pegasus-wms 4.0.1+dfsg-8 / usr / bin / pegasus-analyzer

#!/usr/bin/env python

"""
Pegasus utility for pasing jobstate.log and reporting succesful and failed jobs

Usage: pegasus-analyzer [options]

"""

##
#  Copyright 2007-2012 University Of Southern California
#
#  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.
##

# Revision : $Revision: 2012 $

# Import Python modules
import os
import re
import sys
import time
import errno
import logging
import commands
import optparse
import traceback
import subprocess
import tempfile

# Initialize logging object
logger = logging.getLogger()
# Set default level to WARNING
logger.setLevel(logging.WARNING)

# Use pegasus-config to find our lib path
bin_dir = os.path.normpath(os.path.join(os.path.dirname(sys.argv[0])))
pegasus_config = os.path.join(bin_dir, "pegasus-config") + " --noeoln --python"
lib_dir = subprocess.Popen(pegasus_config, stdout=subprocess.PIPE, shell=True).communicate()[0]
pegasus_config = os.path.join(bin_dir, "pegasus-config") + " --noeoln --python-externals"
lib_ext_dir = subprocess.Popen(pegasus_config, stdout=subprocess.PIPE, shell=True).communicate()[0]

# Insert this directory in our search path
os.sys.path.insert(0, lib_ext_dir)
os.sys.path.insert(0, lib_dir)

# Import our modules
import Pegasus.common
from Pegasus.tools import utils
from Pegasus.tools import db_utils
from Pegasus.tools import kickstart_parser
from netlogger.analysis.workflow import stampede_statistics
from netlogger.analysis.schema.schema_check import SchemaVersionError

# --- regular expressions -------------------------------------------------------------

re_parse_property = re.compile(r'([^:= \t]+)\s*[:=]?\s*(.*)')
re_parse_script_pre = re.compile(r'^SCRIPT PRE (\S+) (.*)')
re_parse_condor_subs = re.compile(r'(\S+)="([^"]+)"')
re_collapse_condor_subs = re.compile(r'\$\([^\)]*\)') 

# --- classes -------------------------------------------------------------------------

class Job:

    def __init__(self, job_name, job_state=""):
	"""
	Initializes the Job class, setting job name,
	and state, if provided
	"""
	self.name = job_name		# Job name
	self.state = job_state		# Job state
        self.sub_file = ""	    	# Submit file for this job
        self.out_file = ""		# Output file for this job
        self.err_file = ""		# Error file for this job
        self.cluster = ""		# Cluster id for this job (from Condor)
        self.process = ""		# Process id for this job (from Condor)
        self.sub_file_parsed = False 	# Flag to tell if we were able to parse this job's submit file
        self.site = ""			# Site where the job ran
        self.executable = ""		# Job's executable
        self.arguments = ""		# Job's arguments
        self.initial_dir = ""		# Job's initial dir (from submit file)
        self.transfer_input_files = ""	# Files to transfer when debugging a job
        self.retries = None             # Keep track of how many times a job is submitted
        self.is_subdax = False          # Flag to tell if job is a SUBDAX/pegasus-plan job
        self.is_subdag = False		# Flag to tell if job is a SUBDAG job in the dag file
        self.subdag_dir = ""		# Subdag directory from a SUBDAG job in the dag file
        self.dag_path = ""		# Full path to the dag file from a SUBDAG job
        self.dagman_out = ""		# dagman.out file for this job (only for clustered jobs)
        self.pre_script = ""		# SCRIPT PRE line from the dag file
        self.condor_subs = {}		# Lits of condor substitutions rom DAG VARS line

    def set_state(self, new_state):
	"""
	This function updates a job state
	"""
	self.state = new_state

# --- constants -----------------------------------------------------------------------

MAXLOGFILE = 1000				# For log file rotation, check files .000 to .999
debug_level = logging.WARNING			# For now

# --- global variables ----------------------------------------------------------------

prog_base = os.path.split(sys.argv[0])[1]	# Name of this program
input_dir = None				# Directory given in -i command line option
dag_path = None					# Path of the dag file
tsdl_path = None				# Path to monitord's log file
temp_dir = None					# Temporary log file created
debug_job = None				# Path of a submit file to debug
debug_dir = None				# Temp directory to use while debugging a job
workflow_type = None				# Type of the workflow being debugged
workflow_base_dir = ""				# Workflow submit_dir or dirname(jsd) from braindump file
run_monitord = 0				# Run monitord before trying to analyze the output
output_dir = None                               # Output_dir for all files written by monitord
top_dir = None                                  # Top_dir of the main workflow, for obtaining the db location
use_files = False                               # Flag for using files in the workflow dir instead of the db
quiet_mode = 0					# Prints out/err filenames instead of dumping their contents
strict_mode = 0					# Gets out/err filenames from submit file
summary_mode = 0                                # Print just the summary output
debug_mode = 0					# Mode that enables debugging a single job
print_invocation = 0				# Prints invocation command for failed jobs
print_pre_script = 0				# Prints the SCRIPT PRE line for failed jobs, if present
jsdl_filename = "jobstate.log"			# Default name of the log file to use
jobs = {}					# List of jobs found in the jobstate.log file
total = 0					# Number of total jobs
success = 0					# Number of successful jobs
failed = 0					# Number of failed jobs
unsubmitted = 0					# Number of unsubmitted jobs
unknown = 0					# Number of jobs in an unknown state
failed_jobs = []				# List of jobs that failed
unknown_jobs = []				# List of jobs that neither succeeded nor failed

# --- functions -----------------------------------------------------------------------

def get_jsdl_filename(input_dir):
    """
    This function parses the braindump file in the input_dir,
    retrieving the wf_uuid and assembling the filename for the
    jobstate.log file.
    """
    try:
        my_wf_params = utils.slurp_braindb(input_dir)
    except:
        logger.error("cannot read braindump.txt file... exiting...")
        sys.exit(1)

    if "wf_uuid" in my_wf_params:
        return my_wf_params["wf_uuid"]+"-"+jsdl_filename

    logger.error("braindump.txt does not contain wf_uuid... exiting...")
    sys.exit(1)

def create_temp_logfile(name):
    """
    This function uses tempfile.mkstemp to create a temporary 
    log filename in the /tmp directory
    """
    try:
	tmp_file = tempfile.mkstemp(prefix="%s-" % (name), suffix=".log", dir="/tmp")
    except:
	return None

    # Close file, we will use it later
    os.close(tmp_file[0])

    # Return filename
    return tmp_file[1]

def has_seen(job_name):
    """
    This function returns true if we are already tracking job_name
    """
    if job_name in jobs:
	return True
    return False

def add_job(job_name, job_state=""):
    """
    This function adds a job to our list
    """
    # Don't add the same job twice
    if job_name in jobs:
	return

    newjob = Job(job_name, job_state)
    jobs[job_name] = newjob

def update_job_state(job_name, job_state=""):
    """
    This function updates the job state of a given job
    """
    # Make sure we have this job
    if not job_name in jobs:
	# Print a warning message
	logger.error("could not find job %s" % (job_name))
	return

    jobs[job_name].set_state(job_state)

def update_job_condor_info(job_name, condor_id="-"):
    """
    This function updates a job's condor_id (it splits it into process
    and cluster)
    """
    # Make sure we have this job
    if not job_name in jobs:
	# Print a warning message
	logger.error("could not find job %s" % (job_name))
	return

    # Nothing to do if condor_id is not defined
    if condor_id == "-":
	return

    my_split = condor_id.split(".")

    # First part is cluster id
    jobs[job_name].cluster = my_split[0]

    # If we have two pieces, second piece is process
    if len(my_split) >= 2:
	jobs[job_name].process = my_split[1]

def analyze():
    """
    This function processes all currently known jobs, generating some statistics
    """
    global total, success, failed, unsubmitted, unknown

    for my_job in jobs:
        total = total + 1
	if (jobs[my_job].state == "POST_SCRIPT_SUCCESS" or
	    jobs[my_job].state == "JOB_SUCCESS"):
	    success = success + 1
	elif (jobs[my_job].state == "POST_SCRIPT_FAILURE" or
	      jobs[my_job].state == "JOB_FAILURE"):
	    failed_jobs.append(my_job)
            failed = failed + 1
	elif (jobs[my_job].state == "UNSUBMITTED"):
	    unsubmitted = unsubmitted + 1
	else:
	    # It seems we don't have a final result for this job
	    unknown_jobs.append(my_job)
            unknown = unknown + 1

def parse_submit_file(my_job):
    """
    This function opens a submit file and reads site
    and condor dagman log information
    """
    # First we check if this is a SUBDAG job from the dag file
    if my_job.is_subdag:
	# Nothing to do here
	return

    if my_job.sub_file == "":
	# Create full path for the submit file if we already don't have the sub file set up
	my_job.sub_file = os.path.join(input_dir, my_job.name + ".sub")
    my_job.out_file = os.path.join(input_dir, my_job.name + ".out")
    my_job.err_file = os.path.join(input_dir, my_job.name + ".err")

    # Try to access submit file
    if os.access(my_job.sub_file, os.R_OK):
	# Open submit file
	try:
	    SUB = open(my_job.sub_file, "r")
	except:
	    # print "error opening submit file: %s" % (my_job.sub_file)
	    # fail silently for now...
	    return
	
	# submit file found
	my_job.sub_file_parsed = True

	# Check if this job includes sub workflows
	if my_job.is_subdax:
	    has_sub_workflow = True
	else:
	    has_sub_workflow = False

	# Parse submit file
	for line in SUB:
	    # First we need to do some trimming...
	    line = line.strip(" \t") # Remove leading and trailing spaces
	    if line.startswith('#'):
		# Skip comments
		continue
	    line = line.rstrip("\n\r") # Remove new lines, if any
	    line = line.split('#')[0] # Remove inline comments too
	    line = line.strip() # Remove any remaining spaces at both ends
	    if len(line) == 0:
		# Skip empty lines
		continue
	    prop = re_parse_property.search(line)
	    if prop:
		# Parse successful
		k = prop.group(1)
		v = prop.group(2)

		# See if it is one of the properties we are looking for...
		if k == "+pegasus_site":
		    my_job.site = v.strip('"')
		    continue
		if k == "arguments":
		    my_job.arguments = v.strip('"')
		if k == "executable":
		    my_job.executable = v
		if k == "environment" and has_sub_workflow:
		    # Ok, we need to find the CONDOR_DAGMAN_LOG entry now...
		    sub_props = v.split(';')
		    for sub_prop_line in sub_props:
			sub_prop_line = sub_prop_line.strip() # Remove any spaces
			if len(sub_prop_line) == 0:
			    continue
			sub_prop = re_parse_property.search(sub_prop_line)
			if sub_prop:
			    if sub_prop.group(1) == "_CONDOR_DAGMAN_LOG":
				my_job.dagman_out = sub_prop.group(2)
                                my_job.dagman_out = os.path.normpath(my_job.dagman_out)
                                if my_job.dagman_out.find(workflow_base_dir) >= 0:
                                    # Path to dagman_out file includes original submit_dir, let's try to change it
                                    my_job.dagman_out = os.path.normpath(my_job.dagman_out.replace((workflow_base_dir + os.sep), '', 1))
                                    # Join with current input_dir
                                    my_job.dagman_out = os.path.join(input_dir, my_job.dagman_out)

                                    # Now, figure out the correct directory, accounting for
                                    # replanning and rescue modes
                                    
                                    # Split filename into dir and base names
                                    my_dagman_dir = os.path.dirname(my_job.dagman_out)
                                    my_dagman_file = os.path.basename(my_job.dagman_out)
                                    my_retry = my_job.retries
                                    if my_retry is None:
                                        logger.warning("sub-workflow retry counter not initialized... continuing...")
                                        continue

                                    # Compose directory... assuming replanning mode
                                    my_retry_dir = my_dagman_dir + ".%03d" % (my_retry)
                                    # If directory doesn't exist, let's change to rescue mode
                                    if not os.path.isdir(my_retry_dir):
                                        logger.debug("sub-workflow directory %s does not exist, shifting to rescue mode..." % (my_retry_dir))
                                        my_retry_dir = my_dagman_dir + ".000"
                                        if not os.path.isdir(my_retry_dir):
                                            # Still not able to find it, output warning message
                                            logger.warning("sub-workflow directory %s does not exist!"
                                                           % (my_retry_dir))
                                            continue

                                    # Found sub-workflow directory, let's compose the final path to the new dagman.out file...
                                    my_job.dagman_out = os.path.join(my_retry_dir, my_dagman_file)



		# Only parse following keys if we are running in strict mode
		if strict_mode:
		    # Get initial dir
		    if k == "initialdir":
			my_job.initial_dir = v
		    # Parse error and output keys
		    if k == "output" or k == "error":
			# Take care of basic substitutions first
			v = v.replace("$(cluster)", my_job.cluster)
			v = v.replace("$(process)", my_job.process)

			# Now we do any substitutions from the DAG's VAR line (if any)
			for my_key in my_job.condor_subs:
			    v = v.replace("$(%s)" % (my_key), my_job.condor_subs[my_key])

			# Now, we collapse any remaining substitutions (not found in the VAR line)
			v = re_collapse_condor_subs.sub('', v)

			# Make sure we have an absolute path
			if not os.path.isabs(v):
			    v = os.path.join(input_dir, v)

			# Done! Replace out/err filenames with what we have
			if k == "output":
			    my_job.out_file = v
			else:
			    my_job.err_file = v
		# Only parse following keys if we are debugging a job
		if debug_mode:
		    # Get transfer input files
		    if k == "transfer_input_files":
			my_job.transfer_input_files = v
	SUB.close()
	# If initialdir was specified, we need to make both error and output files relative to that
	if len(my_job.initial_dir):
	    my_job.out_file = os.path.join(my_job.initial_dir, my_job.out_file)
	    my_job.err_file = os.path.join(my_job.initial_dir, my_job.err_file)
    else:
	# Was not able to access submit file
	# fail silently for now...
	# print "cannot access submit file: %s" % (my_job.sub_file)
	pass

def find_file(input_dir, file_type):
    """
    This function finds a file with the suffix file_type
    in the input directory. We assume there is just one
    file of the requested type in the directory (otherwise
    the function will return the first file matching the type
    """
    try:
	file_list = os.listdir(input_dir)
    except:
	logger.error("cannot read directory: %s" % (input_dir))
	sys.exit(1)

    for file in file_list:
	if file.endswith(file_type):
	    return os.path.join(input_dir, file)

    logger.error("could not find any %s file in %s" % (file_type, input_dir))
    sys.exit(1)

def parse_dag_file(dag_fn):
    """
    This function walks through the dag file, learning about
    all jobs before hand.
    """
    # Open dag file
    try:
	DAG = open(dag_fn, "r")
    except:
	logger.error("could not open dag file %s: exiting..." % (dag_fn))
	sys.exit(1)

    # Loop through the dag file
    for line in DAG:
	line = line.strip(" \t")
	if line.startswith("#"):
	    # Skip comments
	    continue
	line = line.rstrip("\n\r") # Remove new lines, if any
	line = line.split('#')[0] # Remove inline comments too
	line = line.strip() # Remove any remaining spaces at both ends
	if len(line) == 0:
	    # Skip empty lines
	    continue
	if line.startswith("JOB"):
	    # This is a job line, let's parse it
	    my_job = line.split()
	    if len(my_job) != 3:
		logger.warn("confused parsing dag line: %s" % (line))
		continue
	    if not has_seen(my_job[1]):
		add_job(my_job[1], "UNSUBMITTED")
		# Get submit file information from dag file
		jobs[my_job[1]].sub_file = os.path.join(input_dir, my_job[2])
                if my_job[1].startswith("pegasus-plan") or my_job[1].startswith("subdax_"):
                    # Mark job as subdax
                    jobs[my_job[1]].is_subdax = True
	    else:
		logger.warn("job appears twice in dag file: %s" % (my_job[1]))
	if line.startswith("SUBDAG EXTERNAL"):
	    # This is a subdag line, parse it to get job name and directory
	    my_job = line.split()
	    if len(my_job) != 6:
		logger.warn("confused parsing dag line: %s" % (line))
		continue
	    if not has_seen(my_job[2]):
		add_job(my_job[2], "UNSUBMITTED")
		jobs[my_job[2]].is_subdag = True
		jobs[my_job[2]].dag_path = my_job[3]
		jobs[my_job[2]].subdag_dir = my_job[5]
	    else:
		logger.warn("job appears twice in dag file: %s" % (my_job[2]))
	if line.startswith("SCRIPT PRE"):
	    # This is a SCRIPT PRE line, parse it to get the script for the job
	    my_script = re_parse_script_pre.search(line)
	    if my_script is None:
		# Couldn't parse line
		logger.warn("confused parsing dag line: %s" % (line))
		continue
	    # Get job name, and check if we have it
	    my_job = my_script.group(1)
	    if not has_seen(my_job):
		# Cannot find this job, ignore this line
		logger.warn("couldn't find job: %s for PRE SCRIPT line in dag file" % (my_job))
		continue
	    # Good, copy PRE script line to our job structure
	    jobs[my_job].pre_script = my_script.group(2)
	if line.startswith("VARS"):
	    # This is a VARS line, parse it to get the condor substitutions
	    if len(line.split()) > 2:
		# Line looks promising...
		my_job = line.split()[1]
		if not has_seen(my_job):
		    # Cannot find this job, ignore this line
		    logger.warn("couldn't find job: %s for VARS line in dag file" % (my_job))
		    continue
		# Good, parse the condor substitutions, and create substitution dictionary
		for my_key, my_val in re_parse_condor_subs.findall(line):
		    jobs[my_job].condor_subs[my_key] = my_val

def parse_jobstate_log(jobstate_fn):
    """
    This function parses the jobstate.log file, loading all job information
    """
    # Open log file
    try:
	JSDL = open(jobstate_fn, "r")
    except:
	logger.error("could not open file %s: exiting..." % (jobstate_fn))
	sys.exit(1)

    # Loop through the log file
    for line in JSDL:
	sp = line.split()
	# Skip lines that don't have enough items
	if len(sp) < 6:
	    continue
	# Skip monitord comments
	if sp[1] == "INTERNAL":
	    continue

	# Ok, we have a valid job
	jobname = sp[1]
	jobstate = sp[2]
	condor_id = sp[3]

	# Add to job list if we have never seen this job before
	if not has_seen(jobname):
	    logger.warn("job %s not present in dag file" % (jobname))
	    add_job(jobname, jobstate)
            if jobname.startswith("pegasus-plan") or jobname.startswith("subdax_"):
                # Mark job as subdax
                jobs[jobname].is_subdax = True
	else:
	    # Update job state
	    update_job_state(jobname, jobstate)

	# Update condor id if we reached the SUBMIT state
	if jobstate == "SUBMIT":
	    update_job_condor_info(jobname, condor_id)
            # Keep track of retries
            if jobs[jobname].retries is None:
                jobs[jobname].retries = 0
            else:
                jobs[jobname].retries = jobs[jobname].retries + 1

    # Close log file
    JSDL.close()

def find_latest_log(log_file_base):
    """
    This function tries to locate the latest log file
    """
    last_log = None
    curr_log = None

    if os.access(log_file_base, os.F_OK):
	last_log = log_file_base

    # Starts from .000
    sf = 0

    while (sf < MAXLOGFILE):
	curr_log = log_file_base + ".%03d" % (sf)
	if os.access(curr_log, os.F_OK):
	    last_log = curr_log
	    sf = sf + 1
	else:
	    break

    return last_log

def invoke_monitord(dagman_out_file, output_dir):
    """
    This function runs monitord on the given dagman_out_file.
    """
    monitord_cmd = "pegasus-monitord -r --no-events"
    if output_dir is not None:
        # Add output_dir, if given
        monitord_cmd = monitord_cmd + " --output-dir " + output_dir
    monitord_cmd = monitord_cmd + " " + dagman_out_file
    logger.info("running: %s" % (monitord_cmd))

    try:
	status, output = commands.getstatusoutput(monitord_cmd)
    except:
	logger.error("could not invoke monitord, exiting...")
	sys.exit(1)

def dump_file(file):
    """
    This function dumps a file to our stdout
    """
    if file is not None:
	try:
	    OUT = open(file, 'r')
	except:
	    logger.warn("*** Cannot access: %s" % (file))
	    print
	else:
	    print os.path.split(file)[1].center(80, '-')
	    print
	    # Dump file contents to terminal
	    line = OUT.readline()
	    while line:
		line = line.strip()
		print line
		line = OUT.readline()

	    OUT.close()
	    print

def print_output_error(job):
    """
    This function outputs both output and error files for a given job.
    """
    out_file = find_latest_log(job.out_file)
    err_file = find_latest_log(job.err_file)

    my_parser = kickstart_parser.Parser(out_file)
    my_output = my_parser.parse_stdout_stderr()
    my_task_id = 0

    if len(my_output) > 0:
        # Ok, we got valid kickstart records, output stdout and stderr for tasks that failed
        for entry in my_output:
            # Count tasks, the same way as pegasus-monitord for Stampede
            my_task_id = my_task_id + 1
            if not "derivation" in entry or not "transformation" in entry:
                continue
            if not "exitcode" in entry and not "error" in entry:
                continue
            if "exitcode" in entry:
                try:
                    if int(entry["exitcode"]) == 0:
                        # Skip tasks with exitcode equals to zero
                        continue
                except:
                    logger.warn("couldn't convert exitcode to integer!")
                    continue
            else:
                # We must have "error" in entry
                pass
            # Got a task with a non-zero exitcode
            print ("Task #" + str(my_task_id) + " - Summary").center(80, '-')
            print
            if "resource" in entry:
                print "site        : %s" % (entry["resource"])
            if "hostname" in entry:
                print "hostname    : %s" % (entry["hostname"])
            if "name" in entry:
                print "executable  : %s" % (entry["name"])
            if "argument-vector" in entry:
                print "arguments   : %s" % (entry["argument-vector"])
            if "exitcode" in entry:
                print "exitcode    : %s" % (entry["exitcode"])
            else:
                if "error" in entry:
                    print "error       : %s" % (entry["error"])
            if "cwd" in entry:
                print "working dir : %s" % (entry["cwd"])
            print
            # Now let's display stdout and stderr
            if "stdout" in entry:
                if len(entry["stdout"]) > 0:
                    # Something to display
                    print ("Task #" + str(my_task_id) + " - " + entry["transformation"] + " - " + entry["derivation"] + " - stdout").center(80, '-')
                    print
                    print entry["stdout"]
                    print
            if "stderr" in entry:
                if len(entry["stderr"]) > 0:
                    # Something to display
                    print ("Task #" + str(my_task_id) + " - " + entry["transformation"] + " - " + entry["derivation"] + " - stderr").center(80, '-')
                    print
                    print entry["stderr"]
                    print
    else:
        # Not able to parse the kickstart output file, let's just dump the out and err files

	# Print outfile to screen
        dump_file(out_file)

	# Print errfile to screen
        dump_file(err_file)

def print_job_info(job):
    """
    This function prints the information about a particular job
    """
    print
    print job.center(80, '=')
    print
    print " last state: %s" % (jobs[job].state)
    parse_submit_file(jobs[job])

    # Handle subdag jobs from the dag file
    if jobs[job].is_subdag == True:
	print " This is a SUBDAG job:"
	print " For more information, please run the following command:"
	user_cmd = " %s -s " % (prog_base)
        if output_dir is not None:
            user_cmd = user_cmd + " --output-dir %s" % (output_dir)
	print "%s -f %s" % (user_cmd, jobs[job].dag_path)
	print
	return

    if jobs[job].sub_file_parsed == False:
	print "       site: submit file not available"
    else:
	print "       site: %s" % (jobs[job].site or '-')
    print "submit file: %s" % (jobs[job].sub_file)
    print "output file: %s" % (find_latest_log(jobs[job].out_file))
    print " error file: %s" % (find_latest_log(jobs[job].err_file))
    if print_invocation:
	print
	print "To re-run this job, use: %s %s" % (jobs[job].executable, jobs[job].arguments)
	print
    if print_pre_script and len(jobs[job].pre_script) > 0:
	print
	print "SCRIPT PRE:"
	print jobs[job].pre_script
	print
    if len(jobs[job].dagman_out) > 0:
	# This job has a sub workflow
	print " This job contains sub workflows!"
	print " Please run the command below for more information:"
	user_cmd = " %s" % (prog_base)
        if output_dir is not None:
            user_cmd = user_cmd + " --output-dir %s" % (output_dir)
	print "%s -d %s" % (user_cmd, os.path.split(jobs[job].dagman_out)[0])
	print
    print

    # Now dump file contents to screen if we are not in quiet mode
    if not quiet_mode:
        print_output_error(jobs[job])

def print_top_summary():
    """
    This function prints the summary for the analyzer report,
    which is the same for the long and short output versions
    """
    print
    print "Summary".center(80, '*')
    print
    print " Total jobs         : % 6d (%3.2f%%)" % (total, 100 * (1.0 * total/(total or 1)))
    print " # jobs succeeded   : % 6d (%3.2f%%)" % (success, 100 * (1.0 * success/(total or 1)))
    print " # jobs failed      : % 6d (%3.2f%%)" % (failed, 100 * (1.0 * failed/(total or 1)))
    print " # jobs unsubmitted : % 6d (%3.2f%%)" % (unsubmitted, 100 * (1.0 * unsubmitted/(total or 1)))
    if unknown > 0:
	print " # jobs unknown     : % 6d (%3.2f%%)" % (unknown, 100 * (1.0 * unknown/(total or 1)))
    print

def print_summary():
    """
    This function prints the analyzer report summary
    """

    # First print the summary section
    print_top_summary()

    # Print information about failed jobs
    if len(failed_jobs):
	print "Failed jobs' details".center(80, '*')
	for job in failed_jobs:
	    print_job_info(job)

    # Print information about unknown jobs
    if len(unknown_jobs):
	print "Unknown jobs' details".center(80, '*')
	for job in unknown_jobs:
	    print_job_info(job)

def analyze_files():
    """
    This function runs the analyzer using the files in the workflow
    directory as the data source.
    """
    jsdl_path = None                            # Path of the jobstate.log file
    run_directory_writable = False              # Flag to indicate if we can write to the run directory
    dagman_out_path = None			# Path to the dagman.out file

    global workflow_base_dir, dag_path

    # Get the dag file if it was not specified by the user
    if dag_path is None:
        dag_path = find_file(input_dir, ".dag")
        logger.info("using %s, use the --dag option to override" % (dag_path))

    # Build dagman.out path
    dagman_out_path = dag_path + ".dagman.out"

    # Check if we can write to the run directory
    run_directory_writable = os.access(input_dir, os.W_OK)

    # Invoke monitord if requested
    if run_monitord:
        if output_dir is not None:
            # If output_dir is specified, invoke monitord with that path
            invoke_monitord("%s.dagman.out" % (dag_path), output_dir)
            # jobstate.log file uses wf_uuid as prefix
            jsdl_path = os.path.join(output_dir, get_jsdl_filename(input_dir))
        else:
            if run_directory_writable:
                # Run directory is writable, write monitord output to jobstate.log file
                jsdl_path = os.path.join(input_dir, jsdl_filename)
                # Invoke monitord
                invoke_monitord("%s.dagman.out" % (dag_path), None)
            else:
                # User must provide the --output-dir option
                logger.error("%s is not writable" % (input_dir))
                logger.error("user must specify directory for new monitord logs with the --output-dir option")
                logger.error("exiting...")
                sys.exit(1)
    else:
        if output_dir is not None:
            # jobstate.log file uses wf_uuid as prefix and is inside output_dir
            jsdl_path = os.path.join(output_dir, get_jsdl_filename(input_dir))
        else:
            jsdl_path = os.path.join(input_dir, jsdl_filename)

    # Compare timestamps of jsdl_path with dagman_out_path
    try:
        jsdl_stat = os.stat(jsdl_path)
    except:
        logger.error("could not access %s, exiting..." % (jsdl_path))
        sys.exit(1)

    try:
        dagman_out_stat = os.stat(dagman_out_path)
    except:
        logger.error("could not access %s, exiting..." % (dagman_out_path))
        sys.exit(1)

    # Compare mtime for both files
    if dagman_out_stat[8] > jsdl_stat[8]:
        logger.warning("jobstate.log older than the dagman.out file, workflow logs may not be up to date...")

    # Try to parse workflow parameters from braindump.txt file
    wfparams = utils.slurp_braindb(input_dir)
    if "submit_dir" in wfparams:
        workflow_base_dir = os.path.normpath(wfparams["submit_dir"])
    elif "jsd" in wfparams:
        workflow_base_dir = os.path.dirname(os.path.normpath(wfparams["jsd"]))

    # First we learn about jobs by going through the dag file
    parse_dag_file(dag_path)

    # Read logfile
    parse_jobstate_log(jsdl_path)

    # Process our jobs
    analyze()

    # Print summary of our analysis
    if summary_mode :
        print_top_summary()
    else :
        # This is non summary mode despite of the name (go figure)    
        print_summary()

    if failed > 0:
        # Workflow has failures, exit with exitcode 2
        sys.exit(2)
    # Workflow has no failures, exit with exitcode 0
    sys.exit(0)

def analyze_db(config_properties):
    """
    This function runs the analyzer using data from the database.
    """
    global total, success, failed, unsubmitted, unknown

    # Get the database URL
    output_db_url, wf_uuid = db_utils.get_db_url_wf_uuid(input_dir, config_properties, top_dir)

    # Nothing to do if we cannot resolve the database URL
    if output_db_url is None:
        logger.error("cannot find database URL, exiting...")
        sys.exit(1)

    # Now, let's try to access the database
    try:
        workflow_stats = stampede_statistics.StampedeStatistics(output_db_url, False)
        workflow_stats.initialize(wf_uuid)
    except SchemaVersionError:
        logger.error("------------------------------------------------------")
        logger.error("Database schema mismatch! Please run the upgrade tool")
        logger.error("to upgrade the database to the latest schema version.")
        sys.exit(1)
    except:
        logger.error("Failed to load the database." + output_db_url )
        logger.warning(traceback.format_exc())
        sys.exit(1)

    total = workflow_stats.get_total_jobs_status()
    success = workflow_stats.get_total_succeeded_jobs_status()
    failed = workflow_stats.get_total_failed_jobs_status()
    unsubmitted = total - success - failed

    # Let's print the results
    
    print_top_summary()

    # Exit if summary mode is on
    if summary_mode :
         if failed > 0:
             # Workflow has failures, exit with exitcode 2
             sys.exit(2)
         # Workflow has no failures, exit with exitcode 0  
         sys.exit(0)

    # Now, print information about jobs that failed...
    if failed > 0:
        # Get list of failed jobs from database
        my_failed_jobs = workflow_stats.get_failed_job_instances(final=True, all_jobs=True)
        
        # Print header
	print "Failed jobs' details".center(80, '*')

        # Now process one by one...
        for my_job in my_failed_jobs:
            my_info = workflow_stats.get_job_instance_info(my_job[0])[0]
            my_tasks = workflow_stats.get_invocation_info(my_job[0])

            # Unquote stdout and stderr
            my_info.stdout_text = utils.unquote(my_info.stdout_text or "")
            my_info.stderr_text = utils.unquote(my_info.stderr_text or "")
            my_info.stdout_text = my_info.stdout_text.strip(" \n\r\t")
            my_info.stderr_text = my_info.stderr_text.strip(" \n\r\t")
            
            if my_job[0] == my_info.job_instance_id:

                print
                print my_info.job_name.center(80, '=')
                print
                print " last state: %s" % (my_info.state or '-')

                print "       site: %s" % (my_info.site or '-')
                print "submit file: %s" % (my_info.submit_file or '-')
                print "output file: %s" % (my_info.stdout_file or '-')
                print " error file: %s" % (my_info.stderr_file or '-')
                if print_invocation:
                    print
                    print "To re-run this job, use: %s %s" % ((my_info.executable or '-'), (my_info.argv or '-'))
                    print
                if print_pre_script and len((my_info.pre_executable or "")) > 0:
                    print
                    print "SCRIPT PRE:"
                    print "%s %s" % ((my_info.pre_executable or ""), (my_info.pre_argv or ""))
                    print
                if my_info.subwf_dir is not None:
                    # This job has a sub workflow
                    print " This job contains sub workflows!"
                    print " Please run the command below for more information:"
                    user_cmd = " %s" % (prog_base)
                    my_wfdir = os.path.normpath(my_info.subwf_dir)
                    if my_wfdir.find(my_info.submit_dir) >= 0:
                        # Path to dagman_out file includes original submit_dir, let's try to change it...
                        my_wfdir = os.path.normpath(my_wfdir.replace((my_info.submit_dir + os.sep), '', 1))
                        my_wfdir = os.path.join(input_dir, my_wfdir)
                    print "%s -d %s --top-dir %s" % (user_cmd, my_wfdir, (top_dir or input_dir))
                    print
                print

                # Now, print task information
                for my_task in my_tasks:
                    if my_task[0] < 1:
                        # Skip pre, and post script tasks
                        continue
                    if my_task[1] == 0:
                        # Skip tasks that succeeded
                        continue

                    # Got a task with a non-zero exitcode
                    my_exitcode = utils.raw_to_regular(my_task[1])

                    # Print task summary
                    print ("Task #" + str(my_task[0]) + " - Summary").center(80, '-')
                    print
                    print "site        : %s" % (my_info.site or '-')
                    print "hostname    : %s" % (my_info.hostname or '-')
                    print "executable  : %s" % (str(my_task[2] or '-'))
                    print "arguments   : %s" % (str(my_task[3] or '-'))
                    print "exitcode    : %s" % (str(my_exitcode))
                    print "working dir : %s" % (my_info.work_dir or '-')
                    print

                    if quiet_mode:
                        continue

                    # Now, print task stdout and stderr, if anything is there
                    my_stdout_str = "#@ %d stdout" % (my_task[0])
                    my_stderr_str = "#@ %d stderr" % (my_task[0])

                    # Start with stdout
                    my_stdout_start = my_info.stdout_text.find(my_stdout_str)
                    if my_stdout_start >= 0:
                        my_stdout_start = my_stdout_start + len(my_stdout_str) + 1
                        my_stdout_end = my_info.stdout_text.find("#@", my_stdout_start)
                        if my_stdout_end < 0:
                            # Next comment not found, possibly the last entry
                            my_stdout_end = len(my_info.stdout_text)
                        else:
                            my_stdout_end = my_stdout_end - 1

                        if my_stdout_end - my_stdout_start > 0:
                            # Something to display
                            print ("Task #" + str(my_task[0]) + " - " + str(my_task[4]) + " - " + str(my_task[5]) + " - stdout").center(80, '-')
                            print
                            print my_info.stdout_text[my_stdout_start:my_stdout_end]
                            print

                    # Now print stderr (from the kickstart output file)
                    my_stderr_start = my_info.stdout_text.find(my_stderr_str)
                    if my_stderr_start >= 0:
                        my_stderr_start = my_stderr_start + len(my_stderr_str) + 1
                        my_stderr_end = my_info.stdout_text.find("#@", my_stderr_start)
                        if my_stderr_end < 0:
                            # Next comment not found, possibly the last entry
                            my_stderr_end = len(my_info.stdout_text)
                        else:
                            my_stderr_end = my_stderr_end - 1

                        if my_stderr_end - my_stderr_start > 0:
                            # Something to display
                            print ("Task #" + str(my_task[0]) + " - " + str(my_task[4]) + " - " + str(my_task[5]) + " - Kickstart stderr").center(80, '-')
                            print
                            print my_info.stdout_text[my_stderr_start:my_stderr_end]
                            print

                    # Now print the stderr output from the .err file
                    if my_info.stderr_text.strip("\n\t \r") != "":
                        # Something to display
                        print ("Task #" + str(my_task[0]) + " - " + str(my_task[4]) + " - " + str(my_task[5]) + " - stderr").center(80, '-')
                        print
                        print my_info.stderr_text
                        print
            else:
                log.error("unexpected job instance returned by database!")
                log.error("returned: %d - expected: %d" % (my_info[0], my_job[0]))
                continue

    # Done with the database
    workflow_stats.close()

    if failed > 0:
        # Workflow has failures, exit with exitcode 2
        sys.exit(2)

    # Workflow has no failures, exit with exitcode 0
    sys.exit(0)

def debug_condor(my_job):
    """
    This function is used to debug a condor job. It creates a
    shell script in the debug_dir directory that is used to
    copy all necessary files to the (local) debug_dir directory
    and then execute the job locally.
    """
    global strict_mode

    # Set strict mode in order to parse everything in the submit file
    strict_mode = 1
    # Parse submit file
    parse_submit_file(my_job)

    # Create script name
    debug_script_basename = "debug_" + my_job.name + ".sh"
    debug_script_name = os.path.join(debug_dir, debug_script_basename)

    job_executable = os.path.join(debug_dir, my_job.executable) + my_job.arguments

    try:
	debug_script = open(debug_script_name, "w")
    except:
	logger.error("cannot create debug script %s" % (debug_script))
	sys.exit(1)

    try:
	# Start with the bash line
	debug_script.write("#!/bin/bash\n")
	debug_script.write("\n")
	debug_script.write("set -e\n")
	debug_script.write("\n")
	debug_script.write("# Copy any files that are needed\n")

	debug_script.write('echo "copying input files..."\n')
	debug_script.write("\n")
	# Copy all files that we need
	for my_file in my_job.transfer_input_files.split(","):
	    if len(my_file):
		if len(my_job.initial_dir):
		    # Add the initial dir to all files to be copied
		    my_file = os.path.join(my_job.initial_dir, my_file)
		debug_script.write("cp %s %s\n" % (my_file, debug_dir))

	# Extra newline before executing the job
	debug_script.write("\n")
	debug_script.write('echo "copying input files completed."\n')
	debug_script.write("\n")
	debug_script.write("# Set the execute bit on the executable\n")
	debug_script.write("chmod +x %s\n" % (os.path.join(debug_dir, my_job.executable)))
	debug_script.write("\n")
	debug_script.write('echo "executing job: %s"\n' % (job_executable))
	debug_script.write("\n")
	debug_script.write("# Now, execute the job\n")
	debug_script.write("%s\n" % (job_executable))
	debug_script.write("\n")
        # Remember not to put anything between running the executable
        # and checking the exit code, otherwise $? will break...
        debug_script.write("# Check error code\n")
        debug_script.write("if [ $? -eq 0 ]; then\n")
        debug_script.write('   echo "executable ran successfully"\n')
        debug_script.write("else\n")
        debug_script.write('   echo "executable failed with error $?"\n')
    except:
	logger.error("cannot write to file %s" % (debug_script))
	sys.exit(1)

    # We are done writing the file!
    debug_script.close()

    try:
	# Make our debug script executable
	os.chmod(debug_script_name, 0755)
    except:
	logger.error("cannot change permissions for the debug script %s" % (debug_script))
	sys.exit(1)

    # Print next step
    print
    print "%s: finished generating job debug script!" % (prog_base)
    print
    print "To run it, you need to type:"
    print "   $ cd %s" % (debug_dir)
    print "   $ ./%s" % (debug_script_basename)
    print

def debug_workflow():
    """
    This function handles the mode where the analyzer
    is used to debug a job in a workflow
    """
    global debug_job, debug_dir

    # Check if we can find this job's submit file
    if not debug_job.endswith(".sub"):
	debug_job = debug_job + ".sub"
    # Figure out job name
    jobname = os.path.basename(debug_job)
    jobname = jobname[0:jobname.find(".sub")]
    # Create job class
    my_job = Job(jobname)
    my_job.sub_file = debug_job

    if not os.access(debug_job, os.R_OK):
	logger.error("cannot access job submit file: %s" % (debug_job))
	sys.exit(1)

    # Handle the temporary directory option
    if debug_dir is None:
	# Create temporary directory
	try:
	    debug_dir = tempfile.mkdtemp()
	except:
	    logger.error("could not create temporary directory!")
	    sys.exit(1)
    else:
	# Make sure directory specified is writable
	debug_dir = os.path.abspath(debug_dir)
	if not os.access(debug_dir, os.F_OK):
	    # Create directory if it does not exist
	    try:
		os.mkdir(debug_dir)
	    except:
		logger.error("cannot create debug directory: %s" % (debug_dir))

	# Check if we can write to the debug directory
	if not os.access(debug_dir, os.W_OK):
	    logger.error("not able to write to temporary directory: %s" % (debug_dir))
	    sys.exit(1)


    # Handle workflow type
    if workflow_type is not None:
	if workflow_type.lower() == "condor":
	    logger.info("debugging condor type workflow")
	    debug_condor(my_job)
	else:
	    logger.error("workflow type %s not supported!" % (workflow_type))
	    sys.exit(1)
    else:
	logger.info("debugging condor type workflow")
	debug_condor(my_job)

    # All done, in case we are back here!
    sys.exit(0)

# --- main ----------------------------------------------------------------------------

# Configure command line option parser
prog_usage = "usage: %s [options] workflow_directory" % (prog_base)
parser = optparse.OptionParser(usage=prog_usage)
parser.add_option("-v", "--verbose", action="count", default=0, dest="vb",
                  help="Increase verbosity, repeatable")
parser.add_option("-i", "-d", "--dir", action = "store", type = "string", dest = "input_dir",
		  help = "input directory where the jobstate.log file is located, default is the current directory")
parser.add_option("--dag", action = "store", type = "string", dest = "dag_filename",
		  help = "full path to the dag file to use -- this option overrides the -d option")
parser.add_option("-f", "--files", action = "store_const", const = 1, dest = "use_files",
		  help = "disables the database mode and forces the use of workflow directory files")
parser.add_option("-m", "-t", "--monitord", action = "store_const", const = 1, dest = "run_monitord",
		  help = "run pegasus-monitord before analyzing the output")
parser.add_option("-o", "--output-dir", action = "store", type = "string", dest = "output_dir",
                  help = "provides an output directory for all monitord log files")
parser.add_option("--top-dir", action = "store", type = "string", dest = "top_dir",
                  help = "provides the location of the top-level workflow directory, needed to analyze sub-workflows")
parser.add_option("-c","--conf", action = "store", type = "string", dest = "config_properties",
                  help = "Specifies the properties file to use. This overrides all other property files.")
parser.add_option("-q", "--quiet", action = "store_const", const = 1, dest = "quiet_mode",
		  help = "output out/err filenames instead of their contents")
parser.add_option("-p", "--print", action = "store", type = "string" , dest = "print_options",
		  help = "specifies print options from pre,invocation")
parser.add_option("-s", "--strict", action = "store_const", const = 1, dest = "strict_mode",
		  help = "gets a job's out and err files from the submit file")
parser.add_option("-S", "--summary", action = "store_const", const = 1, dest = "summary_mode",
		  help = "Just print the summary and exit")
parser.add_option("--debug-job", action = "store", type = "string", dest = "debug_job",
		  help = "specifies a job to debug (can be either the job base name or the submit file name) -- this option enables debugging a single job")
parser.add_option("--debug-dir", action = "store", type = "string", dest = "debug_dir",
		  help = "specifies the directory to use as debug directory (default is to create a random directory in /tmp)")
parser.add_option("--type", action = "store", type = "string", dest = "workflow_type",
		  help = "specifies what type of workflow we are debugging (available types: condor)")

# Parse command line options
(options, args) = parser.parse_args()

print "%s: initializing..." % (prog_base)

# Copy options from the command line parser

if options.vb == 0:
    lvl = logging.WARN
elif options.vb == 1:
    lvl = logging.INFO
else:
    lvl = logging.DEBUG
# Set logging level
logger.setLevel(lvl)

if options.run_monitord is not None:
    run_monitord = options.run_monitord
if options.strict_mode is not None:
    strict_mode = options.strict_mode
if options.summary_mode is not None:
    summary_mode = options.summary_mode
if options.quiet_mode is not None:
    quiet_mode = options.quiet_mode
if options.use_files is not None:
    use_files = True
if options.print_options is not None:
    my_options = options.print_options.split(",")
    if "pre" in my_options or "all" in my_options:
	print_pre_script = 1
    if "invocation" in my_options or "all" in my_options:
	print_invocation = 1
if options.output_dir is not None:
    output_dir = options.output_dir
if options.top_dir is not None:
    top_dir = os.path.abspath(options.top_dir)
if options.debug_job is not None:
    debug_job = options.debug_job
    # Enables the debugging mode
    debug_mode = 1
if options.debug_dir is not None:
    debug_dir = options.debug_dir
if options.workflow_type is not None:
    workflow_type = options.workflow_type

if options.dag_filename is not None:
    dag_path = options.dag_filename
    input_dir = os.path.abspath(os.path.split(dag_path)[0])
    # Assume current directory if input dir is empty
    if len(input_dir) == 0:
	input_dir = os.getcwd()
else:
    # Select directory where jobstate.log is located
    if options.input_dir is not None:
	input_dir = os.path.abspath(options.input_dir)
    else:
        if len(args) > 1:
            parser.print_help()
            sys.exit(1)
        elif len(args) == 1:
            input_dir = args[0]
        else:
            input_dir = os.getcwd()

if debug_mode == 1:
    # Enter debug mode if job name given
    # This function does not return
    debug_workflow()

# Run the analyzer
if use_files:
    analyze_files()
else:
    analyze_db(options.config_properties)

# Done!
print "Done".center(80, '*')
print
print "%s: end of status report" % (prog_base)
print