def test_log_collector_should_truncate_large_text_files_and_ignore_large_binary_files(
self):
# Set the size limit so that some files are too large to collect in full.
with patch("azurelinuxagent.common.logcollector._FILE_SIZE_LIMIT",
SMALL_FILE_SIZE):
lc = LogCollector(self.manifest_path)
archive = lc.collect_logs()
self._assert_archive_created(archive)
expected_files = [
os.path.join(self.root_collect_dir, "waagent.log"),
self._truncated_path(
os.path.join(
self.root_collect_dir,
"waagent.log.1")), # this file should be truncated
os.path.join(self.root_collect_dir, "waagent.log.2.gz"),
os.path.join(self.root_collect_dir, "less_important_file"),
os.path.join(self.root_collect_dir, "another_dir",
"least_important_file")
]
unexpected_files = [
os.path.join(self.root_collect_dir, "waagent.log.3.gz"
) # binary files cannot be truncated, ignore it
]
self._assert_files_are_in_archive(expected_files)
self._assert_files_are_not_in_archive(unexpected_files)
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
5, no_files,
"Expected 5 files in archive, found {0}!".format(no_files))
def test_log_collector_parses_commands_in_manifest(self):
# Ensure familiar commands are parsed and unknowns are ignored (like diskinfo and malformed entries)
file_to_collect = os.path.join(self.root_collect_dir, "waagent.log")
folder_to_list = self.root_collect_dir
manifest_content = """
echo,### Test header ###
unknown command
ll,{0}
copy,{1}
diskinfo,""".format(folder_to_list, file_to_collect)
manifest_file_path = os.path.join(self.tmp_dir, "manifest")
write_file(manifest_file_path, manifest_content)
lc = LogCollector(manifest_file_path)
archive = lc.collect_logs()
with open(self.output_results_file_path, "r") as fh:
results = fh.readlines()
# Assert echo was parsed
self.assertTrue(
any([line.endswith("### Test header ###\n") for line in results]))
# Assert unknown command was reported
self.assertTrue(
any([
line.endswith("ERROR Couldn\'t parse \"unknown command\"\n")
for line in results
]))
# Assert ll was parsed
self.assertTrue(
any([
"ls -alF {0}".format(folder_to_list) in line
for line in results
]))
# Assert copy was parsed
self._assert_archive_created(archive)
self._assert_files_are_in_archive(expected_files=[file_to_collect])
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
1, no_files,
"Expected 1 file in archive, found {0}!".format(no_files))
def test_log_collector_should_collect_all_files(self):
# All files in the manifest should be collected, since none of them are over the individual file size limit,
# and combined they do not cross the archive size threshold.
lc = LogCollector(self.manifest_path)
archive = lc.collect_logs()
self._assert_archive_created(archive)
expected_files = [
os.path.join(self.root_collect_dir, "waagent.log"),
os.path.join(self.root_collect_dir, "waagent.log.1"),
os.path.join(self.root_collect_dir, "waagent.log.2.gz"),
os.path.join(self.root_collect_dir, "waagent.log.3.gz"),
os.path.join(self.root_collect_dir, "less_important_file"),
os.path.join(self.root_collect_dir, "another_dir",
"least_important_file")
]
self._assert_files_are_in_archive(expected_files)
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
6, no_files,
"Expected 6 files in archive, found {0}!".format(no_files))
def test_log_collector_should_clean_up_uncollected_truncated_files(self):
# Make sure that truncated files that are no longer needed are cleaned up. If an existing truncated file
# from a previous run is not collected in the current run, it should be deleted to free up space.
# Specify files that have priority. The list is ordered, where the first entry has the highest priority.
must_collect_files = [os.path.join(self.root_collect_dir, "waagent*")]
# Set the archive size limit so that not all files can be collected. In that case, files will be added to the
# archive according to their priority.
# Set the size limit so that only two files can be collected, of which one needs to be truncated.
with patch(
"azurelinuxagent.common.logcollector._UNCOMPRESSED_ARCHIVE_SIZE_LIMIT",
2 * SMALL_FILE_SIZE):
with patch(
"azurelinuxagent.common.logcollector._MUST_COLLECT_FILES",
must_collect_files):
with patch(
"azurelinuxagent.common.logcollector._FILE_SIZE_LIMIT",
SMALL_FILE_SIZE):
lc = LogCollector(self.manifest_path)
archive = lc.collect_logs()
self._assert_archive_created(archive)
expected_files = [
os.path.join(self.root_collect_dir, "waagent.log"),
self._truncated_path(
os.path.join(
self.root_collect_dir,
"waagent.log.1")), # this file should be truncated
]
self._assert_files_are_in_archive(expected_files)
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
2, no_files,
"Expected 2 files in archive, found {0}!".format(no_files))
# Remove the original file so it is not collected anymore. In the next collection, the truncated file should be
# removed both from the archive and from the filesystem.
rm_files(os.path.join(self.root_collect_dir, "waagent.log.1"))
with patch(
"azurelinuxagent.common.logcollector._UNCOMPRESSED_ARCHIVE_SIZE_LIMIT",
2 * SMALL_FILE_SIZE):
with patch(
"azurelinuxagent.common.logcollector._MUST_COLLECT_FILES",
must_collect_files):
with patch(
"azurelinuxagent.common.logcollector._FILE_SIZE_LIMIT",
SMALL_FILE_SIZE):
lc = LogCollector(self.manifest_path)
second_archive = lc.collect_logs()
expected_files = [
os.path.join(self.root_collect_dir, "waagent.log"),
os.path.join(self.root_collect_dir, "waagent.log.2.gz"),
]
unexpected_files = [
self._truncated_path(
os.path.join(self.root_collect_dir, "waagent.log.1"))
]
self._assert_files_are_in_archive(expected_files)
self._assert_files_are_not_in_archive(unexpected_files)
self._assert_archive_created(second_archive)
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
2, no_files,
"Expected 2 files in archive, found {0}!".format(no_files))
truncated_files = os.listdir(self.truncated_files_dir)
self.assertEquals(
0, len(truncated_files),
"Uncollected truncated file waagent.log.1 should have been deleted!"
)
def test_log_collector_should_update_archive_when_files_are_new_or_modified_or_deleted(
self):
# Ensure the archive reflects the state of files on the disk at collection time. If a file was updated, it
# needs to be updated in the archive, deleted if removed from disk, and added if not previously seen.
lc = LogCollector(self.manifest_path)
first_archive = lc.collect_logs()
self._assert_archive_created(first_archive)
# Everything should be in the archive
expected_files = [
os.path.join(self.root_collect_dir, "waagent.log"),
os.path.join(self.root_collect_dir, "waagent.log.1"),
os.path.join(self.root_collect_dir, "waagent.log.2.gz"),
os.path.join(self.root_collect_dir, "waagent.log.3.gz"),
os.path.join(self.root_collect_dir, "less_important_file"),
os.path.join(self.root_collect_dir, "another_dir",
"least_important_file")
]
self._assert_files_are_in_archive(expected_files)
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
6, no_files,
"Expected 6 files in archive, found {0}!".format(no_files))
# Update a file and its last modified time to ensure the last modified time and last collection time are not
# the same in this test
file_to_update = os.path.join(self.root_collect_dir, "waagent.log")
self._create_file_of_specific_size(
file_to_update, LARGE_FILE_SIZE) # update existing file
new_time = os.path.getmtime(file_to_update) + 5
os.utime(file_to_update, (new_time, new_time))
# Create a new file (that is covered by the manifest and will be collected) and delete a file
self._create_file_of_specific_size(
os.path.join(self.root_collect_dir, "less_important_file.1"),
LARGE_FILE_SIZE)
rm_files(os.path.join(self.root_collect_dir, "waagent.log.1"))
second_archive = lc.collect_logs()
self._assert_archive_created(second_archive)
expected_files = [
os.path.join(self.root_collect_dir, "waagent.log"),
os.path.join(self.root_collect_dir, "waagent.log.2.gz"),
os.path.join(self.root_collect_dir, "waagent.log.3.gz"),
os.path.join(self.root_collect_dir, "less_important_file"),
os.path.join(self.root_collect_dir, "less_important_file.1"),
os.path.join(self.root_collect_dir, "another_dir",
"least_important_file")
]
unexpected_files = [
os.path.join(self.root_collect_dir, "waagent.log.1")
]
self._assert_files_are_in_archive(expected_files)
self._assert_files_are_not_in_archive(unexpected_files)
file = os.path.join(self.root_collect_dir, "waagent.log")
new_file_size = self._get_uncompressed_file_size(file)
self.assertEquals(
LARGE_FILE_SIZE, new_file_size,
"File {0} hasn't been updated! Size in archive is {1}, but "
"should be {2}.".format(file, new_file_size, LARGE_FILE_SIZE))
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
6, no_files,
"Expected 6 files in archive, found {0}!".format(no_files))
def test_log_collector_should_prioritize_important_files_if_archive_too_big(
self):
# Set the archive size limit so that not all files can be collected. In that case, files will be added to the
# archive according to their priority.
# Specify files that have priority. The list is ordered, where the first entry has the highest priority.
must_collect_files = [
os.path.join(self.root_collect_dir, "waagent*"),
os.path.join(self.root_collect_dir, "less_important_file*")
]
with patch(
"azurelinuxagent.common.logcollector._UNCOMPRESSED_ARCHIVE_SIZE_LIMIT",
10 * 1024 * 1024):
with patch(
"azurelinuxagent.common.logcollector._MUST_COLLECT_FILES",
must_collect_files):
lc = LogCollector(self.manifest_path)
archive = lc.collect_logs()
self._assert_archive_created(archive)
expected_files = [
os.path.join(self.root_collect_dir, "waagent.log"),
os.path.join(self.root_collect_dir, "waagent.log.1"),
os.path.join(self.root_collect_dir, "waagent.log.2.gz")
]
unexpected_files = [
os.path.join(self.root_collect_dir, "waagent.log.3.gz"),
os.path.join(self.root_collect_dir, "less_important_file"),
os.path.join(self.root_collect_dir, "another_dir",
"least_important_file")
]
self._assert_files_are_in_archive(expected_files)
self._assert_files_are_not_in_archive(unexpected_files)
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
3, no_files,
"Expected 3 files in archive, found {0}!".format(no_files))
# Second collection, if a file got deleted, delete it from the archive and add next file on the priority list
# if there is enough space.
rm_files(os.path.join(self.root_collect_dir, "waagent.log.3.gz"))
with patch(
"azurelinuxagent.common.logcollector._UNCOMPRESSED_ARCHIVE_SIZE_LIMIT",
10 * 1024 * 1024):
with patch(
"azurelinuxagent.common.logcollector._MUST_COLLECT_FILES",
must_collect_files):
second_archive = lc.collect_logs()
expected_files = [
os.path.join(self.root_collect_dir, "waagent.log"),
os.path.join(self.root_collect_dir, "waagent.log.1"),
os.path.join(self.root_collect_dir, "waagent.log.2.gz"),
os.path.join(self.root_collect_dir, "less_important_file"),
os.path.join(self.root_collect_dir, "another_dir",
"least_important_file")
]
unexpected_files = [
os.path.join(self.root_collect_dir, "waagent.log.3.gz")
]
self._assert_files_are_in_archive(expected_files)
self._assert_files_are_not_in_archive(unexpected_files)
self._assert_archive_created(second_archive)
no_files = self._get_number_of_files_in_archive()
self.assertEquals(
5, no_files,
"Expected 5 files in archive, found {0}!".format(no_files))