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]
#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 download_haystack_from_part(local_pid, remote_pid, gid):
hpath = get_haystack_path(local_pid, gid)
spath = os.path.join( hpath, SYNC_HAYSTACK_FOLDER )
_mkdir(spath)
files_path = {}
hfs = get_haystack_files_list(remote_pid, gid)
try:
for fn_key, sha1 in hfs['files_sha1'].items():
fpath = os.path.join( spath, 'haystack_%s_%s_'%(hfs['version'], fn_key) + str( time.time() ) )
ips = _get_prt_ips(remote_pid)
port = partition.pid_port( remote_pid, mine = False )
uri = stoclient.get_haystack_file_uri( remote_pid,
{'group_id': gid, 'version': hfs['version'], 'fn_key': fn_key}, timeout=36000 )
s2recovery_util.download(fpath, ips, port, uri, sha1)
files_path[fn_key] = fpath
except Exception as e:
delete_files( files_path.values() )
raise
return hfs['version'], files_path
def download_haystack_file(pid, gid, ver_num, fn_key, sha1):
hpath = get_haystack_path(pid, gid)
spath = os.path.join( hpath, SYNC_HAYSTACK_FOLDER )
_mkdir(spath)
prts = partition_util.get_sorted_prts(gid)
for remote_prt in prts:
remote_pid = remote_prt['partition_id']
if _prt_dead( remote_pid ) or remote_pid == pid or remote_prt['is_del']:
continue
ips = _get_prt_ips(remote_pid)
port = partition.pid_port( remote_pid, mine = False )
uri = stoclient.get_haystack_file_uri( remote_pid,
{'group_id': gid, 'version': ver_num, 'fn_key': fn_key}, timeout=36000 )
fpath = os.path.join( spath, 'haystack_%s_%s_'%(ver_num, fn_key) + str( time.time() ) )
try:
s2recovery_util.download(fpath, ips, port, uri, sha1)
break
except s2recovery_util.DownloadError as e:
logger.info( repr(e) + " s2recovery: download pid:%s gid:%d haytack haystack_%s_%s file fail! "%
(remote_pid, gid, ver_num, fn_key) )
else:
raise
return fpath
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 get_haystack_files_list(pid, gid):
ips = _get_prt_ips(pid)
uri = stoclient.get_haystack_files_list_uri( pid, {'group_id': gid}, timeout=36000)
port = partition.pid_port(pid, mine = False)
buf = s2recovery_util.http_get_to_buf(ips, port, uri)
j = s2json.load( buf )
return j
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))