Exemple #1
0
    def delete_task(self, task_name):
        """ Deletes the named Task in this Job. """

        logger.debug('Deleting task {0}'.format(task_name))
        if not self.state.allow_change_graph:
            raise DagobahError(
                "job's graph is immutable in its current state: %s" %
                self.state.status)

        if task_name not in self.tasks:
            raise DagobahError('task %s does not exist' % task_name)

        self.tasks.pop(task_name)
        self.delete_node(task_name)
        self.commit()
Exemple #2
0
    def start(self):
        """ Begins the job by kicking off all tasks with no dependencies. """

        logger.info('Job {0} starting job run'.format(self.name))
        if not self.state.allow_start:
            raise DagobahError('job cannot be started in its current state; ' +
                               'it is probably already running')

        self.initialize_snapshot()

        # don't increment if the job was run manually
        if self.cron_iter and datetime.utcnow() > self.next_run:
            self.next_run = self.cron_iter.get_next(datetime)

        self.run_log = {
            'job_id': self.job_id,
            'name': self.name,
            'parent_id': self.parent.dagobah_id,
            'log_id': self.backend.get_new_log_id(),
            'start_time': datetime.utcnow(),
            'tasks': {}
        }
        self._set_status('running')

        logger.debug('Job {0} resetting all tasks prior to start'.format(
            self.name))
        for task in self.tasks.values():
            task.reset()

        logger.debug('Job {0} seeding run logs'.format(self.name))
        for task_name in self.ind_nodes(self.snapshot):
            self._put_task_in_run_log(task_name)
            self.tasks[task_name].start()

        self._commit_run_log()
Exemple #3
0
    def schedule(self, cron_schedule, base_datetime=None):
        """ Schedules the job to run periodically using Cron syntax. """

        logger.debug('Scheduling job {0} with cron schedule {1}'.format(
            self.name, cron_schedule))
        if not self.state.allow_change_schedule:
            raise DagobahError(
                "job's schedule cannot be changed in state: %s" %
                self.state.status)

        if cron_schedule is None:
            self.cron_schedule = None
            self.cron_iter = None
            self.next_run = None

        else:
            if base_datetime is None:
                base_datetime = datetime.utcnow()
            self.cron_schedule = cron_schedule
            self.cron_iter = croniter(cron_schedule, base_datetime)
            self.next_run = self.cron_iter.get_next(datetime)

        logger.debug('Determined job {0} next run of {1}'.format(
            self.name, self.next_run))
        self.commit()
Exemple #4
0
    def _tail_temp_file(self, temp_file, num_lines, seek_offset=10000):
        """ Returns a list of the last num_lines lines from a temp file.

        This works by first moving seek_offset chars back from the end of
        the file, then attempting to tail the file from there. It is
        possible that fewer than num_lines will be returned, even if the
        file has more total lines than num_lines.
        """

        if not isinstance(num_lines, int):
            raise DagobahError('num_lines must be an integer')

        temp_file.seek(0, os.SEEK_END)
        size = temp_file.tell()
        temp_file.seek(-1 * min(size, seek_offset), os.SEEK_END)

        result = []
        while True:
            this_line = temp_file.readline()
            if this_line == '':
                break
            result.append(this_line.strip())
            if len(result) > num_lines:
                result.pop(0)
        return result
Exemple #5
0
 def from_backend(self, dagobah_id):
     """ Reconstruct this Dagobah instance from the backend. """
     logger.debug('Reconstructing Dagobah instance from backend with ID {0}'.format(dagobah_id))
     rec = self.backend.get_dagobah_json(dagobah_id)
     if not rec:
         raise DagobahError('dagobah with id %s does not exist '
                            'in backend' % dagobah_id)
     self._construct_from_json(rec)
Exemple #6
0
    def update_job_notes(self, notes):
        logger.debug('Job {0} updating notes'.format(self.name))
        if not self.state.allow_edit_job:
            raise DagobahError('job cannot be edited in its current state')

        setattr(self, 'notes', notes)

        self.parent.commit(cascade=True)
Exemple #7
0
 def delete_job(self, job_name):
     """ Delete a job by name, or error out if no such job exists. """
     logger.debug('Deleting job {0}'.format(job_name))
     for idx, job in enumerate(self.jobs):
         if job.name == job_name:
             self.backend.delete_job(job.job_id)
             del self.jobs[idx]
             self.commit()
             return
     raise DagobahError('no job with name %s exists' % job_name)
Exemple #8
0
 def kill(self):
     """ Send SIGKILL to the task's process. """
     logger.info('Sending SIGKILL to task {0}'.format(self.name))
     if hasattr(self, 'remote_client') and self.remote_client is not None:
         self.kill_sent = True
         self.remote_client.close()
         return
     if not self.process:
         raise DagobahError('task does not have a running process')
     self.kill_sent = True
     self.process.kill()
Exemple #9
0
    def add_task_to_job(self, job_or_job_name, task_command, task_name=None,
                        **kwargs):
        """ Add a task to a job owned by the Dagobah instance. """

        if isinstance(job_or_job_name, Job):
            job = job_or_job_name
        else:
            job = self.get_job(job_or_job_name)

        if not job:
            raise DagobahError('job %s does not exist' % job_or_job_name)

        logger.debug('Adding task with command {0} to job {1}'.format(task_command, job.name))

        if not job.state.allow_change_graph:
            raise DagobahError("job's graph is immutable in its current " +
                               "state: %s"
                               % job.state.status)

        job.add_task(task_command, task_name, **kwargs)
        job.commit()
Exemple #10
0
 def _head_temp_file(self, temp_file, num_lines):
     """ Returns a list of the first num_lines lines from a temp file. """
     if not isinstance(num_lines, int):
         raise DagobahError('num_lines must be an integer')
     temp_file.seek(0)
     result, curr_line = [], 0
     for line in temp_file:
         curr_line += 1
         result.append(line.strip())
         if curr_line >= num_lines:
             break
     return result
Exemple #11
0
    def add_dependency(self, from_task_name, to_task_name):
        """ Add a dependency between two tasks. """

        logger.debug('Adding dependency from {0} to {1}'.format(
            from_task_name, to_task_name))
        if not self.state.allow_change_graph:
            raise DagobahError(
                "job's graph is immutable in its current state: %s" %
                self.state.status)

        self.add_edge(from_task_name, to_task_name)
        self.commit()
Exemple #12
0
    def edit_task(self, task_name, **kwargs):
        """ Change the name of a Task owned by this Job.

        This will affect the historical data available for this
        Task, e.g. past run logs will no longer be accessible.
        """

        logger.debug('Job {0} editing task {1}'.format(self.name, task_name))
        if not self.state.allow_edit_task:
            raise DagobahError("tasks cannot be edited in this job's " +
                               "current state")

        if task_name not in self.tasks:
            raise DagobahError('task %s not found' % task_name)

        if 'name' in kwargs and isinstance(kwargs['name'], str):
            if kwargs['name'] in self.tasks:
                raise DagobahError('task name %s is unavailable' %
                                   kwargs['name'])

        task = self.tasks[task_name]
        for key in ['name', 'command']:
            if key in kwargs and isinstance(kwargs[key], str):
                setattr(task, key, kwargs[key])

        if 'soft_timeout' in kwargs:
            task.set_soft_timeout(kwargs['soft_timeout'])

        if 'hard_timeout' in kwargs:
            task.set_hard_timeout(kwargs['hard_timeout'])

        if 'hostname' in kwargs:
            task.set_hostname(kwargs['hostname'])

        if 'name' in kwargs and isinstance(kwargs['name'], str):
            self.rename_edges(task_name, kwargs['name'])
            self.tasks[kwargs['name']] = task
            del self.tasks[task_name]

        self.parent.commit(cascade=True)
Exemple #13
0
    def edit(self, **kwargs):
        """ Change this Job's name.

        This will affect the historical data available for this
        Job, e.g. past run logs will no longer be accessible.
        """

        logger.debug('Job {0} changing name to {1}'.format(
            self.name, kwargs.get('name')))
        if not self.state.allow_edit_job:
            raise DagobahError('job cannot be edited in its current state')

        if 'name' in kwargs and isinstance(kwargs['name'], str):
            if not self.parent._name_is_available(kwargs['name']):
                raise DagobahError('new job name %s is not available' %
                                   kwargs['name'])

        for key in ['name']:
            if key in kwargs and isinstance(kwargs[key], str):
                setattr(self, key, kwargs[key])

        self.parent.commit(cascade=True)
Exemple #14
0
    def add_job(self, job_name, job_id=None):
        """ Create a new, empty Job. """
        logger.debug('Creating a new job named {0}'.format(job_name))
        if not self._name_is_available(job_name):
            raise DagobahError('name %s is not available' % job_name)

        if not job_id:
            job_id = self.backend.get_new_job_id()
            self.created_jobs += 1

        self.jobs.append(Job(self, self.backend, job_id, job_name))

        job = self.get_job(job_name)
        job.commit()
Exemple #15
0
    def initialize_snapshot(self):
        """ Copy the DAG and validate """
        logger.debug('Initializing DAG snapshot for job {0}'.format(self.name))
        if self.snapshot is not None:
            logger.warn("Attempting to initialize DAG snapshot without " +
                        "first destroying old snapshot.")

        snapshot_to_validate = deepcopy(self.graph)

        is_valid, reason = self.validate(snapshot_to_validate)
        if not is_valid:
            raise DagobahError(reason)

        self.snapshot = snapshot_to_validate
Exemple #16
0
    def add_task(self, command, name=None, **kwargs):
        """ Adds a new Task to the graph with no edges. """

        logger.debug('Adding task with command {0} to job {1}'.format(
            command, self.name))
        if not self.state.allow_change_graph:
            raise DagobahError(
                "job's graph is immutable in its current state: %s" %
                self.state.status)

        if name is None:
            name = command
        new_task = Task(self, command, name, **kwargs)
        self.tasks[name] = new_task
        self.add_node(name)
        self.commit()
Exemple #17
0
    def retry(self):
        """ Restarts failed tasks of a job. """

        logger.info('Job {0} retrying all failed tasks'.format(self.name))
        self.initialize_snapshot()

        failed_task_names = []
        for task_name, log in self.run_log['tasks'].items():
            if log.get('success', True) == False:
                failed_task_names.append(task_name)

        if len(failed_task_names) == 0:
            raise DagobahError('no failed tasks to retry')

        self._set_status('running')
        self.run_log['last_retry_time'] = datetime.utcnow()

        logger.debug('Job {0} seeding run logs'.format(self.name))
        for task_name in failed_task_names:
            self._put_task_in_run_log(task_name)
            self.tasks[task_name].start()

        self._commit_run_log()
Exemple #18
0
 def _set_status(self, status):
     """ Enforces enum-like behavior on the status field. """
     try:
         self.state.set_status(status)
     except:
         raise DagobahError('could not set status %s' % status)
Exemple #19
0
 def _map_string_to_file(self, stream):
     if stream not in ['stdout', 'stderr']:
         raise DagobahError('stream must be stdout or stderr')
     return self.stdout_file if stream == 'stdout' else self.stderr_file