def migrate(muri, dbname, huri): "Migrate data from MongoDB (muri) to HDFS (huri)" mstg = MongoStorage(muri, dbname) hstg = HdfsStorage(huri) # read data from MongoDB query = {'stype': mstg.stype} mdocs = mstg.read(query) mids = [d['wmaid'] for d in mdocs] # do nothing if no documents is found if not len(mdocs): return # store data to HDFS wmaid = hstg.write(mdocs) # read data from HDFS hdocs = hstg.read(wmaid) # now we can compare MongoDB docs with HDFS docs, a la cross-check for mdoc, hdoc in zip(mdocs, hdocs): # drop WMArchive keys for key in ['stype', 'wmaid']: if key in mdoc: del mdoc[key] if key in hdoc: del hdoc[key] if mdoc != hdoc: print("ERROR", mdoc, hdoc) sys.exit(1) # update status attributes of docs in MongoDB query = {'$set': {'stype': hstg.stype}} mstg.update(mids, query)
def migrate(muri, huri): "Migrate data from MongoDB (muri) to HDFS (huri)" mstg = MongoStorage(muri) hstg = HdfsStorage(huri) # read data from MongoDB query = {'stype': mstg.stype} mdocs = mstg.read(query) mids = [d['wmaid'] for d in mdocs] # do nothing if no documents is found if not len(mdocs): return # store data to HDFS wmaid = hstg.write(mdocs) # read data from HDFS hdocs = hstg.read(wmaid) # now we can compare MongoDB docs with HDFS docs, a la cross-check for mdoc, hdoc in zip(mdocs, hdocs): # drop WMArchive keys for key in ['stype', 'wmaid']: if key in mdoc: del mdoc[key] if key in hdoc: del hdoc[key] if mdoc != hdoc: print("ERROR", mdoc, hdoc) sys.exit(1) # update status attributes of docs in MongoDB query = {'$set' : {'stype': hstg.stype}} mstg.update(mids, query)
def migrate(muri, dbname, odir, mdir, avsc, thr, compress, chunk, close2midnight, dtype): "Write data from MongoDB (muri) to avro file(s) on local file system" mstg = MongoStorage(muri, dbname) auri = avsc if avsc.startswith('avroio:') else 'avroio:%s' % avsc astg = AvroStorage(auri) # read data from MongoDB for given storage and document types query = {'stype': mstg.stype, 'dtype': dtype} mdocs = mstg.find(query, None) # with no fields we'll get entire docs # loop over provided docs and write them into avro file on local file system wmaids = [] total = 0 fsize = 0 fname = file_name(odir, mdir, thr, compress, close2midnight) while True: data = [r for r in itertools.islice(mdocs, chunk)] total += len(data) if not len(data): break ids = astg.file_write(fname, data) if os.path.isfile(fname): fsize = os.path.getsize(fname) wmaids += ids if ids: # update status attributes of docs in MongoDB spec = {'$set': {'stype': astg.stype}} mstg.update(ids, spec) try: if PSUTIL: pid = os.getpid() proc = psutil.Process(pid) mem = proc.memory_info_ex() rss = 'RSS:%s' % size_format(mem.rss) else: rss = '' except: rss = '' print(tstamp('mongo2avro'), "%s docs %s %s (%s bytes) %s" \ % (len(ids), fname, size_format(fsize), fsize, rss)) fname = file_name(odir, mdir, thr, compress, close2midnight) print(tstamp('mongo2avro'), "wrote %s docs out of %s" % (len(wmaids), total))
def migrate(muri, odir, mdir, avsc, thr, compress, chunk, close2midnight): "Write data from MongoDB (muri) to avro file(s) on local file system" mstg = MongoStorage(muri) auri = avsc if avsc.startswith('avroio:') else 'avroio:%s' % avsc astg = AvroStorage(auri) # read data from MongoDB, returned mdocs is generator type query = {'stype': mstg.stype} mdocs = mstg.find(query, None) # with no fields we'll get entire docs # loop over provided docs and write them into avro file on local file system wmaids = [] total = 0 fsize = 0 fname = file_name(odir, mdir, thr, compress, close2midnight) while True: data = [r for r in itertools.islice(mdocs, chunk)] total += len(data) if not len(data): break ids = astg.file_write(fname, data) if os.path.isfile(fname): fsize = os.path.getsize(fname) wmaids += ids if ids: # update status attributes of docs in MongoDB spec = {'$set' : {'stype': astg.stype}} mstg.update(ids, spec) try: if PSUTIL: pid = os.getpid() proc = psutil.Process(pid) mem = proc.memory_info_ex() rss = 'RSS:%s' % size_format(mem.rss) else: rss = '' except: rss = '' print(tstamp('mongo2avro'), "%s docs %s %s (%s bytes) %s" \ % (len(ids), fname, size_format(fsize), fsize, rss)) fname = file_name(odir, mdir, thr, compress, close2midnight) print(tstamp('mongo2avro'), "wrote %s docs out of %s" % (len(wmaids), total))