Ticket #3470: VFS s3+ bugfixes & improvements.

* Resolve "Please use  AWS4-HMAC-SHA256" error: enforce the new V4 authentication
method. It is required in many (if not all) locations nowadays.
* Now s3+ works with buckets in different regions: locations are auto-detected.
* Debug level specification support (MCVFS_EXTFS_S3_DEBUGLEVEL).

Signed-off-by: Andrew Borodin <aborodin@vmail.ru>
This commit is contained in:
Dmitry Koterov 2015-05-21 19:26:37 +03:00 committed by Andrew Borodin
parent 4167717e4c
commit e09d64179e

View File

@ -39,8 +39,9 @@
# AWS_ACCESS_KEY_ID : Amazon AWS acces key (required)
# AWS_SECRET_ACCESS_KEY : Amazon AWS secret access key (required)
# Optional:
# MCVFS_EXTFS_S3_LOCATION : where to create new buckets, "EU"(default) or "US"
# MCVFS_EXTFS_S3_DEBUGFILE : write debug info to this file (no info default)
# MCVFS_EXTFS_S3_LOCATION : where to create new buckets: "EU" - default, "USWest", "APNortheast" etc.
# MCVFS_EXTFS_S3_DEBUGFILE : write debug info to this file (no info by default)
# MCVFS_EXTFS_S3_DEBUGLEVEL : debug messages level ("WARNING" - default, "DEBUG" - verbose)
#
#
# Usage:
@ -48,6 +49,12 @@
#
#
# History:
# 2015-05-21 Dmitry Koterov <dmitry.koterov@gmail.com>
# - Resolve "Please use AWS4-HMAC-SHA256" error: enforce the new V4 authentication method.
# It is required in many (if not all) locations nowadays.
# - Now s3+ works with buckets in different regions: locations are auto-detected.
# - Debug level specification support (MCVFS_EXTFS_S3_DEBUGLEVEL).
#
# 2009-02-07 Jakob Kemi <jakob.kemi@gmail.com>
# - Updated instructions.
# - Improved error reporting.
@ -71,7 +78,6 @@ import datetime
import boto
from boto.s3.connection import S3Connection
from boto.s3.key import Key
from boto.exception import BotoServerError
@ -79,8 +85,9 @@ from boto.exception import BotoServerError
USER=os.getenv('USER','0')
AWS_ACCESS_KEY_ID=os.getenv('AWS_ACCESS_KEY_ID')
AWS_SECRET_ACCESS_KEY=os.getenv('AWS_SECRET_ACCESS_KEY')
LOCATION = os.getenv('MCVFS_EXTFS_S3_LOCATION', 'EU').lower()
DEBUGFILE = os.getenv('MCVFS_EXTFS_S3_DEBUGFILE')
S3LOCATION=os.getenv('MCVFS_EXTFS_S3_LOCATION', 'EU')
DEBUGFILE=os.getenv('MCVFS_EXTFS_S3_DEBUGFILE')
DEBUGLEVEL=os.getenv('MCVFS_EXTFS_S3_DEBUGLEVEL', 'WARNING')
if not AWS_ACCESS_KEY_ID or not AWS_SECRET_ACCESS_KEY:
sys.stderr.write('Missing AWS_ACCESS_KEY_ID or AWS_SECRET_ACCESS_KEY environment variables.\n')
@ -93,7 +100,7 @@ if DEBUGFILE:
filename=DEBUGFILE,
level=logging.DEBUG,
format='%(asctime)s %(levelname)s %(message)s')
logging.getLogger('boto').setLevel(logging.WARNING)
logging.getLogger('boto').setLevel(getattr(logging, DEBUGLEVEL))
else:
class Void(object):
def __getattr__(self, attr):
@ -163,17 +170,52 @@ def threadmap(fun, iterable, maxthreads=16):
logger.debug('started')
# Global S3 connection
s3 = S3Connection(AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY)
if LOCATION == 'eu':
logger.debug('Using location EU for new buckets')
S3LOCATION = boto.s3.connection.Location.EU
else:
logger.debug('Using location US for new buckets')
S3LOCATION = boto.s3.connection.Location.US
if S3LOCATION.upper() == "EU":
S3LOCATION = "eu-central-1"
if S3LOCATION.upper() == "US":
S3LOCATION = "us-east-1"
for att in dir(boto.s3.connection.Location):
v = getattr(boto.s3.connection.Location, att)
if type(v) is str and att.lower() == S3LOCATION.lower():
S3LOCATION = v
break
logger.debug('Using location %s for new buckets', S3LOCATION)
def get_connection(location):
"""
Creates a connection to the specified region.
"""
os.environ['S3_USE_SIGV4'] = 'True' # only V4 method is supported in all locations.
return boto.s3.connect_to_region(
location,
aws_access_key_id=AWS_ACCESS_KEY_ID,
aws_secret_access_key=AWS_SECRET_ACCESS_KEY
)
# Global S3 default connection.
s3 = get_connection('us-east-1')
def get_bucket(name):
"""
Returns a bucket by its name, no matter what region is it in.
"""
try:
b = s3.get_bucket(name, validate=False)
b.get_location() # just to raise an exception on error
return b
except boto.exception.S3ResponseError, e:
# Seems this is the only proper way to switch to the bucket's region.
# Requesting of the default region for "?location" does not work unfortunately.
m = re.search(r'<Region>(.*?)</Region>', e.body)
if m:
return get_connection(m.group(1)).get_bucket(name)
raise
logger.debug('argv: ' + str(sys.argv))
try:
cmd = sys.argv[1]
args = sys.argv[2:]
@ -181,6 +223,7 @@ except:
sys.stderr.write('This program should be called from within MC\n')
sys.exit(1)
def handleServerError(msg):
e = sys.exc_info()
msg += ', reason: ' + e[1].reason
@ -242,7 +285,7 @@ if cmd == 'list':
try:
if os.path.islink('/etc/localtime'):
link = os.readlink('/etc/localtime')
tz = '/'.join(p.split(os.path.sep)[-2:])
tz = '/'.join(link.split(os.path.sep)[-2:])
return pytz.timezone(tz)
except:
pass
@ -269,10 +312,16 @@ if cmd == 'list':
def bucketList(b):
b = get_bucket(b.name) # get the bucket at its own region
totsz = 0
mostrecent = '1970-01-01T00:00:00.000Z'
ret = []
for k in b.list():
if k.name.endswith('/'):
# Sometimes someone create S3 keys which are ended with "/".
# Extfs cannot work with them as with files, and such keys may
# hide same-name directories, so we skip them.
continue
mostrecent = max(mostrecent, k.last_modified)
datetime = convDate(k.last_modified)
ret.append('%10s %3d %-8s %-8s %d %s %s\n' % (
@ -301,7 +350,7 @@ elif cmd == 'copyout':
logger.info('copyout bucket: %s, key: %s'%(bucket, key))
try:
b = s3.get_bucket(bucket)
b = get_bucket(bucket)
k = b.get_key(key)
out = open(extractto, 'w')
@ -326,7 +375,7 @@ elif cmd == 'copyin':
logger.info('copyin bucket: %s, key: %s'%(bucket, key))
try:
b = s3.get_bucket(bucket)
b = get_bucket(bucket)
k = b.new_key(key)
k.set_contents_from_file(fp=open(sourcefile,'r'))
except BotoServerError:
@ -343,7 +392,7 @@ elif cmd == 'rm':
logger.info('rm bucket: %s, key: %s'%(bucket, key))
try:
b = s3.get_bucket(bucket)
b = get_bucket(bucket)
b.delete_key(key)
except BotoServerError:
handleServerError('Unable to remove key "%s"' % (key))
@ -362,7 +411,7 @@ elif cmd == 'mkdir':
else:
bucket = dirname
try:
s3.create_bucket(bucket, location=boto.s3.connection.Location.EU)
get_connection(S3LOCATION).create_bucket(bucket, location=S3LOCATION)
except BotoServerError:
handleServerError('Unable to create bucket "%s"' % (bucket))
@ -380,8 +429,8 @@ elif cmd == 'rmdir':
else:
bucket = dirname
try:
b = s3.get_bucket(bucket)
s3.delete_bucket(b)
b = get_bucket(bucket)
b.connection.delete_bucket(b)
except BotoServerError:
handleServerError('Unable to delete bucket "%s"' % (bucket))