def create_haystack(pid, gid, files):
for fn, fpath in files.items():
checksum = fn.split('-', 1)[0]
#一致性 检查
try:
s2recovery_util.check_file_consistent(fpath, checksum)
except s2recovery_util.FileDamaged as e:
logger.info( repr(e) + ' ' + fpath )
raise group.FileDamaged ( ('pid', pid), ('gid', gid), ('fn', fn) )
needle_list = create_needle_list(files)
#创建needle-list 格式如下
#[{'file_names': ['069e6fabca59e1ba3db034950b88aa074708cf30-99532'], 'data_parts': [{'path': #'/data6/064247da00014006a370842b2b0c0467/g553698/040/069e6fabca59e1ba3db034950b88aa074708cf30-99532'#, 'size': 229, 'offset': 0}], 'key': #'\x06\x9eo\xab\xcaY\xe1\xba=\xb04\x95\x0b\x88\xaa\x07G\x08\xcf0'},
hpath = get_haystack_path(pid, gid)
_mkdir(hpath) #创建 haystack目录
port = partition.pid_port(pid)
h = haystack.Haystack(hpath)
try:
h.create( needle_list, chunk_num = CHUNK_NUM,
switch_version_port = port,
switch_version_url = SWITCH_VERSION_PREFIX + pid + '/' + str(gid) )
except haystack_version.VersionExisted as e:
logger.info( 'haystack %s %s current version is existed ' % (pid, gid) )
def recovery_by_haystack(pid, gid):
hpath = get_haystack_path(pid, gid)
h = haystack.Haystack(hpath)
old_num = h.get_latest_version_num()
ver_num, haystack_fpaths = download_haystack(pid, gid)
renamed_paths = []
try:
attr_sha1 = haystack_version.get_file_sha1( haystack_fpaths['attribute'] )
for fn_key, src_path in haystack_fpaths.items():
dst_path = get_haystack_file_path(pid, gid, ver_num, fn_key)
rename_force(src_path, dst_path)
renamed_paths.append(dst_path)
h.reset_version_num( ver_num, attr_sha1 )
except Exception as e:
delete_files( renamed_paths )
delete_files( haystack_fpaths.values() )
raise
switch_port = partition.pid_port(pid)
switch_url = SWITCH_VERSION_PREFIX + pid + '/' + str(gid)
h._switch_version(switch_port, switch_url)
if old_num != ver_num:
h.delete_version(old_num, switch_port, switch_url)
def create_haystack(pid, gid, files):
for fn, fpath in files.items():
checksum = fn.split('-', 1)[0]
try:
s2recovery_util.check_file_consistent(fpath, checksum)
except s2recovery_util.FileDamaged as e:
logger.info( repr(e) + ' ' + fpath )
raise group.FileDamaged ( ('pid', pid), ('gid', gid), ('fn', fn) )
needle_list = create_needle_list(files)
hpath = get_haystack_path(pid, gid)
_mkdir(hpath)
port = partition.pid_port(pid)
h = haystack.Haystack(hpath)
try:
h.create( needle_list, chunk_num = CHUNK_NUM,
switch_version_port = port,
switch_version_url = SWITCH_VERSION_PREFIX + pid + '/' + str(gid) )
except haystack_version.VersionExisted as e:
logger.info( 'haystack %s %s current version is existed ' % (pid, gid) )
def clear_haystack_original_fns(pid, gid):
hpath = get_haystack_path(pid, gid)
h = haystack.Haystack(hpath)
ver_num = h.get_latest_version_num()
if ver_num is not None:
empty_file( get_haystack_file_path(pid, gid, ver_num, 'original_file_names') )
def check_haystack(pid, gid):
gid = int(gid)
hpath = get_haystack_path(pid, gid)
h = haystack.Haystack(hpath)
if not h.exist():
return
logger.info('haystack_check: pid:%s, gid:%d haystack begin checking !'%(pid, gid))
check_global_version_file(pid, gid)
check_haystack_files(pid, gid)
logger.info('haystack_check: pid:%s, gid:%d haystack finish checking !'%(pid, gid))
def check_global_version_file(pid, gid):
if is_global_version_file_ok(pid, gid):
return
logger.info( 'pid: %s, gid:%s global version file damaged!'%(pid, gid) )
hpath = get_haystack_path(pid, gid)
h = haystack.Haystack(hpath)
empty_file(h.global_version)
if recovery_global_version_by_attr_file(pid, gid):
return
recovery_haystack(pid, gid)
def is_global_version_file_ok(pid, gid):
hpath = get_haystack_path(pid, gid)
h = haystack.Haystack(hpath)
try:
ver_num = h.get_latest_version_num()
if ver_num is None:
return False
except (OSError, IOError) as e:
logger.info(repr(e) + 'haystack_check: pid:%s, gid:%d, global_version file damage'%(pid, gid))
return False
if os.path.exists( get_haystack_file_path(pid, gid, ver_num, 'attribute') ):
return True
else:
logger.warn( 'haystack_check: pid:%s, gid:%d, global_version file damage'%(pid, gid) )
return False
def recovery_global_version_by_attr_file(pid, gid):
hpath = get_haystack_path(pid, gid)
h = haystack.Haystack(hpath)
fn = get_latest_attribute_fn(hpath)
if fn is None:
return False
try:
attr_sha1 = haystack_version.get_file_sha1( os.path.join(hpath, fn) )
except haystack_version.FileDamaged as e:
logger.info( 'attribute file is damaged %s' %fn )
return False
ver_num = fn[ len('haystack_') : -len('_attribute') ]
h.reset_version_num(ver_num, attr_sha1)
return True
def remerge_by_original_files(pid, gid):
logger.info('haystack_check: pid:%s, gid:%d, begin creating new haystack by original files!'%(pid, gid))
clear_haystack_original_fns(pid, gid)
try:
s2recovery.recover_group(pid, gid)
except s2recovery.GroupIncomplete as e:
pass
hash_files = get_hash_files(pid, gid)
needle_list = create_needle_list(hash_files)
hpath = get_haystack_path(pid, gid)
h = haystack.Haystack(hpath)
ver_num = h.get_latest_version_num()
if ver_num is not None:
try:
ver = haystack_version.HaystackVersion(hpath, ver_num)
needles = ver.get_valid_needle_list( ver.get_damaged_chunks_indexes() )
needle_list.extend(needles)
except (haystack_version.VersionNotFound, haystack_version.FileDamaged) as e:
logger.info( repr(e) + 'version %s error!' % ver_num )
new_num = h.create_version(needle_list, CHUNK_NUM,
switch_version_port = partition.pid_port(pid),
switch_version_url = SWITCH_VERSION_PREFIX + pid + '/' + str(gid),
force = True)
delete_files( hash_files.values() )
delete_empty_hash_folders(pid, gid)
logger.info('haystack_check: pid:%s, gid:%d, finish creating new haystack by original files, version num is %s !'
%(pid, gid, ver_num))
def check_haystack_files(pid, gid):
hpath = get_haystack_path(pid, gid)
h = haystack.Haystack(hpath)
ver_num = h.get_latest_version_num()
fpath = get_haystack_file_path(pid, gid, ver_num, 'attribute')
if is_attribute_file_ok(fpath) == False:
logger.info( 'pid: %s gid:%s ver_num:%s attribute file damaged!'%(pid, gid, ver_num) )
try:
sha1 = h.get_attr_sha1(ver_num)
recovery_haystack_file(pid, gid, ver_num, 'attribute', sha1)
except s2recovery_util.DownloadError as e:
logger.info( repr(e) + 'pid:%s gid:%d ver_num:%s recovery attribute fail!'%(pid, gid, ver_num) )
recovery_haystack(pid, gid)
return
hfs = get_haystack_files_list(pid, gid)
for fn_key, sha1 in hfs['files_sha1'].items():
if not is_haystack_file_ok(pid, gid, hfs['version'], fn_key, sha1):
logger.info( 'pid: %s gid:%s ver_num:%s fn:%s file damaged!'%(pid, gid, hfs['version'], fn_key) )
try:
recovery_haystack_file(pid, gid, hfs['version'], fn_key, sha1)
except s2recovery_util.DownloadError as e:
logger.info( repr(e) + 'pid:%s gid:%d ver_num:%s recovery fn:%s fail!'%(pid, gid, hfs['version'], fn_key) )
break
else:
return
recovery_haystack(pid, gid)