def payload4(task):
"""
run merge from N inputs
input: Makefile.all, *.fasta.{sfx list}, *1.{N}.fastq, *2.{N}.fastq, {N}reads.tgz, {N}maps.tgz
output: bam file + results.tgz?
:param task:
:return:
"""
logger.debug("payload4: Start")
#### Prepare
# Check type of task
task_type = task.task_type
# Get user
user = users_.get(task.owner_id)
task.tag = "task." + commands.getoutput('uuidgen')
tasks_.save(task)
n = 10
if task.params is not None:
n = int(task.params)
if n == 0:
n = 10
# Get containers
input_cont = conts_.get(task.input)
#TO_DO do smth with output container?
output_cont = conts_.get(task.output)
# Get container
container = Container()
container.guid = task.tag
conts_.save(container)
# Add input files to container
files_template_list = task_type.ifiles_template.split(',')
for item in input_cont.files:
f = item.file
for file_template in files_template_list:
# TO_DO: Change file template here
m = re.match(file_template, f.lfn)
if m is not None:
# Register file in container
fc.reg_file_in_cont(f, container, 'input')
# reg additional output
fc.reg_file_in_cont_byname(user, 'output.bam', container, 'output')
fc.reg_file_in_cont_byname(user, 'myresults.bz2', container, 'output')
# Prepare trf script
script = task.task_type.trf_template
# TO_DO just for test - only emulate, not real jobs
pipeline_path_name = 'paleomix_bam'
swdir = '/s/ls2/users/poyda/swp/' + pipeline_path_name + '/'
script = "/bin/bash " + swdir + "runmerge.sh -t " + str(n)
send_job_(task, container, script)
return True
def get_start_task(p):
"""
Returns start task object
:param p: Pipeline obj
:return: Task obj
"""
if not isinstance(p, Pipeline):
raise Exception("Illegal pipeline class: not Pipeline")
# Create start_task
start_task_type = task_types_.first(method='start')
start_task = new_task(start_task_type)
start_task.owner_id = p.owner_id
tasks_.save(start_task)
# Update Pipeline obj
set_current_task(p, start_task)
return start_task
def check_running_tasks():
"""
Checks PanDA jobs statuses for all running tasks
:return:
"""
# Get tasks in running state
tasks = tasks_.find(status='running')
for task in tasks:
# Check if tag defined
if task.tag is not None and task.tag != "":
# Check failed Panda jobs
jobs = jobs_.find(tags=task.tag, status='failed')
if jobs.count() > 0:
task.status = 'failed'
task.modification_time = datetime.utcnow()
task.comment = "Failed task due to {n} failed jobs".format(
n=jobs.count())
tasks_.save(task)
return False
# Check failed Panda jobs
jobs = jobs_.find(tags=task.tag, status='canceled')
if jobs.count() > 0:
task.status = 'cancelled'
task.modification_time = datetime.utcnow()
tasks_.save(task)
return False
# Check finished Panda jobs
jobs = jobs_.find(tags=task.tag, status='finished')
jobs_all = jobs_.find(tags=task.tag)
if jobs.count() == jobs_all.count():
# Register files from jobs into task container
cont = conts_.get(task.input)
for job in jobs:
files_catalog = job.container.files
for f in files_catalog:
if f.type == 'output':
# Register file in container
fc.reg_file_in_cont(f.file, cont, 'intermediate')
# Change task status
task.status = 'finished'
task.modification_time = datetime.utcnow()
tasks_.save(task)
return True
else:
# If tag is not defined
task.status = 'finished'
task.modification_time = datetime.utcnow()
tasks_.save(task)
return True
def check_running_tasks():
"""
Checks PanDA jobs statuses for all running tasks
:return:
"""
# Get tasks in running state
tasks = tasks_.find(status='running')
for task in tasks:
# Check if tag defined
if task.tag is not None and task.tag != "":
# Check failed Panda jobs
jobs = jobs_.find(tags=task.tag, status='failed')
if jobs.count() > 0:
task.status = 'failed'
task.modification_time = datetime.utcnow()
task.comment = "Failed task due to {n} failed jobs".format(n=jobs.count())
tasks_.save(task)
return False
# Check failed Panda jobs
jobs = jobs_.find(tags=task.tag, status='canceled')
if jobs.count() > 0:
task.status = 'cancelled'
task.modification_time = datetime.utcnow()
tasks_.save(task)
return False
# Check finished Panda jobs
jobs = jobs_.find(tags=task.tag, status='finished')
jobs_all = jobs_.find(tags=task.tag)
if jobs.count() == jobs_all.count():
# Register files from jobs into task container
cont = conts_.get(task.input)
for job in jobs:
files_catalog = job.container.files
for f in files_catalog:
if f.type == 'output':
# Register file in container
fc.reg_file_in_cont(f.file, cont, 'intermediate')
# Change task status
task.status = 'finished'
task.modification_time = datetime.utcnow()
tasks_.save(task)
return True
else:
# If tag is not defined
task.status = 'finished'
task.modification_time = datetime.utcnow()
tasks_.save(task)
return True
def new_pipeline():
form = NewPipelineForm(request.form)
if request.method == 'POST':
ifiles = request.form.getlist('iguids[]')
current_user = g.user
# Prepare pipeline
pp = Pipeline()
pp.status = 'running'
pp.type_id = pipeline_types_.get(1).id
pp.owner_id = current_user.id
pipelines_.save(pp)
# Prepare container
pp_cont = Container()
pp_cont.guid = 'pipeline.' + commands.getoutput('uuidgen')
conts_.save(pp_cont)
# Add guids to container
for item in ifiles:
if item != '':
f = files_.first(guid=item)
if f is not None:
# Register file in catalog
fc.reg_file_in_cont(f, pp_cont, 'input')
else:
pp_cont.status = 'broken'
conts_.save(pp_cont)
return make_response(jsonify({'error': "GUID {} not found".format(f)}))
# Set current task
start_task = pclient.get_start_task(pp)
start_task.input = pp_cont.id
start_task.output = pp_cont.id
tasks_.save(start_task)
return redirect(url_for('pipelines.list_all'))
return render_template('dashboard/pp/new.html', form=form)
def get_next_task(p):
"""
Returns next task object
:param p: Pipeline obj
:return: Task obj
"""
if not isinstance(p, Pipeline):
raise Exception("Illegal pipeline class: not Pipeline")
# Fetch current_task
current_task_id = p.current_task_id
current_task = tasks_.get(current_task_id)
if not isinstance(current_task, Task):
raise Exception("Unable to fetch current_task by id")
current_task_type = current_task.task_type
# Fetch pipeline_type
pipeline_type = p.pipeline_type
if not isinstance(pipeline_type, PipelineType):
raise Exception("Illegal pipeline_type class: not PipelineType")
# Fetch pipeline_catalog item
pp_c = pipeline_catalog_.first(pipeline_type_id=pipeline_type.id, current_task_type_id=current_task_type.id)
if not isinstance(pp_c, PipelineCatalog):
raise Exception("Unable to fetch PipelineCatalog item")
# Return next_task
next_task = new_task(pp_c.next_task_type)
next_task.owner_id = p.owner_id
next_task.input = current_task.input
next_task.output = current_task.output
next_task.params = current_task.params
tasks_.save(next_task)
# Update Pipeline obj
set_current_task(p, next_task)
return next_task
def new_pipeline_from_cont():
form = RunForm(request.form)
if request.method == 'POST':
icont = conts_.first(guid=form.guid.data)
if icont is None:
raise WebpandaError("Container not found")
current_user = g.user
# Prepare pipeline
pp = Pipeline()
pp.status = 'running'
pp.type_id = pipeline_types_.get(1).id
pp.owner_id = current_user.id
pipelines_.save(pp)
# Prepare container
pp_cont = Container()
pp_cont.guid = 'pipeline.' + commands.getoutput('uuidgen')
conts_.save(pp_cont)
# Add guids to container
for item in icont.files:
f = item.file
# Register file in catalog
fc.reg_file_in_cont(f, pp_cont, 'input')
# Set current task
start_task = pclient.get_start_task(pp)
start_task.input = pp_cont.id
start_task.output = pp_cont.id
tasks_.save(start_task)
return redirect(url_for('pipelines.list_all'))
return render_template('dashboard/pp/new.html', form=form)
def run(task):
try:
method = task.task_type.method
if task.status != 'sent':
return False
# raise WebpandaError('Illegal task status to start')
# Change task state to 'running'
task.status = 'preparing'
task.modification_time = datetime.utcnow()
tasks_.save(task)
# Custom payload
if method == 'init_task':
payload1(task)
elif method == 'split_task':
payload2(task)
elif method == 'run1_task':
payload3(task)
elif method == 'merge_task':
payload4(task)
else:
raise WebpandaError("Task payload error: method not found")
# Change task state to 'finished'
task.status = 'running'
task.modification_time = datetime.utcnow()
tasks_.save(task)
return True
except WebpandaError as e:
# Change task state to 'finished'
task.status = 'failed'
task.modification_time = datetime.utcnow()
task.comment = e.msg
tasks_.save(task)
return False
def payload3(task):
"""
run1 - N parallel jobs. {N} = sequence 0..01,0..02,...,N, not less than 2 placeholders
#TODO deal with {N}.fastq.bz2 ??
input: Makefile.{N}, *.fasta.{sfx list}, *1.{N}.fastq, *2.{N}.fastq
output: likely reads{N}.tgz, maps{N}.tgz
:param task:
:return:
"""
logger.debug("payload3: Start")
#### Prepare
# Check type of task
task_type = task.task_type
# if task_type.id != 3or6?:
# raise WebpandaError("Illegal task_type.id")
# Get user
user = users_.get(task.owner_id)
n=10
if task.params is not None:
n=int(task.params)
if n==0:
n=10
task.tag = "task." + commands.getoutput('uuidgen')
tasks_.save(task)
# Get containers
input_cont = conts_.get(task.input)
#TO_DO do smth with output container?
output_cont = conts_.get(task.output)
for jobname in gen_sfx("a",n):
# Get container
container = Container()
container.guid = task.tag + "."+jobname
conts_.save(container)
# Add input files to container
files_template_list = task_type.ifiles_template.split(',')
for item in input_cont.files:
f = item.file
for file_template in files_template_list:
# TO_DO: Change file template here
m = re.match(file_template, f.lfn)
if m is not None:
# Register file in container
fc.reg_file_in_cont(f, container, 'input')
# reg additional output
fc.reg_file_in_cont_byname(user, jobname+'.reads.bz2', container, 'output')
fc.reg_file_in_cont_byname(user, jobname + '.maps.bz2', container, 'output')
# Prepare trf script
script = task.task_type.trf_template
# TO_DO just for test - only emulate, not real jobs
pipeline_path_name = 'paleomix_bam'
swdir = '/s/ls2/users/poyda/swp/' + pipeline_path_name +'/'
script = "/bin/bash "+swdir+"run11.sh -t " +jobname
send_job_(task, container, script)
return True
def payload2(task):
"""
split_task
Split input *.1.fastq and *.2.fastq into 'rn' pieces=
run panda /bin/bash split.sh
:param task:
:return:
"""
logger.debug("payload2: Start")
#### Prepare
# Check type of task
task_type = task.task_type
if task_type.id != 1:
raise WebpandaError("Illegal task_type.id")
logger.debug("payload2: tasktype " + str(task_type.id))
# Get user
user = users_.get(task.owner_id)
logger.debug("payload2: user " + str(user.id))
# Get containers
input_cont = conts_.get(task.input)
#TODO do smth with output container?
output_cont = conts_.get(task.output)
task.tag = "task." + commands.getoutput('uuidgen')
tasks_.save(task)
logger.debug("payload2: tag " + task.tag)
# Get container
container = Container()
container.guid = task.tag + ".0"
conts_.save(container)
logger.debug("payload2: cont " + container.guid)
script_add = ""
rn = 0
# Add input files to container
files_template_list = task_type.ifiles_template.split(',')
for item in input_cont.files:
f = item.file
if rn == 0:
if f.lfn.endswith('fastq'):
rn = getn(f.fsize)
elif f.lfn.endswith('fastq.bz2'):
rn = getn2(f.fsize)
for file_template in files_template_list:
# TODO: Change file template here
m = re.match(file_template, f.lfn)
if m is not None:
# Register file in container
fc.reg_file_in_cont(f, container, 'input')
if f.lfn.endswith('.fastq'):
for fi in gen_sfx(f.lfn[:-5]+'a', rn, '.fastq'):
fc.reg_file_in_cont_byname(user, fi, container, 'output')
if f.lfn.endswith('.fastq.bz2'):
for fi in gen_sfx(f.lfn[:-9]+'a', rn, '.fastq'):
fc.reg_file_in_cont_byname(user, fi, container, 'output')
if f.lfn.endswith('.fasta'):
fn=f.lfn+'.'
fc.reg_file_in_cont_byname(user, fn[:-6]+'dict', container, 'output')
# itert: validated file has null size
#for sfx in ('amb','ann','bwt','fai','pac','sa','validated'):
for sfx in ('amb','ann','bwt','fai','pac','sa', 'validated'):
fc.reg_file_in_cont_byname(user, fn+sfx, container, 'output')
script_add += "; echo 123 > ../{fname}".format(fname=fn+"validated")
logger.debug("payload2: reg Makefile")
#reg additional output
for fi in gen_sfx('Makefile.a', rn, '.yaml'):
fc.reg_file_in_cont_byname(user, fi, container, 'output')
#guids = ["web.it_4b7d4757-9ba4-4ed7-8bc0-6edb8bcc68d2",
# "web.it_3bc78e60-241b-418a-a631-2461d4ba1977",
# "web.it_1b88049e-463b-4b4f-8454-9587301a53e5",
# "web.it_a02271ea-8a9b-42f3-add2-ed6d0f9ff07e",
# "web.it_61bb7c80-e53c-4641-88b0-fbd16b0f3d56",
# "web.it_3930f596-25ea-49b0-8943-7a83c84c7940",
# "web.it_aa7b77a3-c765-464e-a4fa-29ce6dd50346",
# "web.it_211f2187-41f2-489f-ba63-73f004f21c66"
# ]
#for guid in guids:
# fc.reg_file_in_cont(files_.first(guid=guid), container, 'input')
# Prepare trf script
script = task.task_type.trf_template
# TO_DO just for test add "1" - script1.sh- only emulate, not real jobs
pipeline_path_name = 'paleomix_bam'
swdir='/s/ls2/users/poyda/swp/' + pipeline_path_name +'/'
script = "/bin/bash " + swdir + "genref.sh && /bin/bash " + swdir + "runtmplgen.sh -t 1>bam.out 2>bam.err & ;"
script += "/bin/bash " + swdir + "split.sh -t " + str(rn)
script += script_add
# Save rn as task param
task.params = str(rn)
tasks_.save(task)
logger.debug("payload2: script " + script)
logger.debug("payload2: send_job " + container.guid)
send_job_(task, container, script)
return True
def run():
"""
Starts current defined task
:return:
"""
print "main started"
# Fetch pipelines (init state)
#TODO add SQL filter on status if possible
pipelines = pipelines_.all()
for pipeline in pipelines:
# Check if finished
if pipeline.status in ['finished', 'failed', 'cancelled']:
continue
# Fetch task object
current_task = pclient.get_current_task(pipeline)
if current_task is None:
return WebpandaError('Illegal task ID')
if current_task.status == 'failed':
#TODO: What to do if failed?
pipeline.status = 'failed'
current_task.modification_time = datetime.utcnow()
pipelines_.save(pipeline)
continue
if current_task.status == 'cancelled':
#TODO: What to do if cancelled? Who or by whom? If by system - resubmit, if by user -nothing?
pipeline.status = 'cancelled'
current_task.modification_time = datetime.utcnow()
pipelines_.save(pipeline)
continue
if current_task.status == 'finished':
# Get next_task
current_task = pclient.get_next_task(pipeline)
if current_task.status == 'defined':
if current_task.task_type.method == 'start':
# Do some general pipeline checks
current_task.status = 'finished'
current_task.modification_time = datetime.utcnow()
tasks_.save(current_task)
continue
elif current_task.task_type.method == 'finish':
current_task.status = 'finished'
current_task.modification_time = datetime.utcnow()
tasks_.save(current_task)
# Process system finish task
pipeline.status = 'finished'
pipeline.modification_time = datetime.utcnow()
pipelines_.save(pipeline)
continue
else:
# Process system start task
# Do some checks if it usefull - or we have all files already, or there would be never enough of them.
if not paleomix.has_input(current_task):
current_task.status = "failed"
current_task.modification_time = datetime.utcnow()
current_task.comment = "Input files check failed"
tasks_.save(current_task)
pipeline.status = 'failed'
pipeline.modification_time = datetime.utcnow()
pipelines_.save(pipeline)
continue
# Run task if defined
current_task.status = 'sent'
tasks_.save(current_task)
#TO_DO: Run async regime
paleomix.run(current_task) #, current_task.task_type.method) - we already get task from id. Not need to obtain again, is it?
# if we use async run, all params must be serializable (BaseQuery is not)
continue
def run():
"""
Starts current defined task
:return:
"""
print "main started"
# Fetch pipelines (init state)
#TODO add SQL filter on status if possible
pipelines = pipelines_.all()
for pipeline in pipelines:
# Check if finished
if pipeline.status in ['finished', 'failed', 'cancelled']:
continue
# Fetch task object
current_task = pclient.get_current_task(pipeline)
if current_task is None:
return WebpandaError('Illegal task ID')
if current_task.status == 'failed':
#TODO: What to do if failed?
pipeline.status = 'failed'
current_task.modification_time = datetime.utcnow()
pipelines_.save(pipeline)
continue
if current_task.status == 'cancelled':
#TODO: What to do if cancelled? Who or by whom? If by system - resubmit, if by user -nothing?
pipeline.status = 'cancelled'
current_task.modification_time = datetime.utcnow()
pipelines_.save(pipeline)
continue
if current_task.status == 'finished':
# Get next_task
current_task = pclient.get_next_task(pipeline)
if current_task.status == 'defined':
if current_task.task_type.method == 'start':
# Do some general pipeline checks
current_task.status = 'finished'
current_task.modification_time = datetime.utcnow()
tasks_.save(current_task)
continue
elif current_task.task_type.method == 'finish':
current_task.status = 'finished'
current_task.modification_time = datetime.utcnow()
tasks_.save(current_task)
# Process system finish task
pipeline.status = 'finished'
pipeline.modification_time = datetime.utcnow()
pipelines_.save(pipeline)
continue
else:
# Process system start task
# Do some checks if it usefull - or we have all files already, or there would be never enough of them.
if not paleomix.has_input(current_task):
current_task.status = "failed"
current_task.modification_time = datetime.utcnow()
current_task.comment = "Input files check failed"
tasks_.save(current_task)
pipeline.status = 'failed'
pipeline.modification_time = datetime.utcnow()
pipelines_.save(pipeline)
continue
# Run task if defined
current_task.status = 'sent'
tasks_.save(current_task)
#TO_DO: Run async regime
paleomix.run(
current_task
) #, current_task.task_type.method) - we already get task from id. Not need to obtain again, is it?
# if we use async run, all params must be serializable (BaseQuery is not)
continue
