Synnefo

# Copyright 2013 GRNET S.A. All rights reserved.

#

# Redistribution and use in source and binary forms, with or

# without modification, are permitted provided that the following

# conditions are met:

#

#   1. Redistributions of source code must retain the above

#      copyright notice, this list of conditions and the following

#      disclaimer.

#

#   2. Redistributions in binary form must reproduce the above

#      copyright notice, this list of conditions and the following

#      disclaimer in the documentation and/or other materials

#      provided with the distribution.

#

# THIS SOFTWARE IS PROVIDED BY GRNET S.A. ``AS IS'' AND ANY EXPRESS

# OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED

# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR

# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GRNET S.A OR

# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,

# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT

# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF

# USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED

# AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT

# LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN

# ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE

# POSSIBILITY OF SUCH DAMAGE.

#

# The views and conclusions contained in the software and

# documentation are those of the authors and should not be

# interpreted as representing official policies, either expressed

# or implied, of GRNET S.A.

import time

from synnefo.db.models import VirtualMachine, IPAddress, NetworkInterface

from synnefo.logic import servers

from synnefo.logic import ips as logic_ips

from synnefo.logic import backend

MiB = 2 ** 20

GiB = 2 ** 30

def _partition_by(f, l, convert=None):

    if convert is None:

        convert = lambda x: x

    d = {}

    for x in l:

        group = f(x)

        group_l = d.get(group, [])

        group_l.append(convert(x))

        d[group] = group_l

    return d

CHANGE = {

    "cyclades.ram": lambda vm: vm.flavor.ram * MiB,

    "cyclades.cpu": lambda vm: vm.flavor.cpu,

    "cyclades.vm": lambda vm: 1,

    "cyclades.total_ram": lambda vm: vm.flavor.ram * MiB,

    "cyclades.total_cpu": lambda vm: vm.flavor.cpu,

    "cyclades.disk": lambda vm: vm.flavor.disk * GiB,

    "cyclades.floating_ip": lambda vm: 1,

    }

def wait_server_job(server):

    jobID = server.task_job_id

    client = server.get_client()

    status, error = backend.wait_for_job(client, jobID)

    if status != "success":

        raise ValueError(error)

VM_SORT_LEVEL = {

    "ERROR": 4,

    "BUILD": 3,

    "STOPPED": 2,

    "STARTED": 1,

    "RESIZE": 1,

    "DESTROYED": 0,

    }

def sort_vms():

    def f(vm):

        level = VM_SORT_LEVEL[vm.operstate]

        return (level, vm.id)

    return f

def handle_stop_active(viol_id, resource, vms, diff, actions):

    vm_actions = actions["vm"]

    vms = [vm for vm in vms if vm.operstate in ["STARTED", "BUILD", "ERROR"]]

    vms = sorted(vms, key=sort_vms(), reverse=True)

    for vm in vms:

        if diff < 1:

            break

        diff -= CHANGE[resource](vm)

        if vm_actions.get(vm.id) is None:

            action = "REMOVE" if vm.operstate == "ERROR" else "SHUTDOWN"

            vm_actions[vm.id] = viol_id, vm.operstate, action

def handle_destroy(viol_id, resource, vms, diff, actions):

    vm_actions = actions["vm"]

    vms = sorted(vms, key=sort_vms(), reverse=True)

    for vm in vms:

        if diff < 1:

            break

        diff -= CHANGE[resource](vm)

        vm_actions[vm.id] = viol_id, vm.operstate, "REMOVE"

def _state_after_action(vm, action):

    if action == "REMOVE":

        return "ERROR"  # highest

    if action == "SHUTDOWN":

        return "STOPPED"

    return vm.operstate  # no action

def sort_ips(vm_actions):

    def f(ip):

        if not ip.in_use():

            level = 5

        else:

            machine = ip.nic.machine

            _, _, action = vm_actions.get(machine.id, (None, None, None))

            level = VM_SORT_LEVEL[_state_after_action(machine, action)]

        return (level, ip.id)

    return f

def handle_floating_ip(viol_id, resource, ips, diff, actions):

    vm_actions = actions.get("vm", {})

    ip_actions = actions["floating_ip"]

    ips = sorted(ips, key=sort_ips(vm_actions), reverse=True)

    for ip in ips:

        if diff < 1:

            break

        diff -= CHANGE[resource](ip)

        state = "USED" if ip.in_use() else "FREE"

        ip_actions[ip.id] = viol_id, state, "REMOVE"

def get_vms(users=None):

    vms = VirtualMachine.objects.filter(deleted=False).\

        select_related("flavor").order_by('-id')

    if users is not None:

        vms = vms.filter(userid__in=users)

    return _partition_by(lambda vm: vm.userid, vms)

def get_floating_ips(users=None):

    ips = IPAddress.objects.filter(deleted=False, floating_ip=True).\

        select_related("nic__machine")

    if users is not None:

        ips = ips.filter(userid__in=users)

    return _partition_by(lambda ip: ip.userid, ips)

def get_actual_resources(resource_type, users=None):

    ACTUAL_RESOURCES = {

        "vm": get_vms,

        "floating_ip": get_floating_ips,

        }

    return ACTUAL_RESOURCES[resource_type](users=users)

VM_ACTION = {

    "REMOVE": servers.destroy,

    "SHUTDOWN": servers.stop,

}

def apply_to_vm(action, vm_id):

    try:

        vm = VirtualMachine.objects.select_for_update().get(id=vm_id)

        VM_ACTION[action](vm)

        return True

    except BaseException:

        return False

def perform_vm_actions(actions, fix=False):

    log = []

    for vm_id, (viol_id, state, vm_action) in actions.iteritems():

        data = ("vm", vm_id, state, vm_action, viol_id)

        if fix:

            r = apply_to_vm(vm_action, vm_id)

            data += ("DONE" if r else "FAILED",)

        log.append(data)

    return log

def wait_for_ip(ip_id):

    for i in range(100):

        ip = IPAddress.objects.get(id=ip_id)

        if ip.nic_id is None:

            objs = IPAddress.objects.select_for_update()

            return objs.get(id=ip_id)

        time.sleep(1)

    raise ValueError(

        "Floating_ip %s: Waiting for port delete timed out." % ip_id)

def remove_ip(ip_id):

    try:

        ip = IPAddress.objects.select_for_update().get(id=ip_id)

        port_id = ip.nic_id

        if port_id:

            objs = NetworkInterface.objects.select_for_update()

            port = objs.get(id=port_id)

            servers.delete_port(port)

            if port.machine:

                wait_server_job(port.machine)

            ip = wait_for_ip(ip_id)

        logic_ips.delete_floating_ip(ip)

        return True

    except BaseException:

        return False

def perform_floating_ip_actions(actions, fix=False):

    log = []

    for ip_id, (viol_id, state, ip_action) in actions.iteritems():

        data = ("floating_ip", ip_id, state, ip_action, viol_id)

        if ip_action == "REMOVE":

            if fix:

                r = remove_ip(ip_id)

                data += ("DONE" if r else "FAILED",)

        log.append(data)

    return log

def perform_actions(actions, fix=False):

    ACTION_HANDLING = [

        ("floating_ip", perform_floating_ip_actions),

        ("vm", perform_vm_actions),

        ]

    logs = []

    for resource_type, handler in ACTION_HANDLING:

        t_actions = actions.get(resource_type, {})

        log = handler(t_actions, fix=fix)

        logs += log

    return logs

# It is important to check resources in this order, especially

# floating_ip after vm resources.

RESOURCE_HANDLING = [

    ("cyclades.cpu", handle_stop_active, "vm"),

    ("cyclades.ram", handle_stop_active, "vm"),

    ("cyclades.total_cpu", handle_destroy, "vm"),

    ("cyclades.total_ram", handle_destroy, "vm"),

    ("cyclades.disk", handle_destroy, "vm"),

    ("cyclades.vm", handle_destroy, "vm"),

    ("cyclades.floating_ip", handle_floating_ip, "floating_ip"),

    ]