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 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 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 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 send_job(jobid, siteid):
_logger.debug('Jobid: ' + str(jobid))
site = sites_.get(siteid)
job = jobs_.get(int(jobid))
cont = job.container
files_catalog = cont.files
fscope = getScope(job.owner.username)
datasetName = '{}:{}'.format(fscope, cont.guid)
distributive = job.distr.name
release = job.distr.release
# Prepare runScript
parameters = job.distr.command
parameters = parameters.replace("$COMMAND$", job.params)
parameters = parameters.replace("$USERNAME$", job.owner.username)
parameters = parameters.replace("$WORKINGGROUP$", job.owner.working_group)
# Prepare metadata
metadata = dict(user=job.owner.username)
# Prepare PanDA Object
pandajob = JobSpec()
pandajob.jobDefinitionID = int(time.time()) % 10000
pandajob.jobName = cont.guid
pandajob.transformation = client_config.DEFAULT_TRF
pandajob.destinationDBlock = datasetName
pandajob.destinationSE = site.se
pandajob.currentPriority = 1000
pandajob.prodSourceLabel = 'user'
pandajob.computingSite = site.ce
pandajob.cloud = 'RU'
pandajob.VO = 'atlas'
pandajob.prodDBlock = "%s:%s" % (fscope, pandajob.jobName)
pandajob.coreCount = job.corecount
pandajob.metadata = json.dumps(metadata)
#pandajob.workingGroup = job.owner.working_group
if site.encode_commands:
# It requires script wrapper on cluster side
pandajob.jobParameters = '%s %s %s "%s"' % (cont.guid, release,
distributive, parameters)
else:
pandajob.jobParameters = parameters
has_input = False
for fcc in files_catalog:
if fcc.type == 'input':
f = fcc.file
guid = f.guid
fileIT = FileSpec()
fileIT.lfn = f.lfn
fileIT.dataset = pandajob.prodDBlock
fileIT.prodDBlock = pandajob.prodDBlock
fileIT.type = 'input'
fileIT.scope = fscope
fileIT.status = 'ready'
fileIT.GUID = guid
pandajob.addFile(fileIT)
has_input = True
if fcc.type == 'output':
f = fcc.file
fileOT = FileSpec()
fileOT.lfn = f.lfn
fileOT.destinationDBlock = pandajob.prodDBlock
fileOT.destinationSE = pandajob.destinationSE
fileOT.dataset = pandajob.prodDBlock
fileOT.type = 'output'
fileOT.scope = fscope
fileOT.GUID = f.guid
pandajob.addFile(fileOT)
# Save replica meta
fc.new_replica(f, site)
if not has_input:
# Add fake input
fileIT = FileSpec()
fileIT.lfn = "fake.input"
fileIT.dataset = pandajob.prodDBlock
fileIT.prodDBlock = pandajob.prodDBlock
fileIT.type = 'input'
fileIT.scope = fscope
fileIT.status = 'ready'
fileIT.GUID = "fake.guid"
pandajob.addFile(fileIT)
# Prepare lof file
fileOL = FileSpec()
fileOL.lfn = "%s.log.tgz" % pandajob.jobName
fileOL.destinationDBlock = pandajob.destinationDBlock
fileOL.destinationSE = pandajob.destinationSE
fileOL.dataset = '{}:logs'.format(fscope)
fileOL.type = 'log'
fileOL.scope = 'panda'
pandajob.addFile(fileOL)
# Save log meta
log = File()
log.scope = fscope
log.lfn = fileOL.lfn
log.guid = getGUID(log.scope, log.lfn)
log.type = 'log'
log.status = 'defined'
files_.save(log)
# Save replica meta
fc.new_replica(log, site)
# Register file in container
fc.reg_file_in_cont(log, cont, 'log')
# Submit job
o = submitJobs([pandajob])
x = o[0]
try:
#update PandaID
PandaID = int(x[0])
job.pandaid = PandaID
job.ce = site.ce
except:
job.status = 'submit_error'
jobs_.save(job)
return 0
def new_job():
"""Creates new job
"""
g.user = request.oauth.user
scope = getScope(request.oauth.user.username)
js = request.json
data = js['data']
distr_id = data['sw_id']
params = data['script']
corecount = data['cores']
site = sites_.first(ce=current_app.config['DEFAULT_CE'])
distr = distrs_.get(id)
container = Container()
guid = 'job.' + commands.getoutput('uuidgen')
container.guid = guid
container.status = 'open'
conts_.save(container)
# Process ftp files
if 'ftp_dir' in data.keys():
ftp_dir = data['ftp_dir']
register_ftp_files(ftp_dir, scope, container.guid)
# Process guid list
if 'guids' in data.keys():
guids = data['guids']
for f in guids:
if f != '':
file_ = files_.first(guid=f)
if file_ is not None:
# Register file in catalog
fc.reg_file_in_cont(file_, container, 'input')
else:
raise WebpandaError('File with guid %s not found' % f)
ofiles = ['results.tgz']
# Starts cloneReplica tasks
ftasks = prepareInputFiles(container.id, site.se)
# Saves output files meta
for lfn in ofiles:
file = File()
file.scope = scope
file.guid = getGUID(scope, lfn)
file.lfn = lfn
file.status = 'defined'
files_.save(file)
# Register file in catalog
fc.reg_file_in_cont(file, container, 'output')
# Counts files
allfiles = container.files
nifiles = 0
nofiles = 0
for f in allfiles:
if f.type == 'input':
nifiles += 1
if f.type == 'output':
nofiles += 1
# Defines job meta
job = Job()
job.pandaid = None
job.status = 'pending'
job.owner = request.oauth.user
job.params = params
job.distr = distr
job.container = container
job.creation_time = datetime.utcnow()
job.modification_time = datetime.utcnow()
job.ninputfiles = nifiles
job.noutputfiles = nofiles
job.corecount = corecount
job.tags = data['tags'] if 'tags' in data.keys() else ""
jobs_.save(job)
# Async sendjob
res = chord(ftasks)(async_send_job.s(jobid=job.id, siteid=site.id))
return {'id': job.id, 'container_id': guid}
def upload():
form = request.form
# Create a unique container quid for this particular batch of uploads.
cguid = 'job.' + commands.getoutput('uuidgen')
# Is the upload using Ajax, or a direct POST by the form?
is_ajax = False
if form.get("__ajax", None) == "true":
is_ajax = True
# Create new container
container = Container()
container.guid = cguid
container.status = 'open'
conts_.save(container)
# Process files in request
for upload in request.files.getlist("file"):
# Define file params
lfn = upload.filename.rsplit("/")[0]
scope = getScope(g.user.username)
guid = getGUID(scope, lfn)
site = sites_.first(se=current_app.config['DEFAULT_SE'])
# Target folder for these uploads.
dir = '/' + os.path.join('system', scope, guid)
target = site.datadir + dir
try:
os.makedirs(target)
except:
if is_ajax:
return ajax_response(False, "Couldn't create upload directory: %s" % target)
else:
return "Couldn't create upload directory: %s" % target
replfn = os.path.join(dir, lfn)
destination = os.path.join(target, lfn)
upload.save(destination)
if os.path.isfile(destination):
# Check file existence in catalog
adler = adler32(destination)
md5 = md5sum(destination)
size = fsize(destination)
file_id = ddm_checkifexists(lfn, size, adler, md5)
if file_id:
# If file exists
file = files_.get(file_id)
else:
# Otherwise create new
file = File()
file.scope = scope
file.guid = guid
file.type = 'input'
file.lfn = lfn
file.token = ''
file.status = 'defined'
files_.save(file)
setFileMeta(file.id, destination)
replica = Replica()
replica.se = site.se
replica.status = 'ready'
replica.lfn = replfn
replica.original = file
replicas_.save(replica)
# Register file in container
fc.reg_file_in_cont(file, container, 'input')
else:
return ajax_response(False, "Couldn't save file: %s" % target)
if is_ajax:
return ajax_response(True, cguid)
else:
return redirect(url_for("jobs.jobs"))
def file_save(container_guid, lfn):
"""
POST: /pilot/file/<container_guid>/<lfn>/save
Saves file from request, returns file guid
:param container_guid: Guid of container
:type container_guid: str
:param lfn: Local FileName
:type lfn: str
:return: guid
:rtype: json
"""
site = sites_.first(se=current_app.config['DEFAULT_SE'])
if ':' in container_guid:
container_guid = container_guid.split(':')[-1]
container = conts_.first(guid=container_guid)
if container.status != 'open':
raise WebpandaError('Unable to upload: Container is not open')
cc = container.files
ff = None
for c in cc:
f = c.file
if f.lfn == lfn:
ff = f
if not ff:
ff = File()
ff.scope = getScope(g.user.username)
ff.lfn = lfn
ff.guid = getGUID(ff.scope, ff.lfn)
ff.status = 'defined'
files_.save(ff)
# Register file in container
fc.reg_file_in_cont(ff, container, 'input')
path = os.path.join(site.datadir, getScope(g.user.username), container.guid)
replfn = '/' + os.path.join(getScope(g.user.username), container.guid, ff.lfn)
destination = os.path.join(path, ff.lfn)
for r in ff.replicas:
if r.se == site.se:
destination = site.datadir + r.lfn
file_dir = '/'.join(destination.split('/')[:-1])
if r.status == 'ready':
if os.path.isfile(destination): # Check fsize, md5 or adler
raise WebpandaError('Replica exists')
else:
r.status = 'broken'
replicas_.save(r)
raise WebpandaError('Broken replica')
elif r.status == 'defined':
try:
os.makedirs(file_dir)
except(Exception):
pass
f = open(destination, 'wb')
f.write(request.data)
f.close()
# Update file info
setFileMeta(ff.id, destination)
r.status = 'ready'
replicas_.save(r)
return {'guid': ff.guid}
else:
raise WebpandaError('Replica status: %s' % r.status)
replica = Replica()
if os.path.isfile(destination):
raise WebpandaError('Unable to upload: File exists')
try:
os.makedirs(path)
except(Exception):
_logger.debug('Path exists: %s' % path)
f = open(destination, 'wb')
f.write(request.data)
f.close()
# Update file info
setFileMeta(ff.id, destination)
# Create/change replica
replica.se = site.se
replica.status = 'ready'
replica.lfn = replfn
replica.token = ''
replica.original = ff
replicas_.save(replica)
return {'guid': ff.guid}
def upload_dir(user_id, cont_id, se_id, path):
"""
Uploads files from external dir path into defined contained
:param cont_id: id of Container
:param se_id: id of SE
:param path: dir path on SE
:return:
"""
user = users_.get(user_id)
cont = conts_.get(cont_id)
se = sites_.get(se_id)
# Initialize SE connector
print "=Initialize SE connector"
_logger.debug("=Initialize SE connector")
conn_factory = SEFactory()
connector = conn_factory.getSE(se.plugin, None)
# Fetch list of files
print "=Fetch list of files"
_logger.debug("=Fetch list of files")
try:
list_of_lfn = connector.ls(path, rel=False)
for item in list_of_lfn:
# Check empty items
if item == "":
list_of_lfn.remove(item)
except:
raise WebpandaError("Unable to get list of files from SE: " +
str(se_id))
print "=" + str(len(list_of_lfn))
_logger.debug("=" + str(len(list_of_lfn)))
# Create list of File objs
print "=Create list of File objs"
_logger.debug("=Create list of File objs")
list_of_obj = list()
for item in list_of_lfn:
list_of_obj.append(fc.new_file(user, item))
print "=" + str(len(list_of_obj))
_logger.debug("=" + str(len(list_of_obj)))
# Iterate through files objects
print "=IterateLoop:Start"
_logger.debug("=IterateLoop:Start")
for item in list_of_obj:
# Add files to container:
print "=Add file to container"
_logger.debug("=Add file to container")
fc.reg_file_in_cont(item, cont, 'intermediate')
# Copy files into system dir
print "=Copy file into system dir"
_logger.debug("=Copy file into system dir")
connector.link(os.path.join(path, item.lfn),
fc.get_file_dir(item),
rel=False)
# Calculate fsize, adler32, md5hash
print "=Calculate fsize, adler32, md5hash"
_logger.debug("=Calculate fsize, adler32, md5hash")
item.fsize = connector.fsize(fc.get_file_path(item))
item.md5sum = connector.md5sum(fc.get_file_path(item))
item.checksum = connector.adler32(fc.get_file_path(item))
fc.save(item)
# Create list of Replica objs
print "=Create Replica object"
_logger.debug("=Create Replica object")
r = fc.new_replica(item, se)
r.status = 'ready'
fc.save(r)
# Update files' status
print "=Update files' status"
_logger.debug("=Update files' status")
item.status = 'ready'
fc.save(item)
print "=IterateLoop:Finish"
_logger.debug("=IterateLoop:Finish")
# Return container id
print "=Return container id"
_logger.debug("=Return container id")
return cont_id
def upload():
form = request.form
# Create a unique container quid for this particular batch of uploads.
cguid = 'job.' + commands.getoutput('uuidgen')
# Is the upload using Ajax, or a direct POST by the form?
is_ajax = False
if form.get("__ajax", None) == "true":
is_ajax = True
# Create new container
container = Container()
container.guid = cguid
container.status = 'open'
conts_.save(container)
# Process files in request
for upload in request.files.getlist("file"):
# Define file params
lfn = upload.filename.rsplit("/")[0]
scope = getScope(g.user.username)
guid = getGUID(scope, lfn)
site = sites_.first(se=current_app.config['DEFAULT_SE'])
# Target folder for these uploads.
dir = '/' + os.path.join('system', scope, guid)
target = site.datadir + dir
try:
os.makedirs(target)
except:
if is_ajax:
return ajax_response(
False, "Couldn't create upload directory: %s" % target)
else:
return "Couldn't create upload directory: %s" % target
replfn = os.path.join(dir, lfn)
destination = os.path.join(target, lfn)
upload.save(destination)
if os.path.isfile(destination):
# Check file existence in catalog
adler = adler32(destination)
md5 = md5sum(destination)
size = fsize(destination)
file_id = ddm_checkifexists(lfn, size, adler, md5)
if file_id:
# If file exists
file = files_.get(file_id)
else:
# Otherwise create new
file = File()
file.scope = scope
file.guid = guid
file.type = 'input'
file.lfn = lfn
file.token = ''
file.status = 'defined'
files_.save(file)
setFileMeta(file.id, destination)
replica = Replica()
replica.se = site.se
replica.status = 'ready'
replica.lfn = replfn
replica.original = file
replicas_.save(replica)
# Register file in container
fc.reg_file_in_cont(file, container, 'input')
else:
return ajax_response(False, "Couldn't save file: %s" % target)
if is_ajax:
return ajax_response(True, cguid)
else:
return redirect(url_for("jobs.jobs"))
def upload_dir(user_id, cont_id, se_id, path):
"""
Uploads files from external dir path into defined contained
:param cont_id: id of Container
:param se_id: id of SE
:param path: dir path on SE
:return:
"""
user = users_.get(user_id)
cont = conts_.get(cont_id)
se = sites_.get(se_id)
# Initialize SE connector
print "=Initialize SE connector"
_logger.debug("=Initialize SE connector")
conn_factory = SEFactory()
connector = conn_factory.getSE(se.plugin, None)
# Fetch list of files
print "=Fetch list of files"
_logger.debug("=Fetch list of files")
try:
list_of_lfn = connector.ls(path, rel=False)
for item in list_of_lfn:
# Check empty items
if item == "":
list_of_lfn.remove(item)
except:
raise WebpandaError("Unable to get list of files from SE: " + str(se_id))
print "=" + str(len(list_of_lfn))
_logger.debug("=" + str(len(list_of_lfn)))
# Create list of File objs
print "=Create list of File objs"
_logger.debug("=Create list of File objs")
list_of_obj = list()
for item in list_of_lfn:
list_of_obj.append(fc.new_file(user, item))
print "=" + str(len(list_of_obj))
_logger.debug("=" + str(len(list_of_obj)))
# Iterate through files objects
print "=IterateLoop:Start"
_logger.debug("=IterateLoop:Start")
for item in list_of_obj:
# Add files to container:
print "=Add file to container"
_logger.debug("=Add file to container")
fc.reg_file_in_cont(item, cont, 'intermediate')
# Copy files into system dir
print "=Copy file into system dir"
_logger.debug("=Copy file into system dir")
connector.link(os.path.join(path, item.lfn), fc.get_file_dir(item), rel=False)
# Calculate fsize, adler32, md5hash
print "=Calculate fsize, adler32, md5hash"
_logger.debug("=Calculate fsize, adler32, md5hash")
item.fsize = connector.fsize(fc.get_file_path(item))
item.md5sum = connector.md5sum(fc.get_file_path(item))
item.checksum = connector.adler32(fc.get_file_path(item))
fc.save(item)
# Create list of Replica objs
print "=Create Replica object"
_logger.debug("=Create Replica object")
r = fc.new_replica(item, se)
r.status = 'ready'
fc.save(r)
# Update files' status
print "=Update files' status"
_logger.debug("=Update files' status")
item.status = 'ready'
fc.save(item)
print "=IterateLoop:Finish"
_logger.debug("=IterateLoop:Finish")
# Return container id
print "=Return container id"
_logger.debug("=Return container id")
return cont_id
def registerLocalFile(arg, dirname, names, scope):
"""Register files from local dir to container
:param arg: Container guid
:param dirname: Abs dir
:param names: File name
:param scope: Scope to upload files in
:return:
"""
site = sites_.first(se=client_config.DEFAULT_SE)
_logger.debug(str(arg))
cont = conts_.first(guid=arg)
files = cont.files
for name in names:
fpath = os.path.join(dirname, name)
fobj = None
# Check in container
for file in files:
if file.lfn == name:
fobj = file
# Check in catalog
if not fobj:
destination = os.path.join(dirname, name)
adler = adler32(destination)
md5 = md5sum(destination)
size = fsize(destination)
file_id = ddm_checkifexists(name, size, adler, md5)
if file_id:
# If file exists
fobj = files_.get(file_id)
if not fobj:
fobj = File()
fobj.scope = scope
fobj.lfn = name
fobj.guid = getGUID(fobj.scope, fobj.lfn)
fobj.type = 'input'
fobj.status = 'defined'
files_.save(fobj)
setFileMeta(fobj.id, fpath)
# Register file in catalog
fc.reg_file_in_cont(fobj, cont, "input")
replicas = fobj.replicas
replica = None
for r in replicas:
if r.se == site.se and r.status == 'ready':
replica = r
if not replica:
ldir = '/' + os.path.join('system', fobj.scope, fobj.guid)
ddm_localmakedirs(ldir)
ddm_localcp(fpath[len(site.datadir):], ldir)
replica = Replica()
replica.se = site.se
replica.status = 'ready'
replica.token = ''
replica.lfn = os.path.join(ldir, fobj.lfn)
replica.original = fobj
replicas_.save(replica)
def job():
"""
New job form view
:return: Response obj
"""
form = NewJobForm(request.form)
if request.method == 'POST':
site = sites_.get(int(form.site.data))
distr_name, distr_release = form.distr.data.split(':')
distr = distrs_.first(name=distr_name, release=int(distr_release))
container_guid = form.container.data
try:
container = conts_.first(guid=container_guid)
except(Exception):
_logger.error(Exception.message)
return make_response(jsonify({'error': 'Container not found'}), 404)
if site.encode_commands:
# By default frontend encodes with base64 job script parts separated by ";"
# It requires script wrapper on cluster side
jparams = form.params.data
else:
# Set site.encode_commands as False if you want to send command string without base64 encoding
jparams = ';'.join([b64decode(command) for command in form.params.data.split(';')])
ifiles = request.form.getlist('ifiles[]')
iguids = request.form.getlist('iguids[]')
iconts = request.form.getlist('iconts[]')
ofiles = ['{guid}.out.tgz'.format(guid=container.guid)]
scope = getScope(g.user.username)
# Process ftp files
ftp_dir = form.ftpdir.data
register_ftp_files(ftp_dir, scope, container.guid)
# Process guid list
for f in iguids:
if f != '':
file = files_.first(guid=f)
if file is not None:
# Register files in container
fc.reg_file_in_cont(file, container, 'input')
else:
return make_response(jsonify({'error': "GUID {} not found".format(f)}))
# Process containers
for c in iconts:
if c != '':
try:
form_cont = conts_.first(guid=c)
except(Exception):
_logger.error(Exception.message)
return make_response(jsonify({'error': 'Container in form not found'}), 404)
for f in form_cont.files:
# Register file in catalog
fc.reg_file_in_cont(f.file, container, 'input')
# Processes urls
for f in ifiles:
if f != '':
from_se, path, token = getUrlInfo(f)
replfn = ':/'.join([from_se, path])
# Check if used before
file_id = ddm_checkexternalifexists('', replfn)
if file_id:
file = files_.get(file_id)
else:
lfn = path.split('/')[-1]
guid = getGUID(scope, lfn)
file = File()
file.scope = scope
file.guid = guid
file.type = 'input'
file.lfn = lfn
file.status = 'defined'
files_.save(file)
replica = Replica()
replica.se = from_se
replica.status = 'link'
# Separate url & token
replica.lfn = replfn
replica.token = token
replica.original = file
replicas_.save(replica)
# Register file in container
fc.reg_file_in_cont(file, container, 'input')
# Starts cloneReplica tasks
ftasks = prepareInputFiles(container.id, site.se)
# Saves output files meta
for lfn in ofiles:
file = File()
file.scope = scope
file.guid = getGUID(scope, lfn)
file.type = 'output'
file.lfn = lfn
file.status = 'defined'
files_.save(file)
# Register file in container
fc.reg_file_in_cont(file, container, 'output')
# Counts files
allfiles = container.files
nifiles = 0
nofiles = 0
for f in allfiles:
if f.type == 'input':
nifiles += 1
if f.type == 'output':
nofiles += 1
# Defines job meta
job = Job()
job.pandaid = None
job.status = 'pending'
job.owner = g.user
job.params = jparams
job.distr = distr
job.container = container
job.creation_time = datetime.utcnow()
job.modification_time = datetime.utcnow()
job.ninputfiles = nifiles
job.noutputfiles = nofiles
job.corecount = form.corecount.data
job.tags = form.tags.data if form.tags.data != "" else None
jobs_.save(job)
# Async sendjob
res = chord(ftasks)(async_send_job.s(jobid=job.id, siteid=site.id))
return redirect(url_for('jobs.jobs'))
form.distr.choices = [("%s:%s" % (distr.name, distr.release), "%s: %s" % (distr.name, distr.version)) for distr in distrs_.find().order_by('name').order_by('version')]
form.site.choices = [(site.id, "{ce}".format(ce=site.ce)) for site in sites_.find(active=1).order_by('ce')]
return render_template("dashboard/jobs/new.html", form=form)
def send_job(jobid, siteid):
_logger.debug('Jobid: ' + str(jobid))
site = sites_.get(siteid)
job = jobs_.get(int(jobid))
cont = job.container
files_catalog = cont.files
fscope = getScope(job.owner.username)
datasetName = '{}:{}'.format(fscope, cont.guid)
distributive = job.distr.name
release = job.distr.release
# Prepare runScript
parameters = job.distr.command
parameters = parameters.replace("$COMMAND$", job.params)
parameters = parameters.replace("$USERNAME$", job.owner.username)
parameters = parameters.replace("$WORKINGGROUP$", job.owner.working_group)
# Prepare metadata
metadata = dict(user=job.owner.username)
# Prepare PanDA Object
pandajob = JobSpec()
pandajob.jobDefinitionID = int(time.time()) % 10000
pandajob.jobName = cont.guid
pandajob.transformation = client_config.DEFAULT_TRF
pandajob.destinationDBlock = datasetName
pandajob.destinationSE = site.se
pandajob.currentPriority = 1000
pandajob.prodSourceLabel = 'user'
pandajob.computingSite = site.ce
pandajob.cloud = 'RU'
pandajob.VO = 'atlas'
pandajob.prodDBlock = "%s:%s" % (fscope, pandajob.jobName)
pandajob.coreCount = job.corecount
pandajob.metadata = json.dumps(metadata)
#pandajob.workingGroup = job.owner.working_group
if site.encode_commands:
# It requires script wrapper on cluster side
pandajob.jobParameters = '%s %s %s "%s"' % (cont.guid, release, distributive, parameters)
else:
pandajob.jobParameters = parameters
has_input = False
for fcc in files_catalog:
if fcc.type == 'input':
f = fcc.file
guid = f.guid
fileIT = FileSpec()
fileIT.lfn = f.lfn
fileIT.dataset = pandajob.prodDBlock
fileIT.prodDBlock = pandajob.prodDBlock
fileIT.type = 'input'
fileIT.scope = fscope
fileIT.status = 'ready'
fileIT.GUID = guid
pandajob.addFile(fileIT)
has_input = True
if fcc.type == 'output':
f = fcc.file
fileOT = FileSpec()
fileOT.lfn = f.lfn
fileOT.destinationDBlock = pandajob.prodDBlock
fileOT.destinationSE = pandajob.destinationSE
fileOT.dataset = pandajob.prodDBlock
fileOT.type = 'output'
fileOT.scope = fscope
fileOT.GUID = f.guid
pandajob.addFile(fileOT)
# Save replica meta
fc.new_replica(f, site)
if not has_input:
# Add fake input
fileIT = FileSpec()
fileIT.lfn = "fake.input"
fileIT.dataset = pandajob.prodDBlock
fileIT.prodDBlock = pandajob.prodDBlock
fileIT.type = 'input'
fileIT.scope = fscope
fileIT.status = 'ready'
fileIT.GUID = "fake.guid"
pandajob.addFile(fileIT)
# Prepare lof file
fileOL = FileSpec()
fileOL.lfn = "%s.log.tgz" % pandajob.jobName
fileOL.destinationDBlock = pandajob.destinationDBlock
fileOL.destinationSE = pandajob.destinationSE
fileOL.dataset = '{}:logs'.format(fscope)
fileOL.type = 'log'
fileOL.scope = 'panda'
pandajob.addFile(fileOL)
# Save log meta
log = File()
log.scope = fscope
log.lfn = fileOL.lfn
log.guid = getGUID(log.scope, log.lfn)
log.type = 'log'
log.status = 'defined'
files_.save(log)
# Save replica meta
fc.new_replica(log, site)
# Register file in container
fc.reg_file_in_cont(log, cont, 'log')
# Submit job
o = submitJobs([pandajob])
x = o[0]
try:
#update PandaID
PandaID = int(x[0])
job.pandaid = PandaID
job.ce = site.ce
except:
job.status = 'submit_error'
jobs_.save(job)
return 0
def new_job():
"""Creates new job
"""
g.user = request.oauth.user
scope = getScope(request.oauth.user.username)
js = request.json
data = js['data']
distr_id = data['sw_id']
params = data['script']
corecount = data['cores']
site = sites_.first(ce=current_app.config['DEFAULT_CE'])
distr = distrs_.get(id)
container = Container()
guid = 'job.' + commands.getoutput('uuidgen')
container.guid = guid
container.status = 'open'
conts_.save(container)
# Process ftp files
if 'ftp_dir' in data.keys():
ftp_dir = data['ftp_dir']
register_ftp_files(ftp_dir, scope, container.guid)
# Process guid list
if 'guids' in data.keys():
guids = data['guids']
for f in guids:
if f != '':
file_ = files_.first(guid=f)
if file_ is not None:
# Register file in catalog
fc.reg_file_in_cont(file_, container, 'input')
else:
raise WebpandaError('File with guid %s not found' % f)
ofiles = ['results.tgz']
# Starts cloneReplica tasks
ftasks = prepareInputFiles(container.id, site.se)
# Saves output files meta
for lfn in ofiles:
file = File()
file.scope = scope
file.guid = getGUID(scope, lfn)
file.lfn = lfn
file.status = 'defined'
files_.save(file)
# Register file in catalog
fc.reg_file_in_cont(file, container, 'output')
# Counts files
allfiles = container.files
nifiles = 0
nofiles = 0
for f in allfiles:
if f.type == 'input':
nifiles += 1
if f.type == 'output':
nofiles += 1
# Defines job meta
job = Job()
job.pandaid = None
job.status = 'pending'
job.owner = request.oauth.user
job.params = params
job.distr = distr
job.container = container
job.creation_time = datetime.utcnow()
job.modification_time = datetime.utcnow()
job.ninputfiles = nifiles
job.noutputfiles = nofiles
job.corecount = corecount
job.tags = data['tags'] if 'tags' in data.keys() else ""
jobs_.save(job)
# Async sendjob
res = chord(ftasks)(async_send_job.s(jobid=job.id, siteid=site.id))
return {'id': job.id, 'container_id': guid}
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 file_save(container_guid, lfn):
"""
POST: /pilot/file/<container_guid>/<lfn>/save
Saves file from request, returns file guid
:param container_guid: Guid of container
:type container_guid: str
:param lfn: Local FileName
:type lfn: str
:return: guid
:rtype: json
"""
site = sites_.first(se=current_app.config['DEFAULT_SE'])
if ':' in container_guid:
container_guid = container_guid.split(':')[-1]
container = conts_.first(guid=container_guid)
if container.status != 'open':
raise WebpandaError('Unable to upload: Container is not open')
cc = container.files
ff = None
for c in cc:
f = c.file
if f.lfn == lfn:
ff = f
if not ff:
ff = File()
ff.scope = getScope(g.user.username)
ff.lfn = lfn
ff.guid = getGUID(ff.scope, ff.lfn)
ff.status = 'defined'
files_.save(ff)
# Register file in container
fc.reg_file_in_cont(ff, container, 'input')
path = os.path.join(site.datadir, getScope(g.user.username),
container.guid)
replfn = '/' + os.path.join(getScope(g.user.username), container.guid,
ff.lfn)
destination = os.path.join(path, ff.lfn)
for r in ff.replicas:
if r.se == site.se:
destination = site.datadir + r.lfn
file_dir = '/'.join(destination.split('/')[:-1])
if r.status == 'ready':
if os.path.isfile(destination): # Check fsize, md5 or adler
raise WebpandaError('Replica exists')
else:
r.status = 'broken'
replicas_.save(r)
raise WebpandaError('Broken replica')
elif r.status == 'defined':
try:
os.makedirs(file_dir)
except (Exception):
pass
f = open(destination, 'wb')
f.write(request.data)
f.close()
# Update file info
setFileMeta(ff.id, destination)
r.status = 'ready'
replicas_.save(r)
return {'guid': ff.guid}
else:
raise WebpandaError('Replica status: %s' % r.status)
replica = Replica()
if os.path.isfile(destination):
raise WebpandaError('Unable to upload: File exists')
try:
os.makedirs(path)
except (Exception):
_logger.debug('Path exists: %s' % path)
f = open(destination, 'wb')
f.write(request.data)
f.close()
# Update file info
setFileMeta(ff.id, destination)
# Create/change replica
replica.se = site.se
replica.status = 'ready'
replica.lfn = replfn
replica.token = ''
replica.original = ff
replicas_.save(replica)
return {'guid': ff.guid}
def registerLocalFile(arg, dirname, names, scope):
"""Register files from local dir to container
:param arg: Container guid
:param dirname: Abs dir
:param names: File name
:param scope: Scope to upload files in
:return:
"""
site = sites_.first(se=client_config.DEFAULT_SE)
_logger.debug(str(arg))
cont = conts_.first(guid=arg)
files = cont.files
for name in names:
fpath = os.path.join(dirname, name)
fobj = None
# Check in container
for file in files:
if file.lfn == name:
fobj = file
# Check in catalog
if not fobj:
destination = os.path.join(dirname, name)
adler = adler32(destination)
md5 = md5sum(destination)
size = fsize(destination)
file_id = ddm_checkifexists(name, size, adler, md5)
if file_id:
# If file exists
fobj = files_.get(file_id)
if not fobj:
fobj = File()
fobj.scope = scope
fobj.lfn = name
fobj.guid = getGUID(fobj.scope, fobj.lfn)
fobj.type = 'input'
fobj.status = 'defined'
files_.save(fobj)
setFileMeta(fobj.id, fpath)
# Register file in catalog
fc.reg_file_in_cont(fobj, cont, "input")
replicas = fobj.replicas
replica = None
for r in replicas:
if r.se == site.se and r.status == 'ready':
replica = r
if not replica:
ldir = '/' + os.path.join('system', fobj.scope, fobj.guid)
ddm_localmakedirs(ldir)
ddm_localcp(fpath[len(site.datadir):], ldir)
replica = Replica()
replica.se = site.se
replica.status = 'ready'
replica.token = ''
replica.lfn = os.path.join(ldir, fobj.lfn)
replica.original = fobj
replicas_.save(replica)