source: trunk/packages/invirt-web/code/main.py @ 3005

#!/usr/bin/python
"""Main CGI script for web interface"""

import base64
import cPickle
import cgi
import datetime
import hmac
import os
import random
import sha
import sys
import time
import urllib
import socket
import cherrypy
from cherrypy import _cperror
from StringIO import StringIO

def printError():
    """Revert stderr to stdout, and print the contents of stderr"""
    if isinstance(sys.stderr, StringIO):
        print revertStandardError()

if __name__ == '__main__':
    import atexit
    atexit.register(printError)

import validation
import cache_acls
from webcommon import State
import controls
from getafsgroups import getAfsGroupMembers
from invirt import database
from invirt.database import Machine, CDROM, session, connect, MachineAccess, Type, Autoinstall
from invirt.config import structs as config
from invirt.common import InvalidInput, CodeError

from view import View, revertStandardError


static_dir = os.path.join(os.path.dirname(__file__), 'static')
InvirtStatic = cherrypy.tools.staticdir.handler(
    root=static_dir,
    dir=static_dir,
    section='/static')

class InvirtUnauthWeb(View):
    static = InvirtStatic

    @cherrypy.expose
    @cherrypy.tools.mako(filename="/unauth.mako")
    def index(self):
        return dict(simple=True)

class InvirtWeb(View):
    def __init__(self):
        super(self.__class__,self).__init__()
        connect()
        self._cp_config['tools.require_login.on'] = True
        self._cp_config['tools.catch_stderr.on'] = True
        self._cp_config['tools.mako.imports'] = ['from invirt.config import structs as config',
                                                 'from invirt import database']
        self._cp_config['request.error_response'] = self.handle_error

    static = InvirtStatic

    @cherrypy.expose
    @cherrypy.tools.mako(filename="/invalid.mako")
    def invalidInput(self):
        """Print an error page when an InvalidInput exception occurs"""
        err = cherrypy.request.prev.params["err"]
        emsg = cherrypy.request.prev.params["emsg"]
        d = dict(err_field=err.err_field,
                 err_value=str(err.err_value), stderr=emsg,
                 errorMessage=str(err))
        return d

    @cherrypy.expose
    @cherrypy.tools.mako(filename="/error.mako")
    def error(self):
        """Print an error page when an exception occurs"""
        op = cherrypy.request.prev.path_info
        username = cherrypy.request.login
        err = cherrypy.request.prev.params["err"]
        emsg = cherrypy.request.prev.params["emsg"]
        traceback = cherrypy.request.prev.params["traceback"]
        d = dict(op=op, user=username, fields=cherrypy.request.prev.params,
                 errorMessage=str(err), stderr=emsg, traceback=traceback)
        error_raw = cherrypy.request.lookup.get_template("/error_raw.mako")
        details = error_raw.render(**d)
        exclude = config.web.errormail_exclude
        if username not in exclude and '*' not in exclude:
            send_error_mail('xvm error on %s for %s: %s' % (op, cherrypy.request.login, err),
                            details)
        d['details'] = details
        return d

    def __getattr__(self, name):
        # At the point __getattr__ is called, tools haven't been run. Make sure the user is logged in.
        cherrypy.tools.remote_user_login.callable()

        if name in ("admin", "overlord"):
            if not cherrypy.request.login in getAfsGroupMembers(config.adminacl, config.authz.afs.cells[0].cell):
                raise InvalidInput('username', cherrypy.request.login,
                                   'Not in admin group %s.' % config.adminacl)
            cherrypy.request.state = State(cherrypy.request.login, isadmin=True)
            return self
        else:
            return super(InvirtWeb, self).__getattr__(name)

    def handle_error(self):
        err = sys.exc_info()[1]
        if isinstance(err, InvalidInput):
            cherrypy.request.params['err'] = err
            cherrypy.request.params['emsg'] = revertStandardError()
            raise cherrypy.InternalRedirect('/invalidInput')
        if not cherrypy.request.prev or 'err' not in cherrypy.request.prev.params:
            cherrypy.request.params['err'] = err
            cherrypy.request.params['emsg'] = revertStandardError()
            cherrypy.request.params['traceback'] = _cperror.format_exc()
            raise cherrypy.InternalRedirect('/error')
        # fall back to cherrypy default error page
        cherrypy.HTTPError(500).set_response()

    @cherrypy.expose
    @cherrypy.tools.mako(filename="/list.mako")
    def list(self, result=None):
        """Handler for list requests."""
        d = getListDict(cherrypy.request.login, cherrypy.request.state)
        if result is not None:
            d['result'] = result
        return d
    index=list

    @cherrypy.expose
    @cherrypy.tools.mako(filename="/help.mako")
    def help(self, subject=None, simple=False):
        """Handler for help messages."""

        help_mapping = {
            'Autoinstalls': """
The autoinstaller builds a minimal Debian or Ubuntu system to run as a
ParaVM.  You can access the resulting system by logging into the <a
href="help?simple=true&subject=ParaVM+Console">serial console server</a>
with your Kerberos tickets; there is no root password so sshd will
refuse login.</p>

<p>Under the covers, the autoinstaller uses our own patched version of
xen-create-image, which is a tool based on debootstrap.  If you log
into the serial console while the install is running, you can watch
it.
""",
            'ParaVM Console': """
ParaVM machines do not support local console access over VNC.  To
access the serial console of these machines, you can SSH with Kerberos
to %s, using the name of the machine as your
username.""" % config.console.hostname,
            'HVM/ParaVM': """
HVM machines use the virtualization features of the processor, while
ParaVM machines rely on a modified kernel to communicate directly with
the hypervisor.  HVMs support boot CDs of any operating system, and
the VNC console applet.  The three-minute autoinstaller produces
ParaVMs.  ParaVMs typically are more efficient, and always support the
<a href="help?subject=ParaVM+Console">console server</a>.</p>

<p>More details are <a
href="https://xvm.scripts.mit.edu/wiki/Paravirtualization">on the
wiki</a>, including steps to prepare an HVM guest to boot as a ParaVM
(which you can skip by using the autoinstaller to begin with.)</p>

<p>We recommend using a ParaVM when possible and an HVM when necessary.
""",
            'CPU Weight': """
Don't ask us!  We're as mystified as you are.""",
            'Owner': """
The owner field is used to determine <a
href="help?subject=Quotas">quotas</a>.  It must be the name of a
locker that you are an AFS administrator of.  In particular, you or an
AFS group you are a member of must have AFS rlidwka bits on the
locker.  You can check who administers the LOCKER locker using the
commands 'attach LOCKER; fs la /mit/LOCKER' on Athena.)  See also <a
href="help?subject=Administrator">administrator</a>.""",
            'Administrator': """
The administrator field determines who can access the console and
power on and off the machine.  This can be either a user or a moira
group.""",
            'Quotas': """
Quotas are determined on a per-locker basis.  Each locker may have a
maximum of 512 mebibytes of active ram, 50 gibibytes of disk, and 4
active machines.""",
            'Console': """
<strong>Framebuffer:</strong> At a Linux boot prompt in your VM, try
setting <tt>fb=false</tt> to disable the framebuffer.  If you don't,
your machine will run just fine, but the applet's display of the
console will suffer artifacts.
""",
            'Windows': """
<strong>Windows Vista:</strong> The Vista image is licensed for all MIT students and will automatically activate off the network; see <a href="/static/msca-email.txt">the licensing confirmation e-mail</a> for details. The installer requires 512 MiB RAM and at least 7.5 GiB disk space (15 GiB or more recommended).<br>
<strong>Windows XP:</strong> This is the volume license CD image. You will need your own volume license key to complete the install. We do not have these available for the general MIT community; ask your department if they have one, or visit <a href="http://msca.mit.edu/">http://msca.mit.edu/</a> if you are staff/faculty to request one.
"""
            }

        if not subject:
            subject = sorted(help_mapping.keys())
        if not isinstance(subject, list):
            subject = [subject]

        return dict(simple=simple,
                    subjects=subject,
                    mapping=help_mapping)
    help._cp_config['tools.require_login.on'] = False

    def parseCreate(self, fields):
        kws = dict([(kw, fields[kw]) for kw in
         'name description owner memory disksize vmtype cdrom autoinstall'.split()
                    if fields[kw]])
        validate = validation.Validate(cherrypy.request.login,
                                       cherrypy.request.state,
                                       strict=True, **kws)
        return dict(contact=cherrypy.request.login, name=validate.name,
                    description=validate.description, memory=validate.memory,
                    disksize=validate.disksize, owner=validate.owner,
                    machine_type=getattr(validate, 'vmtype', Defaults.type),
                    cdrom=getattr(validate, 'cdrom', None),
                    autoinstall=getattr(validate, 'autoinstall', None))

    @cherrypy.expose
    @cherrypy.tools.mako(filename="/list.mako")
    @cherrypy.tools.require_POST()
    def create(self, **fields):
        """Handler for create requests."""
        try:
            parsed_fields = self.parseCreate(fields)
            machine = controls.createVm(cherrypy.request.login,
                                        cherrypy.request.state, **parsed_fields)
        except InvalidInput, err:
            pass
        else:
            err = None
        cherrypy.request.state.clear() #Changed global state
        d = getListDict(cherrypy.request.login, cherrypy.request.state)
        d['err'] = err
        if err:
            for field, value in fields.items():
                setattr(d['defaults'], field, value)
        else:
            d['new_machine'] = parsed_fields['name']
        return d

    @cherrypy.expose
    @cherrypy.tools.mako(filename="/helloworld.mako")
    def helloworld(self, **kwargs):
        return {'request': cherrypy.request, 'kwargs': kwargs}
    helloworld._cp_config['tools.require_login.on'] = False

    @cherrypy.expose
    def errortest(self):
        """Throw an error, to test the error-tracing mechanisms."""
        print >>sys.stderr, "look ma, it's a stderr"
        raise RuntimeError("test of the emergency broadcast system")

    class MachineView(View):
        def __getattr__(self, name):
            """Synthesize attributes to allow RESTful URLs like
            /machine/13/info. This is hairy. CherryPy 3.2 adds a
            method called _cp_dispatch that allows you to explicitly
            handle URLs that can't be mapped, and it allows you to
            rewrite the path components and continue processing.

            This function gets the next path component being resolved
            as a string. _cp_dispatch will get an array of strings
            representing any subsequent path components as well."""

            try:
                cherrypy.request.params['machine_id'] = int(name)
                return self
            except ValueError:
                return None

        @cherrypy.expose
        @cherrypy.tools.mako(filename="/info.mako")
        def info(self, machine_id):
            """Handler for info on a single VM."""
            machine = validation.Validate(cherrypy.request.login,
                                          cherrypy.request.state,
                                          machine_id=machine_id).machine
            d = infoDict(cherrypy.request.login, cherrypy.request.state, machine)
            return d
        index = info

        @cherrypy.expose
        @cherrypy.tools.mako(filename="/info.mako")
        @cherrypy.tools.require_POST()
        def modify(self, machine_id, **fields):
            """Handler for modifying attributes of a machine."""
            try:
                modify_dict = modifyDict(cherrypy.request.login,
                                         cherrypy.request.state,
                                         machine_id, fields)
            except InvalidInput, err:
                result = None
                machine = validation.Validate(cherrypy.request.login,
                                              cherrypy.request.state,
                                              machine_id=machine_id).machine
            else:
                machine = modify_dict['machine']
                result = 'Success!'
                err = None
            info_dict = infoDict(cherrypy.request.login,
                                 cherrypy.request.state, machine)
            info_dict['err'] = err
            if err:
                for field, value in fields.items():
                    setattr(info_dict['defaults'], field, value)
            info_dict['result'] = result
            return info_dict

        @cherrypy.expose
        @cherrypy.tools.mako(filename="/vnc.mako")
        def vnc(self, machine_id):
            """VNC applet page.

            Note that due to same-domain restrictions, the applet connects to
            the webserver, which needs to forward those requests to the xen
            server.  The Xen server runs another proxy that (1) authenticates
            and (2) finds the correct port for the VM.

            You might want iptables like:

            -t nat -A PREROUTING -s ! 18.181.0.60 -i eth1 -p tcp -m tcp \
            --dport 10003 -j DNAT --to-destination 18.181.0.60:10003
            -t nat -A POSTROUTING -d 18.181.0.60 -o eth1 -p tcp -m tcp \
            --dport 10003 -j SNAT --to-source 18.187.7.142
            -A FORWARD -d 18.181.0.60 -i eth1 -o eth1 -p tcp -m tcp \
            --dport 10003 -j ACCEPT

            Remember to enable iptables!
            echo 1 > /proc/sys/net/ipv4/ip_forward
            """
            machine = validation.Validate(cherrypy.request.login,
                                          cherrypy.request.state,
                                          machine_id=machine_id).machine
            token = controls.vnctoken(machine)
            host = controls.listHost(machine)
            if host:
                port = 10003 + [h.hostname for h in config.hosts].index(host)
            else:
                port = 5900 # dummy

            status = controls.statusInfo(machine)
            has_vnc = hasVnc(status)

            d = dict(on=status,
                     has_vnc=has_vnc,
                     machine=machine,
                     hostname=cherrypy.request.local.name,
                     port=port,
                     authtoken=token)
            return d

        @cherrypy.expose
        @cherrypy.tools.mako(filename="/command.mako")
        @cherrypy.tools.require_POST()
        def command(self, command_name, machine_id, **kwargs):
            """Handler for running commands like boot and delete on a VM."""
            back = kwargs.get('back')
            if command_name == 'delete':
                back = 'list'
            try:
                d = controls.commandResult(cherrypy.request.login,
                                           cherrypy.request.state,
                                           command_name, machine_id, kwargs)
            except InvalidInput, err:
                if not back:
                    raise
                print >> sys.stderr, err
                result = str(err)
            else:
                result = 'Success!'
                if not back:
                    return d
            if back == 'list':
                cherrypy.request.state.clear() #Changed global state
                raise cherrypy.InternalRedirect('/list?result=%s'
                                                % urllib.quote(result))
            elif back == 'info':
                raise cherrypy.HTTPRedirect(cherrypy.request.base
                                            + '/machine/%d/' % machine_id,
                                            status=303)
            else:
                raise InvalidInput('back', back, 'Not a known back page.')

    machine = MachineView()


class Defaults:
    """Class to store default values for fields."""
    memory = 256
    disk = 4.0
    cdrom = ''
    autoinstall = ''
    name = ''
    description = ''
    administrator = ''
    type = 'linux-hvm'

    def __init__(self, max_memory=None, max_disk=None, **kws):
        if max_memory is not None:
            self.memory = min(self.memory, max_memory)
        if max_disk is not None:
            self.disk = min(self.disk, max_disk)
        for key in kws:
            setattr(self, key, kws[key])

def hasVnc(status):
    """Does the machine with a given status list support VNC?"""
    if status is None:
        return False
    for l in status:
        if l[0] == 'device' and l[1][0] == 'vfb':
            d = dict(l[1][1:])
            return 'location' in d
    return False


def getListDict(username, state):
    """Gets the list of local variables used by list.tmpl."""
    machines = state.machines
    on = {}
    has_vnc = {}
    installing = {}
    xmlist = state.xmlist
    for m in machines:
        if m not in xmlist:
            has_vnc[m] = 'Off'
            m.uptime = None
        else:
            m.uptime = xmlist[m]['uptime']
            installing[m] = bool(xmlist[m].get('autoinstall'))
            if xmlist[m]['console']:
                has_vnc[m] = True
            elif m.type.hvm:
                has_vnc[m] = "WTF?"
            else:
                has_vnc[m] = "ParaVM"
    max_memory = validation.maxMemory(username, state)
    max_disk = validation.maxDisk(username)
    defaults = Defaults(max_memory=max_memory,
                        max_disk=max_disk,
                        owner=username)
    def sortkey(machine):
        return (machine.owner != username, machine.owner, machine.name)
    machines = sorted(machines, key=sortkey)
    d = dict(user=username,
             cant_add_vm=validation.cantAddVm(username, state),
             max_memory=max_memory,
             max_disk=max_disk,
             defaults=defaults,
             machines=machines,
             has_vnc=has_vnc,
             installing=installing)
    return d

def getHostname(nic):
    """Find the hostname associated with a NIC.

    XXX this should be merged with the similar logic in DNS and DHCP.
    """
    if nic.hostname:
        hostname = nic.hostname
    elif nic.machine:
        hostname = nic.machine.name
    else:
        return None
    if '.' in hostname:
        return hostname
    else:
        return hostname + '.' + config.dns.domains[0]

def getNicInfo(data_dict, machine):
    """Helper function for info, get data on nics for a machine.

    Modifies data_dict to include the relevant data, and returns a list
    of (key, name) pairs to display "name: data_dict[key]" to the user.
    """
    data_dict['num_nics'] = len(machine.nics)
    nic_fields_template = [('nic%s_hostname', 'NIC %s Hostname'),
                           ('nic%s_mac', 'NIC %s MAC Addr'),
                           ('nic%s_ip', 'NIC %s IP'),
                           ]
    nic_fields = []
    for i in range(len(machine.nics)):
        nic_fields.extend([(x % i, y % i) for x, y in nic_fields_template])
        data_dict['nic%s_hostname' % i] = getHostname(machine.nics[i])
        data_dict['nic%s_mac' % i] = machine.nics[i].mac_addr
        data_dict['nic%s_ip' % i] = machine.nics[i].ip
    if len(machine.nics) == 1:
        nic_fields = [(x, y.replace('NIC 0 ', '')) for x, y in nic_fields]
    return nic_fields

def getDiskInfo(data_dict, machine):
    """Helper function for info, get data on disks for a machine.

    Modifies data_dict to include the relevant data, and returns a list
    of (key, name) pairs to display "name: data_dict[key]" to the user.
    """
    data_dict['num_disks'] = len(machine.disks)
    disk_fields_template = [('%s_size', '%s size')]
    disk_fields = []
    for disk in machine.disks:
        name = disk.guest_device_name
        disk_fields.extend([(x % name, y % name) for x, y in
                            disk_fields_template])
        data_dict['%s_size' % name] = "%0.1f GiB" % (disk.size / 1024.)
    return disk_fields

def modifyDict(username, state, machine_id, fields):
    """Modify a machine as specified by CGI arguments.

    Return a dict containing the machine that was modified.
    """
    olddisk = {}
    session.begin()
    try:
        kws = dict((kw, fields[kw]) for kw in
         'owner admin contact name description memory vmtype disksize'.split()
                    if fields.get(kw))
        kws['machine_id'] = machine_id
        validate = validation.Validate(username, state, **kws)
        machine = validate.machine
        oldname = machine.name

        if hasattr(validate, 'memory'):
            machine.memory = validate.memory

        if hasattr(validate, 'vmtype'):
            machine.type = validate.vmtype

        if hasattr(validate, 'disksize'):
            disksize = validate.disksize
            disk = machine.disks[0]
            if disk.size != disksize:
                olddisk[disk.guest_device_name] = disksize
                disk.size = disksize
                session.save_or_update(disk)

        update_acl = False
        if hasattr(validate, 'owner') and validate.owner != machine.owner:
            machine.owner = validate.owner
            update_acl = True
        if hasattr(validate, 'name'):
            machine.name = validate.name
            for n in machine.nics:
                if n.hostname == oldname:
                    n.hostname = validate.name
        if hasattr(validate, 'description'):
            machine.description = validate.description
        if hasattr(validate, 'admin') and validate.admin != machine.administrator:
            machine.administrator = validate.admin
            update_acl = True
        if hasattr(validate, 'contact'):
            machine.contact = validate.contact

        session.save_or_update(machine)
        if update_acl:
            cache_acls.refreshMachine(machine)
        session.commit()
    except:
        session.rollback()
        raise
    for diskname in olddisk:
        controls.resizeDisk(oldname, diskname, str(olddisk[diskname]))
    if hasattr(validate, 'name'):
        controls.renameMachine(machine, oldname, validate.name)
    return dict(machine=machine)

def infoDict(username, state, machine):
    """Get the variables used by info.tmpl."""
    status = controls.statusInfo(machine)
    has_vnc = hasVnc(status)
    if status is None:
        main_status = dict(name=machine.name,
                           memory=str(machine.memory))
        uptime = None
        cputime = None
    else:
        main_status = dict(status[1:])
        main_status['host'] = controls.listHost(machine)
        start_time = float(main_status.get('start_time', 0))
        uptime = datetime.timedelta(seconds=int(time.time()-start_time))
        cpu_time_float = float(main_status.get('cpu_time', 0))
        cputime = datetime.timedelta(seconds=int(cpu_time_float))
    display_fields = [('name', 'Name'),
                      ('description', 'Description'),
                      ('owner', 'Owner'),
                      ('administrator', 'Administrator'),
                      ('contact', 'Contact'),
                      ('type', 'Type'),
                      'NIC_INFO',
                      ('uptime', 'uptime'),
                      ('cputime', 'CPU usage'),
                      ('host', 'Hosted on'),
                      ('memory', 'RAM'),
                      'DISK_INFO',
                      ('state', 'state (xen format)'),
                      ]
    fields = []
    machine_info = {}
    machine_info['name'] = machine.name
    machine_info['description'] = machine.description
    machine_info['type'] = machine.type.hvm and 'HVM' or 'ParaVM'
    machine_info['owner'] = machine.owner
    machine_info['administrator'] = machine.administrator
    machine_info['contact'] = machine.contact

    nic_fields = getNicInfo(machine_info, machine)
    nic_point = display_fields.index('NIC_INFO')
    display_fields = (display_fields[:nic_point] + nic_fields +
                      display_fields[nic_point+1:])

    disk_fields = getDiskInfo(machine_info, machine)
    disk_point = display_fields.index('DISK_INFO')
    display_fields = (display_fields[:disk_point] + disk_fields +
                      display_fields[disk_point+1:])

    main_status['memory'] += ' MiB'
    for field, disp in display_fields:
        if field in ('uptime', 'cputime') and locals()[field] is not None:
            fields.append((disp, locals()[field]))
        elif field in machine_info:
            fields.append((disp, machine_info[field]))
        elif field in main_status:
            fields.append((disp, main_status[field]))
        else:
            pass
            #fields.append((disp, None))

    max_mem = validation.maxMemory(machine.owner, state, machine, False)
    max_disk = validation.maxDisk(machine.owner, machine)
    defaults = Defaults()
    for name in 'machine_id name description administrator owner memory contact'.split():
        if getattr(machine, name):
            setattr(defaults, name, getattr(machine, name))
    defaults.type = machine.type.type_id
    defaults.disk = "%0.2f" % (machine.disks[0].size/1024.)
    d = dict(user=username,
             on=status is not None,
             machine=machine,
             defaults=defaults,
             has_vnc=has_vnc,
             uptime=str(uptime),
             ram=machine.memory,
             max_mem=max_mem,
             max_disk=max_disk,
             fields = fields)
    return d

def send_error_mail(subject, body):
    import subprocess

    to = config.web.errormail
    mail = """To: %s
From: root@%s
Subject: %s

%s
""" % (to, config.web.hostname, subject, body)
    p = subprocess.Popen(['/usr/sbin/sendmail', '-f', to, to],
                         stdin=subprocess.PIPE)
    p.stdin.write(mail)
    p.stdin.close()
    p.wait()

random.seed() #sigh