory from host host = session.get_xenapi_host() @@ -304,7 +305,6 @@ class VMHelper(HelperBase): % locals()) vm_vdi_ref, vm_vdi_rec = cls.get_vdi_for_vm_safely(session, vm_ref) - vm_vdi_uuid = vm_vdi_rec["uuid"] sr_ref = vm_vdi_rec["SR"] original_parent_uuid = get_vhd_parent_uuid(session, vm_vdi_ref) @@ -446,6 +446,12 @@ class VMHelper(HelperBase): if image_type == ImageType.DISK: # Make room for MBR. vdi_size += MBR_SIZE_BYTES + elif image_type == ImageType.KERNEL_RAMDISK and \ + vdi_size > FLAGS.max_kernel_ramdisk_size: + max_size = FLAGS.max_kernel_ramdisk_size + raise exception.Error( + _("Kernel/Ramdisk image is too large: %(vdi_size)d bytes, " + "max %(max_size)d bytes") % locals()) name_label = get_name_label_for_image(image) vdi_ref = cls.create_vdi(session, sr_ref, name_label, vdi_size, False) @@ -508,9 +514,7 @@ class VMHelper(HelperBase): try: return glance_disk_format2nova_type[disk_format] except KeyError: - raise exception.NotFound( - _("Unrecognized disk_format '%(disk_format)s'") - % locals()) + raise exception.InvalidDiskFormat(disk_format=disk_format) def determine_from_instance(): if instance.kernel_id: @@ -645,8 +649,7 @@ class VMHelper(HelperBase): if n == 0: return None elif n > 1: - raise exception.Duplicate(_('duplicate name found: %s') % - name_label) + raise exception.InstanceExists(name=name_label) else: return vm_refs[0] @@ -753,14 +756,14 @@ class VMHelper(HelperBase): session.call_xenapi('SR.scan', sr_ref) -def get_rrd(host, uuid): +def get_rrd(host, vm_uuid): """Return the VM RRD XML as a string""" try: xml = urllib.urlopen("http://%s:%s@%s/vm_rrd?uuid=%s" % ( FLAGS.xenapi_connection_username, FLAGS.xenapi_connection_password, host, - uuid)) + vm_uuid)) return xml.read() except IOError: return None @@ -855,7 +858,7 @@ def safe_find_sr(session): """ sr_ref = find_sr(session) if sr_ref is None: - raise exception.NotFound(_('Cannot find SR to read/write VDI')) + raise exception.StorageRepositoryNotFound() return sr_ref @@ -896,6 +899,16 @@ def remap_vbd_dev(dev): return remapped_dev +def _wait_for_device(dev): + """Wait for device node to appear""" + for i in xrange(0, FLAGS.block_device_creation_timeout): + if os.path.exists('/dev/%s' % dev): + return + time.sleep(1) + + raise StorageError(_('Timeout waiting for device %s to be created') % dev) + + def with_vdi_attached_here(session, vdi_ref, read_only, f): this_vm_ref = get_this_vm_ref(session) vbd_rec = {} @@ -924,6 +937,11 @@ def with_vdi_attached_here(session, vdi_ref, read_only, f): if dev != orig_dev: LOG.debug(_('VBD %(vbd_ref)s plugged into wrong dev, ' 'remapping to %(dev)s') % locals()) + if dev != 'autodetect': + # NOTE(johannes): Unit tests will end up with a device called + # 'autodetect' which obviously won't exist. It's not ideal, + # but the alternatives were much messier + _wait_for_device(dev) return f(dev) finally: LOG.debug(_('Destroying VBD for VDI %s ... '), vdi_ref) @@ -1003,7 +1021,6 @@ def _stream_disk(dev, image_type, virtual_size, image_file): def _write_partition(virtual_size, dev): dest = '/dev/%s' % dev - mbr_last = MBR_SIZE_SECTORS - 1 primary_first = MBR_SIZE_SECTORS primary_last = MBR_SIZE_SECTORS + (virtual_size / SECTOR_SIZE) - 1 @@ -1130,7 +1147,7 @@ def _prepare_injectables(inst, networks_info): 'dns': dns, 'address_v6': ip_v6 and ip_v6['ip'] or '', 'netmask_v6': ip_v6 and ip_v6['netmask'] or '', - 'gateway_v6': ip_v6 and ip_v6['gateway'] or '', + 'gateway_v6': ip_v6 and info['gateway6'] or '', 'use_ipv6': FLAGS.use_ipv6} interfaces_info.append(interface_info) diff --git a/nova/virt/xenapi/vmops.py b/nova/virt/xenapi/vmops.py index c96c35a6e..be6ef48ea 100644 --- a/nova/virt/xenapi/vmops.py +++ b/nova/virt/xenapi/vmops.py @@ -25,15 +25,15 @@ import M2Crypto import os import pickle import subprocess -import tempfile import uuid -from nova import db from nova import context -from nova import log as logging +from nova import db from nova import exception -from nova import utils from nova import flags +from nova import ipv6 +from nova import log as logging +from nova import utils from nova.auth.manager import AuthManager from nova.compute import power_state @@ -127,8 +127,7 @@ class VMOps(object): instance_name = instance.name vm_ref = VMHelper.lookup(self._session, instance_name) if vm_ref is not None: - raise exception.Duplicate(_('Attempted to create' - ' non-unique name %s') % instance_name) + raise exception.InstanceExists(name=instance_name) #ensure enough free memory is available if not VMHelper.ensure_free_mem(self._session, instance): @@ -176,7 +175,7 @@ class VMOps(object): vdi_ref, network_info) self.create_vifs(vm_ref, network_info) - self.inject_network_info(instance, vm_ref, network_info) + self.inject_network_info(instance, network_info, vm_ref) return vm_ref def _spawn(self, instance, vm_ref): @@ -203,6 +202,13 @@ class VMOps(object): for path, contents in instance.injected_files: LOG.debug(_("Injecting file path: '%s'") % path) self.inject_file(instance, path, contents) + + def _set_admin_password(): + admin_password = instance.admin_pass + if admin_password: + LOG.debug(_("Setting admin password")) + self.set_admin_password(instance, admin_password) + # NOTE(armando): Do we really need to do this in virt? # NOTE(tr3buchet): not sure but wherever we do it, we need to call # reset_network afterwards @@ -211,20 +217,15 @@ class VMOps(object): def _wait_for_boot(): try: state = self.get_info(instance_name)['state'] - db.instance_set_state(context.get_admin_context(), - instance['id'], state) if state == power_state.RUNNING: LOG.debug(_('Instance %s: booted'), instance_name) timer.stop() _inject_files() + _set_admin_password() return True except Exception, exc: LOG.warn(exc) - LOG.exception(_('instance %s: failed to boot'), - instance_name) - db.instance_set_state(context.get_admin_context(), - instance['id'], - power_state.SHUTDOWN) + LOG.exception(_('Instance %s: failed to boot'), instance_name) timer.stop() return False @@ -260,8 +261,8 @@ class VMOps(object): instance_name = instance_or_vm.name vm_ref = VMHelper.lookup(self._session, instance_name) if vm_ref is None: - raise exception.NotFound( - _('Instance not present %s') % instance_name) + raise exception.NotFound(_("No opaque_ref could be determined " + "for '%s'.") % instance_or_vm) return vm_ref def _acquire_bootlock(self, vm): @@ -387,7 +388,6 @@ class VMOps(object): def link_disks(self, instance, base_copy_uuid, cow_uuid): """Links the base copy VHD to the COW via the XAPI plugin.""" - vm_ref = VMHelper.lookup(self._session, instance.name) new_base_copy_uuid = str(uuid.uuid4()) new_cow_uuid = str(uuid.uuid4()) params = {'instance_id': instance.id, @@ -437,11 +437,12 @@ class VMOps(object): """ # Need to uniquely identify this request. - transaction_id = str(uuid.uuid4()) + key_init_transaction_id = str(uuid.uuid4()) # The simple Diffie-Hellman class is used to manage key exchange. dh = SimpleDH() - args = {'id': transaction_id, 'pub': str(dh.get_public())} - resp = self._make_agent_call('key_init', instance, '', args) + key_init_args = {'id': key_init_transaction_id, + 'pub': str(dh.get_public())} + resp = self._make_agent_call('key_init', instance, '', key_init_args) if resp is None: # No response from the agent return @@ -455,8 +456,9 @@ class VMOps(object): dh.compute_shared(agent_pub) enc_pass = dh.encrypt(new_pass) # Send the encrypted password - args['enc_pass'] = enc_pass - resp = self._make_agent_call('password', instance, '', args) + password_transaction_id = str(uuid.uuid4()) + password_args = {'id': password_transaction_id, 'enc_pass': enc_pass} + resp = self._make_agent_call('password', instance, '', password_args) if resp is None: # No response from the agent return @@ -464,6 +466,9 @@ class VMOps(object): # Successful return code from password is '0' if resp_dict['returncode'] != '0': raise RuntimeError(resp_dict['message']) + db.instance_update(context.get_admin_context(), + instance['id'], + dict(admin_pass=new_pass)) return resp_dict['message'] def inject_file(self, instance, path, contents): @@ -579,9 +584,8 @@ class VMOps(object): if not (instance.kernel_id and instance.ramdisk_id): # 2. We only have kernel xor ramdisk - raise exception.NotFound( - _("Instance %(instance_id)s has a kernel or ramdisk but not " - "both" % locals())) + raise exception.InstanceUnacceptable(instance_id=instance_id, + reason=_("instance has a kernel or ramdisk but not both")) # 3. We have both kernel and ramdisk (kernel, ramdisk) = VMHelper.lookup_kernel_ramdisk(self._session, @@ -722,8 +726,7 @@ class VMOps(object): "%s-rescue" % instance.name) if not rescue_vm_ref: - raise exception.NotFound(_( - "Instance is not in Rescue Mode: %s" % instance.name)) + raise exception.InstanceNotInRescueMode(instance_id=instance.id) original_vm_ref = VMHelper.lookup(self._session, instance.name) instance._rescue = False @@ -760,7 +763,6 @@ class VMOps(object): instance))) for vm in rescue_vms: - rescue_name = vm["name"] rescue_vm_ref = vm["vm_ref"] self._destroy_rescue_instance(rescue_vm_ref) @@ -798,15 +800,17 @@ class VMOps(object): def _get_network_info(self, instance): """Creates network info list for instance.""" admin_context = context.get_admin_context() - IPs = db.fixed_ip_get_all_by_instance(admin_context, + ips = db.fixed_ip_get_all_by_instance(admin_context, instance['id']) networks = db.network_get_all_by_instance(admin_context, instance['id']) - flavor = db.instance_type_get_by_name(admin_context, - instance['instance_type']) + + inst_type = db.instance_type_get_by_id(admin_context, + instance['instance_type_id']) + network_info = [] for network in networks: - network_IPs = [ip for ip in IPs if ip.network_id == network.id] + network_ips = [ip for ip in ips if ip.network_id == network.id] def ip_dict(ip): return { @@ -814,12 +818,12 @@ class VMOps(object): "netmask": network["netmask"], "enabled": "1"} - def ip6_dict(ip6): + def ip6_dict(): return { - "ip": utils.to_global_ipv6(network['cidr_v6'], - instance['mac_address']), + "ip": ipv6.to_global(network['cidr_v6'], + instance['mac_address'], + instance['project_id']), "netmask": network['netmask_v6'], - "gateway": network['gateway_v6'], "enabled": "1"} info = { @@ -827,23 +831,41 @@ class VMOps(object): 'gateway': network['gateway'], 'broadcast': network['broadcast'], 'mac': instance.mac_address, - 'rxtx_cap': flavor['rxtx_cap'], + 'rxtx_cap': inst_type['rxtx_cap'], 'dns': [network['dns']], - 'ips': [ip_dict(ip) for ip in network_IPs]} + 'ips': [ip_dict(ip) for ip in network_ips]} if network['cidr_v6']: - info['ip6s'] = [ip6_dict(ip) for ip in network_IPs] + info['ip6s'] = [ip6_dict()] + if network['gateway_v6']: + info['gateway6'] = network['gateway_v6'] network_info.append((network, info)) return network_info - def inject_network_info(self, instance, vm_ref, network_info): + #TODO{tr3buchet) remove this shim with nova-multi-nic + def inject_network_info(self, instance, network_info=None, vm_ref=None): + """ + shim in place which makes inject_network_info work without being + passed network_info. + shim goes away after nova-multi-nic + """ + if not network_info: + network_info = self._get_network_info(instance) + self._inject_network_info(instance, network_info, vm_ref) + + def _inject_network_info(self, instance, network_info, vm_ref=None): """ Generate the network info and make calls to place it into the xenstore and the xenstore param list. + vm_ref can be passed in because it will sometimes be different than + what VMHelper.lookup(session, instance.name) will find (ex: rescue) """ logging.debug(_("injecting network info to xs for vm: |%s|"), vm_ref) - # this function raises if vm_ref is not a vm_opaque_ref - self._session.get_xenapi().VM.get_record(vm_ref) + if vm_ref: + # this function raises if vm_ref is not a vm_opaque_ref + self._session.get_xenapi().VM.get_record(vm_ref) + else: + vm_ref = VMHelper.lookup(self._session, instance.name) for (network, info) in network_info: location = 'vm-data/networking/%s' % info['mac'].replace(':', '') @@ -875,8 +897,10 @@ class VMOps(object): VMHelper.create_vif(self._session, vm_ref, network_ref, mac_address, device, rxtx_cap) - def reset_network(self, instance, vm_ref): + def reset_network(self, instance, vm_ref=None): """Creates uuid arg to pass to make_agent_call and calls it.""" + if not vm_ref: + vm_ref = VMHelper.lookup(self._session, instance.name) args = {'id': str(uuid.uuid4())} # TODO(tr3buchet): fix function call after refactor #resp = self._make_agent_call('resetnetwork', instance, '', args) @@ -902,7 +926,7 @@ class VMOps(object): try: ret = self._make_xenstore_call('read_record', vm, path, {'ignore_missing_path': 'True'}) - except self.XenAPI.Failure, e: + except self.XenAPI.Failure: return None ret = json.loads(ret) if ret == "None": @@ -1150,23 +1174,22 @@ class SimpleDH(object): return mpi def _run_ssl(self, text, which): - base_cmd = ('cat %(tmpfile)s | openssl enc -aes-128-cbc ' - '-a -pass pass:%(shared)s -nosalt %(dec_flag)s') + base_cmd = ('openssl enc -aes-128-cbc -a -pass pass:%(shared)s ' + '-nosalt %(dec_flag)s') if which.lower()[0] == 'd': dec_flag = ' -d' else: dec_flag = '' - fd, tmpfile = tempfile.mkstemp() - os.close(fd) - file(tmpfile, 'w').write(text) shared = self._shared cmd = base_cmd % locals() proc = _runproc(cmd) + proc.stdin.write(text + '\n') + proc.stdin.close() proc.wait() err = proc.stderr.read() if err: raise RuntimeError(_('OpenSSL error: %s') % err) - return proc.stdout.read() + return proc.stdout.read().strip('\n') def encrypt(self, text): return self._run_ssl(text, 'enc') diff --git a/nova/virt/xenapi/volume_utils.py b/nova/virt/xenapi/volume_utils.py index 72284ac02..7821a4f7e 100644 --- a/nova/virt/xenapi/volume_utils.py +++ b/nova/virt/xenapi/volume_utils.py @@ -204,14 +204,17 @@ def _get_volume_id(path_or_id): if isinstance(path_or_id, int): return path_or_id # n must contain at least the volume_id - # /vol- is for remote volumes - # -vol- is for local volumes + # :volume- is for remote volumes + # -volume- is for local volumes # see compute/manager->setup_compute_volume - volume_id = path_or_id[path_or_id.find('/vol-') + 1:] + volume_id = path_or_id[path_or_id.find(':volume-') + 1:] if volume_id == path_or_id: - volume_id = path_or_id[path_or_id.find('-vol-') + 1:] - volume_id = volume_id.replace('--', '-') - return volume_id + volume_id = path_or_id[path_or_id.find('-volume--') + 1:] + volume_id = volume_id.replace('volume--', '') + else: + volume_id = volume_id.replace('volume-', '') + volume_id = volume_id[0:volume_id.find('-')] + return int(volume_id) def _get_target_host(iscsi_string): @@ -244,25 +247,23 @@ def _get_target(volume_id): Gets iscsi name and portal from volume name and host. For this method to work the following are needed: 1) volume_ref['host'] must resolve to something rather than loopback - 2) ietd must bind only to the address as resolved above - If any of the two conditions are not met, fall back on Flags. """ - volume_ref = db.volume_get_by_ec2_id(context.get_admin_context(), - volume_id) + volume_ref = db.volume_get(context.get_admin_context(), + volume_id) result = (None, None) try: - (r, _e) = utils.execute("sudo iscsiadm -m discovery -t " - "sendtargets -p %s" % - volume_ref['host']) + (r, _e) = utils.execute('sudo', 'iscsiadm', + '-m', 'discovery', + '-t', 'sendtargets', + '-p', volume_ref['host']) except exception.ProcessExecutionError, exc: LOG.exception(exc) else: - targets = r.splitlines() - if len(_e) == 0 and len(targets) == 1: - for target in targets: - if volume_id in target: - (location, _sep, iscsi_name) = target.partition(" ") - break - iscsi_portal = location.split(",")[0] - result = (iscsi_name, iscsi_portal) + volume_name = "volume-%08x" % volume_id + for target in r.splitlines(): + if FLAGS.iscsi_ip_prefix in target and volume_name in target: + (location, _sep, iscsi_name) = target.partition(" ") + break + iscsi_portal = location.split(",")[0] + result = (iscsi_name, iscsi_portal) return result diff --git a/nova/virt/xenapi/volumeops.py b/nova/virt/xenapi/volumeops.py index 757ecf5ad..afcb8cf47 100644 --- a/nova/virt/xenapi/volumeops.py +++ b/nova/virt/xenapi/volumeops.py @@ -45,8 +45,7 @@ class VolumeOps(object): # Before we start, check that the VM exists vm_ref = VMHelper.lookup(self._session, instance_name) if vm_ref is None: - raise exception.NotFound(_('Instance %s not found') - % instance_name) + raise exception.InstanceNotFound(instance_id=instance_name) # NOTE: No Resource Pool concept so far LOG.debug(_("Attach_volume: %(instance_name)s, %(device_path)s," " %(mountpoint)s") % locals()) @@ -98,8 +97,7 @@ class VolumeOps(object): # Before we start, check that the VM exists vm_ref = VMHelper.lookup(self._session, instance_name) if vm_ref is None: - raise exception.NotFound(_('Instance %s not found') - % instance_name) + raise exception.InstanceNotFound(instance_id=instance_name) # Detach VBD from VM LOG.debug(_("Detach_volume: %(instance_name)s, %(mountpoint)s") % locals()) diff --git a/nova/virt/xenapi_conn.py b/nova/virt/xenapi_conn.py index 99fd35c61..6d828e109 100644 --- a/nova/virt/xenapi_conn.py +++ b/nova/virt/xenapi_conn.py @@ -57,19 +57,24 @@ reactor thread if the VM.get_by_name_label or VM.get_record calls block. - suffix "_rec" for record objects """ +import json +import random import sys import urlparse import xmlrpclib from eventlet import event from eventlet import tpool +from eventlet import timeout from nova import context from nova import db +from nova import exception from nova import utils from nova import flags from nova import log as logging from nova.virt import driver +from nova.virt.xenapi import vm_utils from nova.virt.xenapi.vmops import VMOps from nova.virt.xenapi.volumeops import VolumeOps @@ -140,6 +145,9 @@ flags.DEFINE_bool('xenapi_remap_vbd_dev', False, flags.DEFINE_string('xenapi_remap_vbd_dev_prefix', 'sd', 'Specify prefix to remap VBD dev to ' '(ex. /dev/xvdb -> /dev/sdb)') +flags.DEFINE_integer('xenapi_login_timeout', + 10, + 'Timeout in seconds for XenAPI login.') def get_connection(_): @@ -161,9 +169,16 @@ class XenAPIConnection(driver.ComputeDriver): def __init__(self, url, user, pw): super(XenAPIConnection, self).__init__() - session = XenAPISession(url, user, pw) - self._vmops = VMOps(session) - self._volumeops = VolumeOps(session) + self._session = XenAPISession(url, user, pw) + self._vmops = VMOps(self._session) + self._volumeops = VolumeOps(self._session) + self._host_state = None + + @property + def HostState(self): + if not self._host_state: + self._host_state = HostState(self._session) + return self._host_state def init_host(self, host): #FIXME(armando): implement this @@ -311,6 +326,16 @@ class XenAPIConnection(driver.ComputeDriver): """This method is supported only by libvirt.""" raise NotImplementedError('This method is supported only by libvirt.') + def update_host_status(self): + """Update the status info of the host, and return those values + to the calling program.""" + return self.HostState.update_status() + + def get_host_stats(self, refresh=False): + """Return the current state of the host. If 'refresh' is + True, run the update first.""" + return self.HostState.get_host_stats(refresh=refresh) + class XenAPISession(object): """The session to invoke XenAPI SDK calls""" @@ -318,8 +343,10 @@ class XenAPISession(object): def __init__(self, url, user, pw): self.XenAPI = self.get_imported_xenapi() self._session = self._create_session(url) - self._session.login_with_password(user, pw) - self.loop = None + exception = self.XenAPI.Failure(_("Unable to log in to XenAPI " + "(is the Dom0 disk full?)")) + with timeout.Timeout(FLAGS.xenapi_login_timeout, exception): + self._session.login_with_password(user, pw) def get_imported_xenapi(self): """Stubout point. This can be replaced with a mock xenapi module.""" @@ -356,57 +383,52 @@ class XenAPISession(object): def wait_for_task(self, task, id=None): """Return the result of the given task. The task is polled - until it completes. Not re-entrant.""" + until it completes.""" done = event.Event() - self.loop = utils.LoopingCall(self._poll_task, id, task, done) - self.loop.start(FLAGS.xenapi_task_poll_interval, now=True) - rv = done.wait() - self.loop.stop() - return rv - - def _stop_loop(self): - """Stop polling for task to finish.""" - #NOTE(sandy-walsh) Had to break this call out to support unit tests. - if self.loop: - self.loop.stop() + loop = utils.LoopingCall(f=None) + + def _poll_task(): + """Poll the given XenAPI task, and return the result if the + action was completed successfully or not. + """ + try: + name = self._session.xenapi.task.get_name_label(task) + status = self._session.xenapi.task.get_status(task) + if id: + action = dict( + instance_id=int(id), + action=name[0:255], # Ensure action is never > 255 + error=None) + if status == "pending": + return + elif status == "success": + result = self._session.xenapi.task.get_result(task) + LOG.info(_("Task [%(name)s] %(task)s status:" + " success %(result)s") % locals()) + done.send(_parse_xmlrpc_value(result)) + else: + error_info = self._session.xenapi.task.get_error_info(task) + action["error"] = str(error_info) + LOG.warn(_("Task [%(name)s] %(task)s status:" + " %(status)s %(error_info)s") % locals()) + done.send_exception(self.XenAPI.Failure(error_info)) + + if id: + db.instance_action_create(context.get_admin_context(), + action) + except self.XenAPI.Failure, exc: + LOG.warn(exc) + done.send_exception(*sys.exc_info()) + loop.stop() + + loop.f = _poll_task + loop.start(FLAGS.xenapi_task_poll_interval, now=True) + return done.wait() def _create_session(self, url): """Stubout point. This can be replaced with a mock session.""" return self.XenAPI.Session(url) - def _poll_task(self, id, task, done): - """Poll the given XenAPI task, and fire the given action if we - get a result. - """ - try: - name = self._session.xenapi.task.get_name_label(task) - status = self._session.xenapi.task.get_status(task) - if id: - action = dict( - instance_id=int(id), - action=name[0:255], # Ensure action is never > 255 - error=None) - if status == "pending": - return - elif status == "success": - result = self._session.xenapi.task.get_result(task) - LOG.info(_("Task [%(name)s] %(task)s status:" - " success %(result)s") % locals()) - done.send(_parse_xmlrpc_value(result)) - else: - error_info = self._session.xenapi.task.get_error_info(task) - action["error"] = str(error_info) - LOG.warn(_("Task [%(name)s] %(task)s status:" - " %(status)s %(error_info)s") % locals()) - done.send_exception(self.XenAPI.Failure(error_info)) - - if id: - db.instance_action_create(context.get_admin_context(), action) - except self.XenAPI.Failure, exc: - LOG.warn(exc) - done.send_exception(*sys.exc_info()) - self._stop_loop() - def _unwrap_plugin_exceptions(self, func, *args, **kwargs): """Parse exception details""" try: @@ -429,6 +451,65 @@ class XenAPISession(object): raise +class HostState(object): + """Manages information about the XenServer host this compute + node is running on. + """ + def __init__(self, session): + super(HostState, self).__init__() + self._session = session + self._stats = {} + self.update_status() + + def get_host_stats(self, refresh=False): + """Return the current state of the host. If 'refresh' is + True, run the update first. + """ + if refresh: + self.update_status() + return self._stats + + def update_status(self): + """Since under Xenserver, a compute node runs on a given host, + we can get host status information using xenapi. + """ + LOG.debug(_("Updating host stats")) + # Make it something unlikely to match any actual instance ID + task_id = random.randint(-80000, -70000) + task = self._session.async_call_plugin("xenhost", "host_data", {}) + task_result = self._session.wait_for_task(task, task_id) + if not task_result: + task_result = json.dumps("") + try: + data = json.loads(task_result) + except ValueError as e: + # Invalid JSON object + LOG.error(_("Unable to get updated status: %s") % e) + return + # Get the SR usage + try: + sr_ref = vm_utils.safe_find_sr(self._session) + except exception.NotFound as e: + # No SR configured + LOG.error(_("Unable to get SR for this host: %s") % e) + return + sr_rec = self._session.get_xenapi().SR.get_record(sr_ref) + total = int(sr_rec["virtual_allocation"]) + used = int(sr_rec["physical_utilisation"]) + data["disk_total"] = total + data["disk_used"] = used + data["disk_available"] = total - used + host_memory = data.get('host_memory', None) + if host_memory: + data["host_memory_total"] = host_memory.get('total', 0) + data["host_memory_overhead"] = host_memory.get('overhead', 0) + data["host_memory_free"] = host_memory.get('free', 0) + data["host_memory_free_computed"] = \ + host_memory.get('free-computed', 0) + del data['host_memory'] + self._stats = data + + def _parse_xmlrpc_value(val): """Parse the given value as if it were an XML-RPC value. This is sometimes used as the format for the task.result field.""" |
