#!/usr/bin/env python
+import logging
+logging.basicConfig(level=logging.INFO, format="run-command: %(message)s")
+
import arvados
import re
import os
import subprocess
import sys
import shutil
-import subst
+import crunchutil.subst as subst
import time
import arvados.commands.put as put
import signal
import traceback
import pprint
import multiprocessing
-import logging
+import crunchutil.robust_put as robust_put
+import crunchutil.vwd as vwd
os.umask(0077)
-logging.basicConfig(format="run-command: %(message)s")
t = arvados.current_task().tmpdir
links = []
-def sub_link(v):
- r = os.path.join(outdir, os.path.basename(v))
- os.symlink(v, r)
- links.append(r)
- return r
-
def sub_tmpdir(v):
return os.path.join(arvados.current_task().tmpdir, 'tmpdir')
def sub_jobsrc(v):
return os.environ['CRUNCH_SRC']
-subst.default_subs["link "] = sub_link
subst.default_subs["task.tmpdir"] = sub_tmpdir
subst.default_subs["task.outdir"] = sub_outdir
subst.default_subs["job.srcdir"] = sub_jobsrc
subst.default_subs["job.uuid"] = sub_jobid
subst.default_subs["task.uuid"] = sub_taskid
-def machine_progress(bytes_written, bytes_expected):
- return "run-command: wrote {} total {}\n".format(
- bytes_written, -1 if (bytes_expected is None) else bytes_expected)
-
class SigHandler(object):
def __init__(self):
self.sig = None
fn = subst.do_substitution(p, value)
mode = os.stat(fn).st_mode
prefix = fn[len(os.environ['TASK_KEEPMOUNT'])+1:]
- if mode != None:
+ if mode is not None:
if stat.S_ISDIR(mode):
items = ["$(dir %s/%s/)" % (prefix, l) for l in os.listdir(fn)]
elif stat.S_ISREG(mode):
stdoutname = None
stdoutfile = None
+stdinname = None
+stdinfile = None
rcode = 1
try:
var = jobp["task.foreach"]
items = get_items(jobp, jobp[var])
logging.info("parallelizing on %s with items %s" % (var, items))
- if items != None:
+ if items is not None:
for i in items:
params = copy.copy(jobp)
params[var] = i
'parameters': params
}
).execute()
- arvados.current_task().set_output(None)
+ if "task.vwd" in jobp:
+ # Base vwd collection will be merged with output fragments from
+ # the other tasks by crunch.
+ arvados.current_task().set_output(subst.do_substitution(jobp, jobp["task.vwd"]))
+ else:
+ arvados.current_task().set_output(None)
sys.exit(0)
else:
sys.exit(1)
else:
taskp = jobp
+ if "task.vwd" in taskp:
+ # Populate output directory with symlinks to files in collection
+ vwd.checkout(subst.do_substitution(taskp, taskp["task.vwd"]), outdir)
+
+ if "task.cwd" in taskp:
+ os.chdir(subst.do_substitution(taskp, taskp["task.cwd"]))
+
cmd = expand_list(taskp, taskp["command"])
- if "save.stdout" in taskp:
- stdoutname = subst.do_substitution(taskp, taskp["save.stdout"])
+ if "task.stdin" in taskp:
+ stdinname = subst.do_substitution(taskp, taskp["task.stdin"])
+ stdinfile = open(stdinname, "rb")
+
+ if "task.stdout" in taskp:
+ stdoutname = subst.do_substitution(taskp, taskp["task.stdout"])
stdoutfile = open(stdoutname, "wb")
- logging.info("{}{}".format(' '.join(cmd), (" > " + stdoutname) if stdoutname != None else ""))
+ logging.info("{}{}{}".format(' '.join(cmd), (" < " + stdinname) if stdinname is not None else "", (" > " + stdoutname) if stdoutname is not None else ""))
except Exception as e:
logging.exception("caught exception")
sys.exit(1)
try:
- sp = subprocess.Popen(cmd, shell=False, stdout=stdoutfile)
+ sp = subprocess.Popen(cmd, shell=False, stdin=stdinfile, stdout=stdoutfile)
sig = SigHandler()
# forward signals to the process.
# wait for process to complete.
rcode = sp.wait()
- if sig.sig != None:
+ if sig.sig is not None:
logging.critical("terminating on signal %s" % sig.sig)
sys.exit(2)
else:
logging.info("start writing output to keep")
-done = False
-resume_cache = put.ResumeCache(os.path.join(arvados.current_task().tmpdir, "upload-output-checkpoint"))
-reporter = put.progress_writer(machine_progress)
-bytes_expected = put.expected_bytes_for(".")
-while not done:
- try:
- out = put.ArvPutCollectionWriter.from_cache(resume_cache, reporter, bytes_expected)
- out.do_queued_work()
- out.write_directory_tree(".", max_manifest_depth=0)
- outuuid = out.finish()
- api.job_tasks().update(uuid=arvados.current_task()['uuid'],
- body={
- 'output':outuuid,
- 'success': (rcode == 0),
- 'progress':1.0
- }).execute()
- done = True
- except KeyboardInterrupt:
- logging.critical("terminating on signal 2")
- sys.exit(2)
- except Exception as e:
- logging.exception("caught exception:")
- time.sleep(5)
+if "task.vwd" in taskp:
+ if "task.foreach" in jobp:
+ # This is a subtask, so don't merge with the original collection, that will happen at the end
+ outcollection = vwd.checkin(subst.do_substitution(taskp, taskp["task.vwd"]), outdir, merge=False).manifest_text()
+ else:
+ # Just a single task, so do merge with the original collection
+ outcollection = vwd.checkin(subst.do_substitution(taskp, taskp["task.vwd"]), outdir, merge=True).manifest_text()
+else:
+ outcollection = robust_put.upload(outdir)
+
+api.job_tasks().update(uuid=arvados.current_task()['uuid'],
+ body={
+ 'output': outcollection,
+ 'success': (rcode == 0),
+ 'progress':1.0
+ }).execute()
sys.exit(rcode)
projects / arvados.git / blobdiff