Retarget hard links pointing to deleted files by emitting the deleted file contents under the first hard link instance. This fixes a breakage in the squashed TAR where we were pointing hard links to missing data.

Fixes https://jira.coreos.com/browse/QUAY-885
This commit is contained in:
Joseph Schorr 2018-03-21 16:05:27 -04:00
parent 041a7fcd36
commit 110366f656
8 changed files with 337 additions and 252 deletions

View file

@ -2,17 +2,16 @@ import unittest
import tarfile
from StringIO import StringIO
from util.registry.streamlayerformat import StreamLayerMerger, AUFS_WHITEOUT
from util.registry.streamlayerformat import StreamLayerMerger
from util.registry.aufs import AUFS_WHITEOUT
from util.registry.tarlayerformat import TarLayerReadException
class TestStreamLayerMerger(unittest.TestCase):
def create_layer(self, **kwargs):
def create_layer(self, *file_pairs):
output = StringIO()
with tarfile.open(fileobj=output, mode='w:gz') as tar:
for current_contents in kwargs:
current_filename = kwargs[current_contents]
if current_contents == '_':
for current_filename, current_contents in file_pairs:
if current_contents is None:
# This is a deleted file.
if current_filename.endswith('/'):
current_filename = current_filename[:-1]
@ -25,9 +24,15 @@ class TestStreamLayerMerger(unittest.TestCase):
current_contents = ''
info = tarfile.TarInfo(name=current_filename)
info.size = len(current_contents)
tar.addfile(info, fileobj=StringIO(current_contents))
if current_contents.startswith('linkto:'):
info = tarfile.TarInfo(name=current_filename)
info.linkname = current_contents[len('linkto:'):]
info.type = tarfile.LNKTYPE
tar.addfile(info)
else:
info = tarfile.TarInfo(name=current_filename)
info.size = len(current_contents)
tar.addfile(info, fileobj=StringIO(current_contents))
return output.getvalue()
@ -35,10 +40,13 @@ class TestStreamLayerMerger(unittest.TestCase):
return ''
def squash_layers(self, layers, path_prefix=None):
def get_layers():
return [StringIO(layer) for layer in layers]
def getter_for_layer(layer):
return lambda: StringIO(layer)
merger = StreamLayerMerger(get_layers, path_prefix=path_prefix)
def layer_stream_getter():
return [getter_for_layer(layer) for layer in layers]
merger = StreamLayerMerger(layer_stream_getter, path_prefix=path_prefix)
merged_data = ''.join(merger.get_generator())
return merged_data
@ -58,9 +66,9 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_single_layer(self):
tar_layer = self.create_layer(
foo = 'some_file',
bar = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', 'bar'),
('third_file', 'meh'))
squashed = self.squash_layers([tar_layer])
@ -70,12 +78,12 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_multiple_layers(self):
second_layer = self.create_layer(
foo = 'some_file',
bar = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', 'bar'),
('third_file', 'meh'))
first_layer = self.create_layer(
top = 'top_file')
('top_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer])
@ -86,12 +94,12 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_multiple_layers_dot(self):
second_layer = self.create_layer(
foo = './some_file',
bar = 'another_file',
meh = './third_file')
('./some_file', 'foo'),
('another_file', 'bar'),
('./third_file', 'meh'))
first_layer = self.create_layer(
top = 'top_file')
('top_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer])
@ -102,12 +110,12 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_multiple_layers_overwrite(self):
second_layer = self.create_layer(
foo = 'some_file',
bar = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', 'bar'),
('third_file', 'meh'))
first_layer = self.create_layer(
top = 'another_file')
('another_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer])
@ -117,12 +125,12 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_multiple_layers_overwrite_base_dot(self):
second_layer = self.create_layer(
foo = 'some_file',
bar = './another_file',
meh = 'third_file')
('some_file', 'foo'),
('./another_file', 'bar'),
('third_file', 'meh'))
first_layer = self.create_layer(
top = 'another_file')
('another_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer])
@ -133,12 +141,12 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_multiple_layers_overwrite_top_dot(self):
second_layer = self.create_layer(
foo = 'some_file',
bar = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', 'bar'),
('third_file', 'meh'))
first_layer = self.create_layer(
top = './another_file')
('./another_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer])
@ -149,12 +157,12 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_deleted_file(self):
second_layer = self.create_layer(
foo = 'some_file',
bar = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', 'bar'),
('third_file', 'meh'))
first_layer = self.create_layer(
_ = 'another_file')
('another_file', None))
squashed = self.squash_layers([first_layer, second_layer])
@ -164,15 +172,15 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_deleted_readded_file(self):
third_layer = self.create_layer(
bar = 'another_file')
('another_file', 'bar'))
second_layer = self.create_layer(
foo = 'some_file',
_ = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', None),
('third_file', 'meh'))
first_layer = self.create_layer(
newagain = 'another_file')
('another_file', 'newagain'))
squashed = self.squash_layers([first_layer, second_layer, third_layer])
@ -182,15 +190,15 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_deleted_in_lower_layer(self):
third_layer = self.create_layer(
bar = 'deleted_file')
('deleted_file', 'bar'))
second_layer = self.create_layer(
foo = 'some_file',
_ = 'deleted_file',
meh = 'third_file')
('some_file', 'foo'),
('deleted_file', None),
('third_file', 'meh'))
first_layer = self.create_layer(
top = 'top_file')
('top_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer, third_layer])
@ -201,31 +209,31 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_deleted_in_lower_layer_with_added_dot(self):
third_layer = self.create_layer(
something = './deleted_file')
('./deleted_file', 'something'))
second_layer = self.create_layer(
_ = 'deleted_file')
('deleted_file', None))
squashed = self.squash_layers([second_layer, third_layer])
self.assertDoesNotHaveFile(squashed, 'deleted_file')
def test_deleted_in_lower_layer_with_deleted_dot(self):
third_layer = self.create_layer(
something = './deleted_file')
('./deleted_file', 'something'))
second_layer = self.create_layer(
_ = './deleted_file')
('./deleted_file', None))
squashed = self.squash_layers([second_layer, third_layer])
self.assertDoesNotHaveFile(squashed, 'deleted_file')
def test_directory(self):
second_layer = self.create_layer(
foo = 'foo/some_file',
bar = 'foo/another_file')
('foo/some_file', 'foo'),
('foo/another_file', 'bar'))
first_layer = self.create_layer(
top = 'foo/some_file')
('foo/some_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer])
@ -234,11 +242,11 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_sub_directory(self):
second_layer = self.create_layer(
foo = 'foo/some_file',
bar = 'foo/bar/another_file')
('foo/some_file', 'foo'),
('foo/bar/another_file', 'bar'))
first_layer = self.create_layer(
top = 'foo/some_file')
('foo/some_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer])
@ -247,11 +255,11 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_delete_directory(self):
second_layer = self.create_layer(
foo = 'foo/some_file',
bar = 'foo/another_file')
('foo/some_file', 'foo'),
('foo/another_file', 'bar'))
first_layer = self.create_layer(
_ = 'foo/')
('foo/', None))
squashed = self.squash_layers([first_layer, second_layer])
@ -260,11 +268,11 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_delete_sub_directory(self):
second_layer = self.create_layer(
foo = 'foo/some_file',
bar = 'foo/bar/another_file')
('foo/some_file', 'foo'),
('foo/bar/another_file', 'bar'))
first_layer = self.create_layer(
_ = 'foo/bar/')
('foo/bar/', None))
squashed = self.squash_layers([first_layer, second_layer])
@ -273,11 +281,11 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_delete_sub_directory_with_dot(self):
second_layer = self.create_layer(
foo = 'foo/some_file',
bar = 'foo/bar/another_file')
('foo/some_file', 'foo'),
('foo/bar/another_file', 'bar'))
first_layer = self.create_layer(
_ = './foo/bar/')
('./foo/bar/', None))
squashed = self.squash_layers([first_layer, second_layer])
@ -286,11 +294,11 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_delete_sub_directory_with_subdot(self):
second_layer = self.create_layer(
foo = './foo/some_file',
bar = './foo/bar/another_file')
('./foo/some_file', 'foo'),
('./foo/bar/another_file', 'bar'))
first_layer = self.create_layer(
_ = 'foo/bar/')
('foo/bar/', None))
squashed = self.squash_layers([first_layer, second_layer])
@ -300,14 +308,14 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_delete_directory_recreate(self):
third_layer = self.create_layer(
foo = 'foo/some_file',
bar = 'foo/another_file')
('foo/some_file', 'foo'),
('foo/another_file', 'bar'))
second_layer = self.create_layer(
_ = 'foo/')
('foo/', None))
first_layer = self.create_layer(
baz = 'foo/some_file')
('foo/some_file', 'baz'))
squashed = self.squash_layers([first_layer, second_layer, third_layer])
@ -316,11 +324,11 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_delete_directory_prefix(self):
third_layer = self.create_layer(
foo = 'foobar/some_file',
bar = 'foo/another_file')
('foobar/some_file', 'foo'),
('foo/another_file', 'bar'))
second_layer = self.create_layer(
_ = 'foo/')
('foo/', None))
squashed = self.squash_layers([second_layer, third_layer])
@ -330,11 +338,11 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_delete_directory_pre_prefix(self):
third_layer = self.create_layer(
foo = 'foobar/baz/some_file',
bar = 'foo/another_file')
('foobar/baz/some_file', 'foo'),
('foo/another_file', 'bar'))
second_layer = self.create_layer(
_ = 'foo/')
('foo/', None))
squashed = self.squash_layers([second_layer, third_layer])
@ -344,11 +352,11 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_delete_root_directory(self):
third_layer = self.create_layer(
foo = 'build/first_file',
bar = 'build/second_file')
('build/first_file', 'foo'),
('build/second_file', 'bar'))
second_layer = self.create_layer(
_ = 'build')
('build', None))
squashed = self.squash_layers([second_layer, third_layer])
@ -358,8 +366,8 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_tar_empty_layer(self):
third_layer = self.create_layer(
foo = 'build/first_file',
bar = 'build/second_file')
('build/first_file', 'foo'),
('build/second_file', 'bar'))
empty_layer = self.create_layer()
@ -371,8 +379,8 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_data_empty_layer(self):
third_layer = self.create_layer(
foo = 'build/first_file',
bar = 'build/second_file')
('build/first_file', 'foo'),
('build/second_file', 'bar'))
empty_layer = self.create_empty_layer()
@ -384,8 +392,8 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_broken_layer(self):
third_layer = self.create_layer(
foo = 'build/first_file',
bar = 'build/second_file')
('build/first_file', 'foo'),
('build/second_file', 'bar'))
broken_layer = 'not valid data'
@ -397,9 +405,9 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_single_layer_with_prefix(self):
tar_layer = self.create_layer(
foo = 'some_file',
bar = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', 'bar'),
('third_file', 'meh'))
squashed = self.squash_layers([tar_layer], path_prefix='foo/')
@ -409,12 +417,12 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_multiple_layers_overwrite_with_prefix(self):
second_layer = self.create_layer(
foo = 'some_file',
bar = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', 'bar'),
('third_file', 'meh'))
first_layer = self.create_layer(
top = 'another_file')
('another_file', 'top'))
squashed = self.squash_layers([first_layer, second_layer], path_prefix='foo/')
@ -425,7 +433,7 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_superlong_filename(self):
tar_layer = self.create_layer(
meh = 'this_is_the_filename_that_never_ends_it_goes_on_and_on_my_friend_some_people_started')
('this_is_the_filename_that_never_ends_it_goes_on_and_on_my_friend_some_people_started', 'meh'))
squashed = self.squash_layers([tar_layer],
path_prefix='foo/')
@ -435,9 +443,9 @@ class TestStreamLayerMerger(unittest.TestCase):
def test_superlong_prefix(self):
tar_layer = self.create_layer(
foo = 'some_file',
bar = 'another_file',
meh = 'third_file')
('some_file', 'foo'),
('another_file', 'bar'),
('third_file', 'meh'))
squashed = self.squash_layers([tar_layer],
path_prefix='foo/bar/baz/something/foo/bar/baz/anotherthing/whatever/this/is/a/really/long/filename/that/goes/here/')
@ -447,5 +455,40 @@ class TestStreamLayerMerger(unittest.TestCase):
self.assertHasFile(squashed, 'foo/bar/baz/something/foo/bar/baz/anotherthing/whatever/this/is/a/really/long/filename/that/goes/here/third_file', 'meh')
def test_hardlink_to_deleted_file(self):
first_layer = self.create_layer(
('tobedeletedfile', 'somecontents'),
('link_to_deleted_file', 'linkto:tobedeletedfile'),
('third_file', 'meh'))
second_layer = self.create_layer(
('tobedeletedfile', None))
squashed = self.squash_layers([second_layer, first_layer], path_prefix='foo/')
self.assertHasFile(squashed, 'foo/third_file', 'meh')
self.assertHasFile(squashed, 'foo/link_to_deleted_file', 'somecontents')
self.assertDoesNotHaveFile(squashed, 'foo/tobedeletedfile')
def test_multiple_hardlink_to_deleted_file(self):
first_layer = self.create_layer(
('tobedeletedfile', 'somecontents'),
('link_to_deleted_file', 'linkto:tobedeletedfile'),
('another_link_to_deleted_file', 'linkto:tobedeletedfile'),
('third_file', 'meh'))
second_layer = self.create_layer(
('tobedeletedfile', None))
squashed = self.squash_layers([second_layer, first_layer], path_prefix='foo/')
self.assertHasFile(squashed, 'foo/third_file', 'meh')
self.assertHasFile(squashed, 'foo/link_to_deleted_file', 'somecontents')
self.assertHasFile(squashed, 'foo/another_link_to_deleted_file', 'somecontents')
self.assertDoesNotHaveFile(squashed, 'foo/tobedeletedfile')
if __name__ == '__main__':
unittest.main()