hadcall = "hadoop fs -cat %s | gunzip | %s > %s" % (filename, javacall,
batchfile)
#print hadcall
os.system(hadcall)
if __name__ == "__main__":
if not len(sys.argv) == 3:
print "Usage BatchSort.py daycode a/b/c"
sys.exit(1)
daycode = sys.argv[1]
if "yest" == daycode:
daycode = SynchUtil.get_yesterday()
(proc_id, num_proc, num_batch) = [int(x) for x in sys.argv[2].split("/")]
if int(proc_id) == 0:
setupManiFile(daycode)
# TODO: don't use temp file, instead hadoop cat into another program
# that does the scrubbing, and direct output to the batch file.
print "proc=%d, num=%d, nbatch=%d" % (proc_id, num_proc, num_batch)
manimap = getManiData(daycode)
# TODO: prefilter files so that we can print "finished with file 52/652..."
targlist = [
onefile for onefile in manimap
if ((hash(onefile) % num_batch) % num_proc) == proc_id
#!/usr/bin/python
###################################################
# Just a wrapper for the two SliceInterest jobs
# Want these to run back to back, hard to do without a wrapper
###################################################
import os, sys
import SynchUtil
def runCall(dmclass, daycode):
hadcall = "hadoop jar /mnt/jars/adnetik.jar com.adnetik.data_management.%s %s" % (dmclass, daycode)
print "\nHadoop call is : \n\t%s" % (hadcall)
os.system(hadcall)
print "\nFinished with %s" % (dmclass)
if __name__ == "__main__":
# this is ALWAYS going to run using "yesterday"
daycode = SynchUtil.get_yesterday()
runCall("SliceInterestActivity", daycode)
runCall("SliceInterestSecond", daycode)
os.system(hadcall)
print "... done"
if __name__ == "__main__":
daylist = []
if len(sys.argv) < 2:
print "Usage UploadExelate.py <daycode|daylist.txt>"
sys.exit(1)
singarg = sys.argv[1]
if singarg.endswith(".txt"):
gimp = [daylist.append(oneday.strip()) for oneday in open(singarg)]
elif singarg == "yest":
daylist.append(SynchUtil.get_yesterday())
else:
daylist.append(singarg)
for daycode in daylist:
grabExelateFile(daycode)
upload2hdfs(daycode)
continue # no point in continuing
subhadls = "hadoop fs -ls %s" % (dirpath)
subfiles = SynchUtil.sysCallResult(subhadls)
if len(subfiles) < FILE_COUNT_WARN_CUTOFF:
logmail.addLogLine("Warning: found only %d files for exchange %s" %
(len(subfiles), excname))
warncount += 1
return warncount
if __name__ == "__main__":
"""
Check to make sure that the daily Synch Jobs ran correctly.
If they didn't, send an AdminMail
"""
daycode = SynchUtil.get_yesterday(
) if "yest" in sys.argv[1] else sys.argv[1]
logmail = SimpleMail.SimpleMail("File Synch Check for %s" % (daycode))
warncount = 0
warncount += checkExchangeDirs(logmail, "no_bid", daycode)
warncount += checkExchangeDirs(logmail, "bid_all", daycode)
if warncount > 0:
logmail.send2admin()
subhadls = "hadoop fs -ls %s" % (dirpath)
subfiles = SynchUtil.sysCallResult(subhadls)
if len(subfiles) < FILE_COUNT_WARN_CUTOFF:
logmail.addLogLine("Warning: found only %d files for exchange %s" % (len(subfiles), excname))
warncount += 1
return warncount
if __name__ == "__main__":
"""
Check to make sure that the daily Synch Jobs ran correctly.
If they didn't, send an AdminMail
"""
daycode = SynchUtil.get_yesterday() if "yest" in sys.argv[1] else sys.argv[1]
logmail = SimpleMail.SimpleMail("File Synch Check for %s" % (daycode))
warncount = 0
warncount += checkExchangeDirs(logmail, "no_bid", daycode)
warncount += checkExchangeDirs(logmail, "bid_all", daycode)
if warncount > 0:
logmail.send2admin()
print "... done"
if __name__ == "__main__":
daylist = []
if len(sys.argv) < 2:
print "Usage UploadExelate.py <daycode|daylist.txt>"
sys.exit(1)
singarg = sys.argv[1]
if singarg.endswith(".txt"):
gimp = [daylist.append(oneday.strip()) for oneday in open(singarg)]
elif singarg == "yest":
daylist.append(SynchUtil.get_yesterday())
else:
daylist.append(singarg)
for daycode in daylist:
grabExelateFile(daycode)
upload2hdfs(daycode)
#pc_set = set()
#for x in targlist:
# pc_set.add(x)
# pc_set.add(pix_comp_map[x])
#
#for onepix in pc_set: