Synnefo

#!/usr/bin/env python

# -*- coding: utf-8 -*-

#

# Copyright 2011 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.

#

"""Ganeti notification daemon with AMQP support

A daemon to monitor the Ganeti job queue and publish job progress

and Ganeti VM state notifications to the ganeti exchange

"""

import sys

import os

path = os.path.normpath(os.path.join(os.getcwd(), '..'))

sys.path.append(path)

# Since Ganeti 2.7, debian package ships the majority of the python code in

# a private module under '/usr/share/ganeti'. Add this directory to path

# in order to be able to import ganeti. Also, add it to the start of path

# to allow conflicts with Ganeti RAPI client.

sys.path.insert(0, "/usr/share/ganeti")

import json

import logging

import pyinotify

import daemon

import daemon.pidlockfile

import daemon.runner

from lockfile import LockTimeout

from signal import signal, SIGINT, SIGTERM

import setproctitle

from ganeti import utils, jqueue, constants, serializer, pathutils, cli

from ganeti import errors as ganeti_errors

from ganeti.ssconf import SimpleStore

from synnefo import settings

from synnefo.lib.amqp import AMQPClient

def get_time_from_status(op, job):

    """Generate a unique message identifier for a ganeti job.

    The identifier is based on the timestamp of the job. Since a ganeti

    job passes from multiple states, we need to pick the timestamp that

    corresponds to each state.

    """

    status = op.status

    if status == constants.JOB_STATUS_QUEUED:

        time = job.received_timestamp

    try:  # Compatibility with Ganeti version

        if status == constants.JOB_STATUS_WAITLOCK:

            time = op.start_timestamp

    except AttributeError:

        if status == constants.JOB_STATUS_WAITING:

            time = op.start_timestamp

    if status == constants.JOB_STATUS_CANCELING:

        time = op.start_timestamp

    if status == constants.JOB_STATUS_RUNNING:

        time = op.exec_timestamp

    if status in constants.JOBS_FINALIZED:

        time = op.end_timestamp

    return time and time or job.end_timestamp

    raise InvalidBackendStatus(status, job)

def get_instance_nics(instance, logger):

    """Query Ganeti to a get the instance's NICs.

    Get instance's NICs from Ganeti configuration data. If running on master,

    query Ganeti via Ganeti CLI client. Otherwise, get the nics from Ganeti

    configuration file.

    @type instance: string

    @param instance: the name of the instance

    @rtype: List of dicts

    @return: Dictionary containing the instance's NICs. Each dictionary

             contains the following keys: 'network', 'ip', 'mac', 'mode',

             'link' and 'firewall'

    """

    try:

        client = cli.GetClient()

        fields = ["nic.networks.names", "nic.ips", "nic.macs", "nic.modes",

                  "nic.links", "tags"]

        info = client.QueryInstances([instance], fields, use_locking=False)

        networks, ips, macs, modes, links, tags = info[0]

        nic_keys = ["network", "ip", "mac", "mode", "link"]

        nics = zip(networks, ips, macs, modes, links)

        nics = map(lambda x: dict(zip(nic_keys, x)), nics)

    except ganeti_errors.OpPrereqError:

        # Not running on master! Load the conf file

        raw_data = utils.ReadFile(constants.CLUSTER_CONF_FILE)

        config = serializer.LoadJson(raw_data)

        i = config["instances"][instance]

        nics = []

        for nic in i["nics"]:

            params = nic.pop("nicparams")

            nic["mode"] = params["mode"]

            nic["link"] = params["link"]

            nics.append(nic)

        tags = i.get("tags", [])

    # Get firewall from instance Tags

    # Tags are of the form synnefo:network:N:firewall_mode

    for tag in tags:

        t = tag.split(":")

        if t[0:2] == ["synnefo", "network"]:

            if len(t) != 4:

                logger.error("Malformed synefo tag %s", tag)

                continue

            try:

                index = int(t[2])

                nics[index]['firewall'] = t[3]

            except ValueError:

                logger.error("Malformed synnefo tag %s", tag)

            except IndexError:

                logger.error("Found tag %s for non-existent NIC %d",

                             tag, index)

    return nics

class InvalidBackendStatus(Exception):

    def __init__(self, status, job):

        self.status = status

        self.job = job

    def __str__(self):

        return repr("Invalid backend status: %s in job %s"

                    % (self.status, self.job))

def prefix_from_name(name):

    return name.split('-')[0]

def get_field(from_, field):

    try:

        return getattr(from_, field)

    except AttributeError:

        None

class JobFileHandler(pyinotify.ProcessEvent):

    def __init__(self, logger, cluster_name):

        pyinotify.ProcessEvent.__init__(self)

        self.logger = logger

        self.cluster_name = cluster_name

        # Set max_retries to 0 for unlimited retries.

        self.client = AMQPClient(hosts=settings.AMQP_HOSTS, confirm_buffer=25,

                                 max_retries=0, logger=logger)

        handler_logger.info("Attempting to connect to RabbitMQ hosts")

        self.client.connect()

        handler_logger.info("Connected succesfully")

        self.client.exchange_declare(settings.EXCHANGE_GANETI, type='topic')

        self.op_handlers = {"INSTANCE": self.process_instance_op,

                            "NETWORK": self.process_network_op}

                            # "GROUP": self.process_group_op}

    def process_IN_CLOSE_WRITE(self, event):

        self.process_IN_MOVED_TO(event)

    def process_IN_MOVED_TO(self, event):

        jobfile = os.path.join(event.path, event.name)

        if not event.name.startswith("job-"):

            self.logger.debug("Not a job file: %s" % event.path)

            return

        try:

            data = utils.ReadFile(jobfile)

        except IOError:

            return

        data = serializer.LoadJson(data)

        job = jqueue._QueuedJob.Restore(None, data, False, False)

        job_id = int(job.id)

        for op in job.ops:

            op_id = op.input.OP_ID

            msg = None

            try:

                handler_fn = self.op_handlers[op_id.split('_')[1]]

                msg, routekey = handler_fn(op, job_id)

            except KeyError:

                pass

            if not msg:

                self.logger.debug("Ignoring job: %s: %s", job_id, op_id)

                continue

            # Generate a unique message identifier

            event_time = get_time_from_status(op, job)

            # Get the last line of the op log as message

            try:

                logmsg = op.log[-1][-1]

            except IndexError:

                logmsg = None

            # Add shared attributes for all operations

            msg.update({"event_time": event_time,

                        "operation": op_id,

                        "status": op.status,

                        "cluster": self.cluster_name,

                        "logmsg": logmsg,

                        "result": op.result,

                        "jobId": job_id})

            if op_id in ["OP_INSTANCE_CREATE", "OP_INSTANCE_SET_PARAMS",

                         "OP_INSTANCE_STARTUP"]:

                if op.status == "success":

                    nics = get_instance_nics(msg["instance"], self.logger)

                    msg["nics"] = nics

            if op_id == "OP_INSTANCE_CREATE" and op.status == "error":

                # In case an instance creation fails send the job input

                # so that the job can be retried if needed.

                msg["job_fields"] = op.Serialize()["input"]

            msg = json.dumps(msg)

            self.logger.debug("Delivering msg: %s (key=%s)", msg, routekey)

            # Send the message to RabbitMQ

            self.client.basic_publish(settings.EXCHANGE_GANETI,

                                      routekey,

                                      msg)

    def process_instance_op(self, op, job_id):

        """ Process OP_INSTANCE_* opcodes.

        """

        input = op.input

        op_id = input.OP_ID

        instances = None

        instances = get_field(input, 'instance_name')

        if not instances:

            instances = get_field(input, 'instances')

            if not instances or len(instances) > 1:

                # Do not publish messages for jobs with no or multiple

                # instances.  Currently snf-dispatcher can not normally handle

                # these messages

                return None, None

            else:

                instances = instances[0]

        self.logger.debug("Job: %d: %s(%s) %s", job_id, op_id,

                          instances, op.status)

        msg = {"type": "ganeti-op-status",

               "instance": instances,

               "operation": op_id}

        routekey = "ganeti.%s.event.op" % prefix_from_name(instances)

        return msg, routekey

    def process_network_op(self, op, job_id):

        """ Process OP_NETWORK_* opcodes.

        """

        input = op.input

        op_id = input.OP_ID

        network_name = get_field(input, 'network_name')

        if not network_name:

            return None, None

        self.logger.debug("Job: %d: %s(%s) %s", job_id, op_id,

                          network_name, op.status)

        msg = {'operation':    op_id,

               'type':         "ganeti-network-status",

               'network':      network_name,

               'subnet':       get_field(input, 'network'),

               # 'network_mode': get_field(input, 'network_mode'),

               # 'network_link': get_field(input, 'network_link'),

               'gateway':      get_field(input, 'gateway'),

               'group_name':   get_field(input, 'group_name')}

        if op_id == "OP_NETWORK_SET_PARAMS":

            msg["add_reserved_ips"] = get_field(input, "add_reserved_ips")

            msg["remove_reserved_ips"] = get_field(input,

                                                   "remove_reserved_ips")

        routekey = "ganeti.%s.event.network" % prefix_from_name(network_name)

        return msg, routekey

    # def process_group_op(self, op, job_id):

    #     """ Process OP_GROUP_* opcodes.

    #     """

    #     return None, None

def find_cluster_name():

    global handler_logger

    try:

        ss = SimpleStore()

        name = ss.GetClusterName()

    except Exception as e:

        handler_logger.error('Can not get the name of the Cluster: %s' % e)

        raise e

    return name

handler_logger = None

def fatal_signal_handler(signum, frame):

    global handler_logger

    handler_logger.info("Caught fatal signal %d, will raise SystemExit",

                        signum)

    raise SystemExit

def parse_arguments(args):

    from optparse import OptionParser

    parser = OptionParser()

    parser.add_option("-d", "--debug", action="store_true", dest="debug",

                      help="Enable debugging information")

    parser.add_option("-l", "--log", dest="log_file",

                      default="/var/log/snf-ganeti-eventd.log",

                      metavar="FILE",

                      help="Write log to FILE instead of %s" %

                           "/var/log/snf-ganeti-eventd.log")

    parser.add_option('--pid-file', dest="pid_file",

                      default="/var/run/snf-ganeti-eventd.pid",

                      metavar='PIDFILE',

                      help="Save PID to file (default: %s)" %

                           "/var/run/snf-ganeti-eventd.pid")

    return parser.parse_args(args)

def main():

    global handler_logger

    (opts, args) = parse_arguments(sys.argv[1:])

    # Initialize logger

    lvl = logging.DEBUG if opts.debug else logging.INFO

    logger = logging.getLogger("ganeti.eventd")

    logger.setLevel(lvl)

    formatter = logging.Formatter(

        "%(asctime)s %(module)s[%(process)d] %(levelname)s: %(message)s",

        "%Y-%m-%d %H:%M:%S")

    handler = logging.FileHandler(opts.log_file)

    handler.setFormatter(formatter)

    logger.addHandler(handler)

    handler_logger = logger

    # Rename this process so 'ps' output looks like this is a native

    # executable.  Can not seperate command-line arguments from actual name of

    # the executable by NUL bytes, so only show the name of the executable

    # instead.  setproctitle.setproctitle("\x00".join(sys.argv))

    setproctitle.setproctitle(sys.argv[0])

    # Create pidfile

    pidf = daemon.pidlockfile.TimeoutPIDLockFile(opts.pid_file, 10)

    # Remove any stale PID files, left behind by previous invocations

    if daemon.runner.is_pidfile_stale(pidf):

        logger.warning("Removing stale PID lock file %s", pidf.path)

        pidf.break_lock()

    # Become a daemon:

    # Redirect stdout and stderr to handler.stream to catch

    # early errors in the daemonization process [e.g., pidfile creation]

    # which will otherwise go to /dev/null.

    daemon_context = daemon.DaemonContext(

        pidfile=pidf,

        umask=022,

        stdout=handler.stream,

        stderr=handler.stream,

        files_preserve=[handler.stream])

    try:

        daemon_context.open()

    except (daemon.pidlockfile.AlreadyLocked, LockTimeout):

        logger.critical("Failed to lock pidfile %s, another instance running?",

                        pidf.path)

        sys.exit(1)

    logger.info("Became a daemon")

    # Catch signals to ensure graceful shutdown

    signal(SIGINT, fatal_signal_handler)

    signal(SIGTERM, fatal_signal_handler)

    # Monitor the Ganeti job queue, create and push notifications

    wm = pyinotify.WatchManager()

    mask = (pyinotify.EventsCodes.ALL_FLAGS["IN_MOVED_TO"] |

            pyinotify.EventsCodes.ALL_FLAGS["IN_CLOSE_WRITE"])

    cluster_name = find_cluster_name()

    handler = JobFileHandler(logger, cluster_name)

    notifier = pyinotify.Notifier(wm, handler)

    try:

        # Fail if adding the inotify() watch fails for any reason

        res = wm.add_watch(pathutils.QUEUE_DIR, mask)

        if res[pathutils.QUEUE_DIR] < 0:

            raise Exception("pyinotify add_watch returned negative descriptor")

        logger.info("Now watching %s of %s" % (pathutils.QUEUE_DIR,

                    cluster_name))

        while True:    # loop forever

            # process the queue of events as explained above

            notifier.process_events()

            if notifier.check_events():

                # read notified events and enqeue them

                notifier.read_events()

    except SystemExit:

        logger.info("SystemExit")

    except:

        logger.exception("Caught exception, terminating")

    finally:

        # destroy the inotify's instance on this interrupt (stop monitoring)

        notifier.stop()

        raise

if __name__ == "__main__":

    sys.exit(main())

# vim: set sta sts=4 shiftwidth=4 sw=4 et ai :