def init(): global HTTP_PORT, PARAMS_FILE, EXT_MAP, EXT_REDUCE,\ PART_SUFFIX, MAP_OUTPUT, REDUCE_DL,\ REDUCE_SORTED, REDUCE_OUTPUT, OOB_FILE,\ JOB_HOME, DISCO_ROOT, JOB_ROOT, PART_OUTPUT,\ MAP_INDEX, REDUCE_INDEX, REQ_FILES, CHDIR_PATH tmp, HTTP_PORT, DISCO_ROOT = load_conf() job_name = this_name() JOB_HOME = "%s/%s/%s/" %\ (this_host(), md5.md5(job_name).hexdigest()[:2], job_name) JOB_ROOT = "%s/data/%s" % (DISCO_ROOT, JOB_HOME) if resultfs_enabled: pp = "%s/temp/%s" % (DISCO_ROOT, JOB_HOME) else: pp = JOB_ROOT CHDIR_PATH = pp PARAMS_FILE = pp + "params.dl" REQ_FILES = pp + "lib" EXT_MAP = pp + "ext.map" EXT_REDUCE = pp + "ext.reduce" PART_SUFFIX = "-%.9d" MAP_OUTPUT = pp + "map-disco-%d" + PART_SUFFIX PART_OUTPUT = pp + "part-disco-%.9d" REDUCE_DL = pp + "reduce-in-%d.dl" REDUCE_SORTED = pp + "reduce-in-%d.sorted" REDUCE_OUTPUT = pp + "reduce-disco-%d" OOB_FILE = pp + "oob/%s" MAP_INDEX = pp + "map-index.txt" REDUCE_INDEX = pp + "reduce-index.txt"
def result_iterator(results, notifier = None,\ proxy = None, reader = func.netstr_reader): res = [] for dir_url in results: if dir_url.startswith("dir://"): res += util.parse_dir(dir_url, proxy) else: res.append(dir_url) x, x, root = util.load_conf() for url in res: if url.startswith("file://"): fname = url[7:] fd = file(fname) sze = os.stat(fname).st_size elif url.startswith("disco://"): host, fname = url[8:].split("/", 1) url = util.proxy_url(proxy, fname, host) if util.resultfs_enabled: f = "%s/data/%s" % (root, fname) fd = file(f) sze = os.stat(f).st_size else: sze, fd = comm.open_remote(url) else: raise JobException("Invalid result url: %s" % url) if notifier: notifier(url) for x in reader(fd, sze, fname): yield x
def init(): global HTTP_PORT, LOCAL_PATH, PARAMS_FILE, EXT_MAP, EXT_REDUCE,\ MAP_OUTPUT, CHUNK_OUTPUT, REDUCE_DL, REDUCE_SORTED, REDUCE_OUTPUT tmp, HTTP_PORT, LOCAL_PATH = load_conf() PARAMS_FILE = LOCAL_PATH + "%s/params" EXT_MAP = LOCAL_PATH + "%s/ext-map" EXT_REDUCE = LOCAL_PATH + "%s/ext-reduce" MAP_OUTPUT = LOCAL_PATH + "%s/map-disco-%d-%.9d" CHUNK_OUTPUT = LOCAL_PATH + "%s/map-chunk-%d" REDUCE_DL = LOCAL_PATH + "%s/reduce-in-%d.dl" REDUCE_SORTED = LOCAL_PATH + "%s/reduce-in-%d.sorted" REDUCE_OUTPUT = LOCAL_PATH + "%s/reduce-disco-%d"
def init(): global HTTP_PORT, LOCAL_PATH, PARAMS_FILE, EXT_MAP, EXT_REDUCE,\ PART_SUFFIX, MAP_OUTPUT, CHUNK_OUTPUT, REDUCE_DL,\ REDUCE_SORTED, REDUCE_OUTPUT, OOB_FILE, OOB_URL tmp, HTTP_PORT, LOCAL_PATH = load_conf() OOB_URL = ("http://%s/disco/ctrl/oob_get?" % this_master())\ + "name=%s&key=%s" PARAMS_FILE = LOCAL_PATH + "%s/params" EXT_MAP = LOCAL_PATH + "%s/ext-map" EXT_REDUCE = LOCAL_PATH + "%s/ext-reduce" PART_SUFFIX = "-%.9d" MAP_OUTPUT = LOCAL_PATH + "%s/map-disco-%d" + PART_SUFFIX CHUNK_OUTPUT = LOCAL_PATH + "%s/map-chunk-%d" REDUCE_DL = LOCAL_PATH + "%s/reduce-in-%d.dl" REDUCE_SORTED = LOCAL_PATH + "%s/reduce-in-%d.sorted" REDUCE_OUTPUT = LOCAL_PATH + "%s/reduce-disco-%d" OOB_FILE = LOCAL_PATH + "%s/oob/%s"