def main():
hosts = None
action = None
path = None
status = None
dryrun = False
store_filepath = "/etc/data_retention/dataretention_rules.sq3"
try:
(options, remainder) = getopt.gnu_getopt(
sys.argv[1:], "H:a:p:s:r:dh",
["hosts=", "action=", "path=", "status=", "dryrun", "help"])
except getopt.GetoptError as err:
usage("Unknown option specified: " + str(err))
for (opt, val) in options:
if opt in ["-H", "--hosts"]:
hosts = val
elif opt in ["-a", "--action"]:
action = val
elif opt in ["-p", "--path"]:
path = val
elif opt in ["-s", "--status"]:
status = val
elif opt in ["-r", "--rulestore"]:
store_filepath = val
elif opt in ["-d", "--dryrun"]:
dryrun = True
elif opt in ["-h", "--help"]:
usage()
else:
usage("Unknown option specified: %s" % opt)
if len(remainder) > 0:
usage("Unknown option specified: <%s>" % remainder[0])
check_args(hosts, action, status)
if not os.path.exists(store_filepath):
usage('no such rulestore at %s' % store_filepath)
cdb = RuleStore(store_filepath)
cdb.store_db_init(None)
hosts, htype = clouseau.retention.utils.utils.get_hosts_expr_type(hosts)
# if we are given one host, check that the host has a table or whine
if htype == 'glob' and '*' not in hosts:
if not clouseau.retention.utils.ruleutils.check_host_table_exists(
cdb, hosts):
usage('no such host in rule store, %s' % hosts)
if htype == 'grain' or htype == 'glob':
client = LocalClientPlus()
hosts = client.cmd_expandminions(hosts, "test.ping", expr_form=htype)
do_action(cdb, action, hosts, status, path, dryrun)
def main():
hosts = None
action = None
path = None
status = None
dryrun = False
store_filepath = "/etc/data_retention/dataretention_rules.sq3"
try:
(options, remainder) = getopt.gnu_getopt(
sys.argv[1:], "H:a:p:s:r:dh",
["hosts=", "action=", "path=",
"status=", "dryrun", "help"])
except getopt.GetoptError as err:
usage("Unknown option specified: " + str(err))
for (opt, val) in options:
if opt in ["-H", "--hosts"]:
hosts = val
elif opt in ["-a", "--action"]:
action = val
elif opt in ["-p", "--path"]:
path = val
elif opt in ["-s", "--status"]:
status = val
elif opt in ["-r", "--rulestore"]:
store_filepath = val
elif opt in ["-d", "--dryrun"]:
dryrun = True
elif opt in ["-h", "--help"]:
usage()
else:
usage("Unknown option specified: %s" % opt)
if len(remainder) > 0:
usage("Unknown option specified: <%s>" % remainder[0])
check_args(hosts, action, status)
if not os.path.exists(store_filepath):
usage('no such rulestore at %s' % store_filepath)
cdb = RuleStore(store_filepath)
cdb.store_db_init(None)
hosts, htype = clouseau.retention.utils.utils.get_hosts_expr_type(hosts)
# if we are given one host, check that the host has a table or whine
if htype == 'glob' and '*' not in hosts:
if not clouseau.retention.utils.ruleutils.check_host_table_exists(cdb, hosts):
usage('no such host in rule store, %s' % hosts)
if htype == 'grain' or htype == 'glob':
client = LocalClientPlus()
hosts = client.cmd_expandminions(hosts, "test.ping", expr_form=htype)
do_action(cdb, action, hosts, status, path, dryrun)
class CommandLine(object):
'''
prompt user at the command line for actions to take on a given
directory or file, show results
'''
# todo: down and up should check you really are (descending,
# ascending path)
def __init__(self,
confdir,
store_filepath,
timeout,
audit_type,
ignore_also=None,
hosts_expr=None):
self.confdir = confdir
self.cdb = RuleStore(store_filepath)
self.cdb.store_db_init(None)
self.timeout = timeout
self.audit_type = audit_type
self.locations = audit_type + "_locations"
self.hosts_expr = hosts_expr
self.basedir = None
clouseau.retention.utils.cliutils.init_readline_hist()
# this is arbitrary, can tweak it later
# how many levels down we keep in our list of
# top-level dirs from which the user can start
# their interactive session
self.max_depth_top_level = 3
self.filtertype = 'all'
# fixme completely wrong
self.batchno = 1
clouseau.retention.utils.config.set_up_conf(self.confdir)
# duplicate all the ignores except for the uh
# ones specific to a host. those will be done
# at host choice time
# this includes rules, we will do those at host choice time too
# we want: global, perhost, ignore_also (if there were any)
self.local_ignored = None
self.ignores = Ignores(self.confdir)
self.ignored_from_rulestore = {}
self.ignored_also = clouseau.retention.utils.ignores.convert_ignore_also_to_ignores(
ignore_also)
self.dircontents = CurrentDirContents(self.timeout)
self.cenv = CurrentEnv()
self.cmpl = Completion(self.dircontents, self.cenv,
self.max_depth_top_level)
def do_one_host(self, host, report):
self.set_host(host)
results = clouseau.retention.utils.ignores.get_ignored_from_rulestore(
self.cdb, [host])
if host in results:
self.ignored_from_rulestore[host] = results[host]
if host not in report:
dirs_problem = None
dirs_skipped = None
else:
dirs_problem, dirs_skipped = get_dirs_toexamine(report[host])
self.cenv.set_reported_dirs(dirs_problem, dirs_skipped)
if self.cenv.problem_dirs is None and self.cenv.skipped_dirs is None:
print "No report available from this host"
elif len(self.cenv.problem_dirs) == 0 and len(
self.cenv.skipped_dirs) == 0:
print "No problem dirs and no skipped dirs on this host"
else:
dirs_problem_to_depth = [
clouseau.retention.utils.cliutils.get_path_prefix(
d, self.max_depth_top_level) for d in dirs_problem
]
dirs_skipped = [
s for s in dirs_skipped if s not in dirs_problem_to_depth
]
relevant_dirs = (sorted(list(set(dirs_problem_to_depth))) +
sorted(list(set(dirs_skipped))))
while True:
dir_todo = self.cmpl.prompt_for_dir()
if dir_todo is None:
print "Done with this host"
break
elif dir_todo not in relevant_dirs:
print "Please choose one of the following directories:"
# fixme another arbitrary setting
clouseau.retention.utils.cliutils.print_columns(
relevant_dirs, 5)
else:
self.basedir = None
self.cenv.cwdir = None
self.do_one_directory(dir_todo)
def run(self, report):
'''
call with full report output (not summary) across
hosts, this will permit the user to examine
directories and files of specified hosts and
add/update rules for those dirs and files
'''
self.cenv.set_hosts(report.keys())
while True:
host_todo = self.cmpl.prompt_for_host()
if host_todo is None:
print "exiting at user request"
break
else:
usercfgrab = RemoteUserCfGrabber(host_todo, self.timeout,
self.audit_type, self.confdir)
to_convert = usercfgrab.run(True)
self.local_ignored = clouseau.retention.utils.ignores.process_local_ignores(
to_convert)
results = clouseau.retention.utils.ignores.get_ignored_from_rulestore(
self.cdb, [host_todo])
if host_todo in results:
self.ignored_from_rulestore[host_todo] = results[host_todo]
self.do_one_host(host_todo, report)
def set_host(self, host):
self.cenv.host = host
def do_one_directory(self, path):
'''
given a list which contains absolute paths for the
subdirectories / files of a given directory, (we don't
go more than one level down, it's likely to be too much),
ask the user what status to give this directory, and
show the user information for each contained dir/file if
desired, as well as info about the directory
'''
while True:
todo = self.get_do_command(path)
if todo is None:
break
def get_do_command(self, path):
command = self.show_menu('top')
return self.do_command(command, 'top', path)
def get_menu_entry(self, choices, default, text):
self.cmpl.set_choices_completion(choices, default)
self.cenv.set_prompt()
command = raw_input(self.cenv.prompt + ' ' + text +
" [%s]: " % default)
command = command.strip()
if command == "":
command = default
return command
def show_menu(self, level):
if level == 'top':
text = ("S(set status)/E(examine directory)/"
"Filter directory listings/"
"I(ignore)/R(manage rules)/Q(quit menu)")
command = self.get_menu_entry(['S', 'E', 'I', 'F', 'R', 'Q'], 'S',
text)
elif level == 'status':
text = Status.get_statuses_prompt(", ") + ", Q(quit status menu)"
command = self.get_menu_entry(Status.STATUSES + ['Q'], text,
Status.text_to_status('good'))
if command == 'Q' or command == 'q':
level = 'top'
elif level == 'examine':
text = ("D(down a level)/U(up a level)/E(show entries)/"
"C(show contents of file)/R(show rules)/"
"F(filter directory listings/"
"M(mark file(s))/Q(quit examine menu)")
command = self.get_menu_entry(
['D', 'U', 'E', 'F', 'C', 'R', 'M', 'Q'], 'E', text)
if command == 'Q' or command == 'q':
level = 'top'
elif level == 'rule':
text = ("S(show all rules of type)/D(show rules covering dir)/"
"C(show rules covering dir contents)/"
"A(add rule to rules store)/"
"R(remove rule from rules store/"
"E(export rules from store to file)/"
"I(import rules from file to store)/Q(quit rule menu)")
command = self.get_menu_entry(['S', 'C', 'A', 'R', 'E', 'I', 'Q'],
'D', text)
if command == 'Q' or command == 'q':
level = 'top'
else:
command = None
return command
def get_file_contents(self, path):
# get 20 lines and hope that's enough for the user to evaluate
# fixme the number of lines should be configurable
fileexamin = RemoteFileExaminer(path,
self.cenv.host,
20,
self.timeout,
quiet=True)
contents = fileexamin.run()
return contents
def get_basedir_from_path(self, path):
for location in clouseau.retention.utils.config.conf[self.locations]:
if path == location or path.startswith(location + os.path.sep):
return location
# fixme is this really the right fallback? check it
return '/'
def entry_is_not_ignored(self, path, entrytype, do_check):
'''
see if the given entry is in NOT in the ingored lists and return
True if so, False otherwise
we only do this check if the do_check argment is set to 'check';
otherwise we default to True
'''
if do_check != 'check':
return True
basedir = self.get_basedir_from_path(path)
if self.audit_type == 'logs' and entrytype == 'file':
path = LocalLogsAuditor.normalize(path)
if entrytype == 'file':
checker = clouseau.retention.utils.ignores.file_is_ignored
dirs = False
else:
checker = clouseau.retention.utils.ignores.dir_is_ignored
dirs = True
for ignored in [self.ignores.global_ignored, self.ignored_also]:
if dirs:
result = checker(path, ignored)
else:
result = checker(path, basedir, ignored)
if result:
return False
for ignored in [
self.ignores.perhost_ignored, self.ignored_from_rulestore
]:
if self.cenv.host in ignored:
if dirs:
result = checker(path, ignored[self.cenv.host])
else:
result = checker(path, basedir,
ignored[self.cenv.host])
if result:
return False
return True
def get_entries_from_wildcard(self, file_expr):
'''
get entries from cwdir that match the
expression
'''
# fixme that dang batchno, what a bad idea it was
self.dircontents.get(self.cenv.host, self.cenv.cwdir, 1)
# one wildcard only, them's the breaks
if '*' in file_expr:
start, end = file_expr.split('*', 1)
return [
c for c in self.dircontents.entries_dict
if (c.startswith(start) and c.endswith(end)
and len(c) >= len(start) + len(end))
]
elif file_expr in self.dircontents.entries_dict:
return [file_expr]
else:
return []
def do_mark(self):
readline.set_completer(self.cmpl.dir_entries_completion)
file_expr = raw_input("file or dirname expression (empty to quit): ")
file_expr = file_expr.strip()
if file_expr == '':
return True
if file_expr[-1] == os.path.sep:
file_expr = file_expr[:-1]
if '*' in file_expr:
entries_todo = self.get_entries_from_wildcard(file_expr)
else:
entries_todo = [file_expr]
self.dircontents.get(self.cenv.host, self.cenv.cwdir, self.batchno)
if not self.dircontents.entries:
print 'failed to get directory contents for', self.cenv.cwdir
print 'marking dirs/files regardless'
for entry in entries_todo:
if entry not in self.dircontents.entries_dict:
print 'skipping %s, not in current dir listing' % entry
print self.dircontents.entries_dict
continue
filetype = clouseau.retention.utils.ruleutils.entrytype_to_text(
self.dircontents.entries_dict[entry]['type'])
if filetype == 'link':
print 'No need to mark', file_expr, 'links are always skipped'
continue
elif filetype != 'dir' and filetype != 'file':
print 'Not a dir or regular file, no need to mark, skipping'
continue
status = Status.text_to_status('good')
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, file_expr, filetype, status, self.cenv.host)
return True
def do_add_rule(self):
# fixme need different completer here I think, that
# completes relative to self.cwdir
readline.set_completer(None)
path = raw_input("path or wildcard expr in rule (empty to quit): ")
path = path.strip()
if path == '':
return True
default = Status.text_to_status('good')
self.cmpl.set_choices_completion(Status.STATUSES + ['Q'], default)
while True:
statuses_text = Status.get_statuses_prompt(", ")
status = raw_input(statuses_text + " Q(quit)) [%s]: " % default)
status = status.strip()
if status == "":
status = default
if status[0].upper() in Status.STATUSES:
status = status[0].upper()
break
elif status == 'q' or status == 'Q':
return None
else:
print "Unknown status type"
continue
# fixme should check that any wildcard is only one and only
# in the last component... someday
if path[0] != os.path.sep:
path = os.path.join(self.cenv.cwdir, path)
if path[-1] == os.path.sep:
path = path[:-1]
filetype = clouseau.retention.utils.ruleutils.text_to_entrytype(
'dir')
else:
filetype = clouseau.retention.utils.ruleutils.text_to_entrytype(
'file')
clouseau.retention.utils.ruleutils.do_add_rule(self.cdb, path,
filetype, status,
self.cenv.host)
# update the ignores list since we have a new rule
results = clouseau.retention.utils.ignores.get_ignored_from_rulestore(
self.cdb, [self.cenv.host])
if self.cenv.host in results:
self.ignored_from_rulestore[self.cenv.host] = results[
self.cenv.host]
return True
def do_show_rules_with_status(self):
default = Status.text_to_status('problem')
self.cmpl.set_choices_completion(['A'] + Status.STATUSES + ['Q'],
default)
while True:
statuses_text = Status.get_statuses_prompt(", ")
status = raw_input("status type A(all), " + statuses_text +
", Q(quit)) [%s]: " % default)
status = status.strip()
if status == "":
status = default
if status == 'q' or status == 'Q':
return None
elif status[0].upper() not in ['A'] + Status.STATUSES:
print "Unknown status type"
continue
readline.set_completer(None)
prefix = raw_input("starting with prefix? [/]: ")
prefix = prefix.strip()
if prefix == "":
prefix = "/"
if status == 'a' or status == 'A':
clouseau.retention.utils.ruleutils.show_rules(self.cdb,
self.cenv.host,
prefix=prefix)
return True
elif status[0].upper() in Status.STATUSES:
clouseau.retention.utils.ruleutils.show_rules(
self.cdb, self.cenv.host, status[0].upper(), prefix=prefix)
return True
def do_remove_rule(self):
# fixme need different completer here I think, that
# completes relative to self.cwdir
readline.set_completer(None)
path = raw_input("path or wildcard expr in rule (empty to quit): ")
path = path.strip()
if path == '':
return True
elif path[0] != os.path.sep:
path = os.path.join(self.cenv.cwdir, path)
if path[-1] == os.path.sep:
path = path[:-1]
clouseau.retention.utils.ruleutils.do_remove_rule(
self.cdb, path, self.cenv.host)
# update the ignores list since we removed a rule
results = clouseau.retention.utils.ignores.get_ignored_from_rulestore(
self.cdb, [self.cenv.host])
if self.cenv.host in results:
self.ignored_from_rulestore[self.cenv.host] = results[
self.cenv.host]
return True
def get_rules_path(self):
readline.set_completer(None)
rules_path = raw_input("full path to rules file (empty to quit): ")
rules_path = rules_path.strip()
if rules_path == '':
return rules_path
if not clouseau.retention.utils.cliutils.check_rules_path(rules_path):
print "bad rules file path specified, aborting"
return ''
return rules_path
def do_rule(self, command):
if command == 'A' or command == 'a':
result = self.do_add_rule()
elif command == 'S' or command == 's':
result = self.do_show_rules_with_status()
elif command == 'D' or command == 'd':
self.dircontents.get(self.cenv.host, self.cenv.cwdir, self.batchno)
clouseau.retention.utils.ruleutils.get_rules_for_path(
self.cdb, self.cenv.cwdir, self.cenv.host)
result = True
elif command == 'C' or command == 'c':
self.dircontents.get(self.cenv.host, self.cenv.cwdir, self.batchno)
clouseau.retention.utils.ruleutils.get_rules_for_entries(
self.cdb, self.cenv.cwdir, self.dircontents.entries_dict,
self.cenv.host)
result = True
elif command == 'R' or command == 'r':
result = self.do_remove_rule()
elif command == 'I' or command == 'i':
rules_path = self.get_rules_path()
if rules_path != '':
clouseau.retention.utils.ruleutils.import_rules(
self.cdb, rules_path, self.cenv.host)
result = True
elif command == 'E' or command == 'e':
rules_path = self.get_rules_path()
if rules_path != '':
clouseau.retention.utils.ruleutils.export_rules(
self.cdb, rules_path, self.cenv.host)
result = True
elif command == 'Q' or command == 'q':
print "quitting this level"
result = None
else:
clouseau.retention.utils.cliutils.show_help('rule')
result = True
return result
def do_file_contents(self):
# fixme need a different completer here... meh
readline.set_completer(None)
filename = raw_input("filename (empty to quit): ")
filename = filename.strip()
if filename == '':
return
if filename[0] != os.path.sep:
filename = os.path.join(self.cenv.cwdir, filename)
contents = self.get_file_contents(filename)
if contents is not None:
print contents
else:
print "failed to get contents of file"
def do_filter(self):
default = 'C'
self.cmpl.set_choices_completion(['A', 'D', 'F', 'C', 'Q'], default)
while True:
filtertype = raw_input("filter A(all), D(directories only),"
" F(files only),"
" C(Entries checked (not ignored),"
" Q(quit)) [?]: ")
filtertype = filtertype.strip()
if filtertype == "":
filtertype = default
if filtertype == 'a' or filtertype == 'A':
self.filtertype = 'all'
return True
elif filtertype == 'D' or filtertype == 'd':
self.filtertype = 'dir'
return True
elif filtertype == 'F' or filtertype == 'f':
self.filtertype = 'file'
return True
elif filtertype == 'C' or filtertype == 'c':
self.filtertype = 'check'
return True
elif filtertype == 'q' or filtertype == 'Q':
return None
else:
print "Unknown filter type"
continue
def do_dir_descend(self, command):
while True:
# prompt user for dir to descend
readline.set_completer(self.cmpl.dir_completion)
self.cenv.set_prompt()
directory = raw_input(self.cenv.prompt + ' ' +
"directory name (empty to quit): ")
directory = directory.strip()
if directory == '':
return command
if directory[-1] == os.path.sep:
directory = directory[:-1]
if (directory[0] == '/' and
not directory.startswith(self.cenv.cwdir + os.path.sep)):
print 'New directory is not a subdirectory of',
print self.cenv.cwdir, "skipping"
else:
self.cenv.cwdir = os.path.join(self.cenv.cwdir, directory)
self.dircontents.clear()
self.cenv.set_prompt()
print 'Now at', self.cenv.cwdir
return True
def do_examine(self, command):
if command == 'D' or command == 'd':
return self.do_dir_descend(command)
elif command == 'U' or command == 'u':
if self.cenv.cwdir != self.basedir:
self.cenv.cwdir = os.path.dirname(self.cenv.cwdir)
self.dircontents.clear()
self.cenv.set_prompt()
print 'Now at', self.cenv.cwdir
else:
print 'Already at top', self.cenv.cwdir
result = True
elif command == 'E' or command == 'e':
self.dircontents.show(self.cenv.host, self.cenv.cwdir, 1,
self.filtertype, self.entry_is_not_ignored)
result = True
elif command == 'C' or command == 'c':
self.do_file_contents()
result = True
elif command == 'F' or command == 'f':
self.do_filter()
result = True
elif command == 'R' or command == 'r':
continuing = True
while continuing:
command = self.show_menu('rule')
continuing = self.do_command(command, 'rule', self.cenv.cwdir)
result = True
elif command == 'M' or command == 'm':
result = self.do_mark()
elif command == 'Q' or command == 'q' or command == '':
print "quitting this level"
result = None
else:
clouseau.retention.utils.cliutils.show_help('examine')
result = True
return result
def do_top(self, command, dir_path):
result = True
if command == 'S' or command == 's':
continuing = True
while continuing:
command = self.show_menu('status')
continuing = self.do_command(command, 'status', dir_path)
elif command == 'E' or command == 'e':
self.dircontents.show(self.cenv.host, self.cenv.cwdir, 1,
self.filtertype, self.entry_is_not_ignored)
continuing = True
while continuing:
# fixme this should let the user page through batches,
# not use '1' every time
command = self.show_menu('examine')
continuing = self.do_command(command, 'examine',
self.cenv.cwdir)
elif command == 'F' or command == 'f':
self.do_filter()
elif command == 'I' or command == 'i':
# do nothing
result = command
elif command == 'R' or command == 'r':
continuing = True
while continuing:
command = self.show_menu('rule')
continuing = self.do_command(command, 'rule', self.cenv.cwdir)
elif command == 'Q' or command == 'q':
result = None
else:
clouseau.retention.utils.cliutils.show_help('top')
return result
def do_command(self, command, level, dir_path):
result = None
if self.basedir is None:
self.basedir = dir_path
if self.cenv.cwdir is None:
self.cenv.cwdir = dir_path
if command is None:
return None
if level == 'top':
result = self.do_top(command, dir_path)
elif level == 'status':
if command in Status.STATUSES:
# this option is invoked on a directory so
# type is dir every time
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, dir_path,
clouseau.retention.utils.ruleutils.text_to_entrytype(
'dir'), command, self.cenv.host)
return None
elif command == 'Q' or command == 'q':
return None
else:
clouseau.retention.utils.cliutils.show_help(level)
result = True
elif level == 'examine':
result = self.do_examine(command)
elif level == 'rule':
result = self.do_rule(command)
return result
class RemoteFilesAuditor(object):
'''
audit files across a set of remote hosts,
in a specified set of directories
'''
def __init__(self,
hosts_expr,
audit_type,
confdir=None,
prettyprint=False,
show_content=False,
dirsizes=False,
summary_report=False,
depth=2,
to_check=None,
ignore_also=None,
timeout=60,
maxfiles=None,
store_filepath=None,
verbose=False):
'''
hosts_expr: list or grain-based or wildcard expr for hosts
to be audited
audit_type: type of audit e.g. 'logs', 'homes'
confdir: directory where the yaml config files are stored
prettyprint: nicely format the output display
show_content: show the first line or so from problematic files
dirsizes: show only directories which have too many files to
audit properly, don't report on files at all
summary_report: do a summary of results instead of detailed
this means different thiings depending on the audit
type
depth: the auditor will give up if a directory has too any files
it (saves it form dying on someone's 25gb homedir).
this option tells it how far down the tree to go from
the top dir of the audit, before starting to count.
e.g. do we count in /home/ariel or separately in
/home/ariel/* or in /home/ariel/*/*, etc.
to_check: comma-separated list of dirs (must end in '/') and/or
files that will be checked; if this is None then
all dirs/files will be checked
ignore_also: comma-separated list of dirs (must end in '/') and/or
files that will be skipped in addition to the ones
in the config, rules, etc.
timeout: salt timeout for running remote commands
maxfiles: how many files in a directory tree is too many to audit
(at which point we warn about that and move on)
store_filepath: full path to rule store (sqlite3 db)
verbose: show informative messages during processing
'''
self.hosts_expr = hosts_expr
self.audit_type = audit_type
self.confdir = confdir
self.locations = audit_type + "_locations"
self.prettyprint = prettyprint
self.show_sample_content = show_content
self.dirsizes = dirsizes
self.show_summary = summary_report
self.depth = depth + 1 # actually count of path separators in dirname
self.to_check = to_check
self.ignore_also = ignore_also
self.timeout = timeout
self.store_filepath = store_filepath
self.verbose = verbose
self.max_files = maxfiles
self.set_up_max_files(maxfiles)
self.magic = clouseau.retention.utils.magic.magic_open(
clouseau.retention.utils.magic.MAGIC_NONE)
self.magic.load()
self.summary = None
self.display_from_dict = FileInfo.display_from_dict
self.runner = None
if self.hosts_expr == "127.0.0.1" or self.hosts_expr == "localhost":
# run locally
self.localaudit = True
self.expanded_hosts = []
self.cdb = None
else:
self.localaudit = False
clouseau.retention.utils.config.set_up_conf(confdir)
client = LocalClientPlus()
hosts, expr_type = Runner.get_hosts_expr_type(self.hosts_expr)
self.expanded_hosts = client.cmd_expandminions(hosts,
"test.ping",
expr_form=expr_type)
self.cdb = RuleStore(self.store_filepath)
self.cdb.store_db_init(self.expanded_hosts)
self.set_up_and_export_rule_store()
def get_audit_args(self):
audit_args = [
self.confdir, self.show_sample_content, self.dirsizes,
self.depth - 1, self.to_check, self.ignore_also, self.max_files
]
return audit_args
def set_up_runner(self):
self.runner = Runner(self.confdir, self.store_filepath,
self.hosts_expr,
self.expanded_hosts, self.audit_type,
self.get_audit_args(), self.show_sample_content,
self.to_check, self.timeout, self.verbose)
def set_up_max_files(self, maxfiles):
'''
more than this many files in a subdir we won't process,
we'll just try to name top offenders
if we've been asked only to report dir trees that are
too large in this manner, we can set defaults mich
higher, since we don't stat files, open them to guess
their filetype, etc; processing then goes much quicker
'''
if maxfiles is None:
if self.dirsizes:
self.max_files = 1000
else:
self.max_files = 100
else:
self.max_files = maxfiles
def set_up_and_export_rule_store(self):
hosts = self.cdb.store_db_list_all_hosts()
destdir = os.path.join(os.path.dirname(self.store_filepath),
"data_retention.d")
if not os.path.isdir(destdir):
os.makedirs(destdir, 0755)
for host in hosts:
all_destpath = os.path.join(destdir, host + "_store.yaml")
clouseau.retention.utils.ruleutils.export_rules(
self.cdb, all_destpath, host)
good_destpath = os.path.join(destdir, host + "_store_good.yaml")
clouseau.retention.utils.ruleutils.export_rules(
self.cdb, good_destpath, host, Status.text_to_status('good'))
def normalize(self, fname):
'''
subclasses may want to do something different, see
LogsAuditor for an example
'''
return fname
@staticmethod
def get_dirname_from_warning(warning):
'''
some audit output lines warn about directory trees
having too many files to audit; grab the dirname
out of such a line and return it
'''
start = "WARNING: directory "
if warning.startswith(start):
# WARNING: directory %s has more than %d files
rindex = warning.rfind(" has more than")
if not rindex:
return None
else:
return warning[len(start):rindex]
start = "WARNING: too many files to audit in directory "
if warning.startswith(start):
return warning[len(start):]
return None
def add_stats(self, item, summary):
'''
gather stats on how many files/dirs
may be problematic; summary is where the results
are collected, item is the item to include in
the summary if needed
'''
dirname = os.path.dirname(item['path'])
if dirname not in summary:
summary[dirname] = {
'binary': {
'old': 0,
'maybe_old': 0,
'nonroot': 0
},
'text': {
'old': 0,
'maybe_old': 0,
'nonroot': 0
}
}
if item['binary'] is True:
group = 'binary'
else:
group = 'text'
if item['old'] == 'T':
summary[dirname][group]['old'] += 1
elif item['old'] == '-':
summary[dirname][group]['maybe_old'] += 1
if item['owner'] != 0:
summary[dirname][group]['nonroot'] += 1
return summary
def display_host_summary(self):
if self.summary is not None:
paths = sorted(self.summary.keys())
for path in paths:
for group in self.summary[path]:
if (self.summary[path][group]['old'] > 0
or self.summary[path][group]['maybe_old'] > 0
or self.summary[path][group]['nonroot'] > 0):
print(
"in directory %s, (%s), %d old,"
" %d maybe old, %d with non root owner" %
(path, group, self.summary[path][group]['old'],
self.summary[path][group]['maybe_old'],
self.summary[path][group]['nonroot']))
def display_summary(self, result):
for host in result:
self.summary = {}
print "host:", host
if result[host]:
try:
lines = result[host].split('\n')
for line in lines:
if display_summary_line(line):
continue
else:
try:
item = json.loads(
line, object_hook=JsonHelper.decode_dict)
if item['empty'] is not True:
self.add_stats(item, self.summary)
except:
print "WARNING: failed to json load from host",
print host, "this line:", line
self.display_host_summary()
except:
print "WARNING: failed to process output from host"
else:
if self.verbose:
print "WARNING: no output from host", host
def display_remote_host(self, result):
try:
lines = result.split('\n')
files = []
for line in lines:
if line == "":
continue
elif line.startswith("WARNING:") or line.startswith("INFO:"):
print line
else:
files.append(
json.loads(line, object_hook=JsonHelper.decode_dict))
if files == []:
return
path_justify = max([len(finfo['path']) for finfo in files]) + 2
for finfo in files:
self.display_from_dict(finfo, self.show_sample_content,
path_justify)
except:
print "WARNING: failed to load json from host"
def get_local_auditor(self):
return LocalFilesAuditor(self.audit_type, self.confdir,
self.show_sample_content, self.dirsizes,
self.depth, self.to_check, self.ignore_also,
self.max_files)
def audit_hosts(self):
# do local audit instead
if self.localaudit:
localauditor = self.get_local_auditor()
result = localauditor.do_local_audit()
self.display_remote_host(result)
return
# proceed to regular remote audit
self.set_up_runner()
result = self.runner.run_remotely()
if result is None:
print "WARNING: failed to get output from audit script on any host"
elif self.show_summary:
self.display_summary(result)
else:
for host in result:
print "host:", host
if result[host]:
self.display_remote_host(result[host])
else:
if self.verbose:
print "no output from host", host
# add some results to rule store
self.update_status_rules_from_report(result)
return result
def update_status_rules_from_report(self, report):
hostlist = report.keys()
for host in hostlist:
try:
problem_rules = clouseau.retention.utils.ruleutils.get_rules(
self.cdb, host, Status.text_to_status('problem'))
except:
print 'WARNING: problem retrieving problem rules for host', host
problem_rules = None
if problem_rules is not None:
existing_problems = [rule['path'] for rule in problem_rules]
else:
existing_problems = []
dirs_problem, dirs_skipped = get_dirs_toexamine(report[host])
if dirs_problem is not None:
dirs_problem = list(set(dirs_problem))
for dirname in dirs_problem:
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, dirname,
clouseau.retention.utils.ruleutils.text_to_entrytype(
'dir'), Status.text_to_status('problem'), host)
if dirs_skipped is not None:
dirs_skipped = list(set(dirs_skipped))
for dirname in dirs_skipped:
if dirname in dirs_problem or dirname in existing_problems:
# problem report overrides 'too many to audit'
continue
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, dirname,
clouseau.retention.utils.ruleutils.text_to_entrytype(
'dir'), Status.text_to_status('unreviewed'), host)
class CommandLine(object):
'''
prompt user at the command line for actions to take on a given
directory or file, show results
'''
# todo: down and up should check you really are (descending,
# ascending path)
def __init__(self, confdir, store_filepath, timeout, audit_type,
ignore_also=None, hosts_expr=None):
self.confdir = confdir
self.cdb = RuleStore(store_filepath)
self.cdb.store_db_init(None)
self.timeout = timeout
self.audit_type = audit_type
self.locations = audit_type + "_locations"
self.hosts_expr = hosts_expr
self.basedir = None
clouseau.retention.utils.cliutils.init_readline_hist()
# this is arbitrary, can tweak it later
# how many levels down we keep in our list of
# top-level dirs from which the user can start
# their interactive session
self.max_depth_top_level = 3
self.filtertype = 'all'
# fixme completely wrong
self.batchno = 1
clouseau.retention.utils.config.set_up_conf(self.confdir)
# duplicate all the ignores except for the uh
# ones specific to a host. those will be done
# at host choice time
# this includes rules, we will do those at host choice time too
# we want: global, perhost, ignore_also (if there were any)
self.local_ignored = None
self.ignores = Ignores(self.confdir)
self.ignored_from_rulestore = {}
self.ignored_also = clouseau.retention.utils.ignores.convert_ignore_also_to_ignores(
ignore_also)
self.dircontents = CurrentDirContents(self.timeout)
self.cenv = CurrentEnv()
self.cmpl = Completion(self.dircontents, self.cenv, self.max_depth_top_level)
def do_one_host(self, host, report):
self.set_host(host)
results = clouseau.retention.utils.ignores.get_ignored_from_rulestore(self.cdb, [host])
if host in results:
self.ignored_from_rulestore[host] = results[host]
if host not in report:
dirs_problem = None
dirs_skipped = None
else:
dirs_problem, dirs_skipped = get_dirs_toexamine(report[host])
self.cenv.set_reported_dirs(dirs_problem, dirs_skipped)
if self.cenv.problem_dirs is None and self.cenv.skipped_dirs is None:
print "No report available from this host"
elif len(self.cenv.problem_dirs) == 0 and len(self.cenv.skipped_dirs) == 0:
print "No problem dirs and no skipped dirs on this host"
else:
dirs_problem_to_depth = [clouseau.retention.utils.cliutils.get_path_prefix(
d, self.max_depth_top_level) for d in dirs_problem]
dirs_skipped = [s for s in dirs_skipped
if s not in dirs_problem_to_depth]
relevant_dirs = (sorted(list(set(dirs_problem_to_depth))) +
sorted(list(set(dirs_skipped))))
while True:
dir_todo = self.cmpl.prompt_for_dir()
if dir_todo is None:
print "Done with this host"
break
elif dir_todo not in relevant_dirs:
print "Please choose one of the following directories:"
# fixme another arbitrary setting
clouseau.retention.utils.cliutils.print_columns(relevant_dirs, 5)
else:
self.basedir = None
self.cenv.cwdir = None
self.do_one_directory(dir_todo)
def run(self, report):
'''
call with full report output (not summary) across
hosts, this will permit the user to examine
directories and files of specified hosts and
add/update rules for those dirs and files
'''
self.cenv.set_hosts(report.keys())
while True:
host_todo = self.cmpl.prompt_for_host()
if host_todo is None:
print "exiting at user request"
break
else:
usercfgrab = RemoteUserCfGrabber(host_todo, self.timeout,
self.audit_type, self.confdir)
to_convert = usercfgrab.run(True)
self.local_ignored = clouseau.retention.utils.ignores.process_local_ignores(
to_convert)
results = clouseau.retention.utils.ignores.get_ignored_from_rulestore(
self.cdb, [host_todo])
if host_todo in results:
self.ignored_from_rulestore[host_todo] = results[host_todo]
self.do_one_host(host_todo, report)
def set_host(self, host):
self.cenv.host = host
def do_one_directory(self, path):
'''
given a list which contains absolute paths for the
subdirectories / files of a given directory, (we don't
go more than one level down, it's likely to be too much),
ask the user what status to give this directory, and
show the user information for each contained dir/file if
desired, as well as info about the directory
'''
while True:
todo = self.get_do_command(path)
if todo is None:
break
def get_do_command(self, path):
command = self.show_menu('top')
return self.do_command(command, 'top', path)
def get_menu_entry(self, choices, default, text):
self.cmpl.set_choices_completion(choices, default)
self.cenv.set_prompt()
command = raw_input(self.cenv.prompt + ' ' + text + " [%s]: " % default)
command = command.strip()
if command == "":
command = default
return command
def show_menu(self, level):
if level == 'top':
text = ("S(set status)/E(examine directory)/"
"Filter directory listings/"
"I(ignore)/R(manage rules)/Q(quit menu)")
command = self.get_menu_entry(['S', 'E', 'I', 'F', 'R', 'Q'], 'S', text)
elif level == 'status':
text = Status.get_statuses_prompt(", ") + ", Q(quit status menu)"
command = self.get_menu_entry(Status.STATUSES + ['Q'], text,
Status.text_to_status('good'))
if command == 'Q' or command == 'q':
level = 'top'
elif level == 'examine':
text = ("D(down a level)/U(up a level)/E(show entries)/"
"C(show contents of file)/R(show rules)/"
"F(filter directory listings/"
"M(mark file(s))/Q(quit examine menu)")
command = self.get_menu_entry(['D', 'U', 'E', 'F', 'C', 'R', 'M', 'Q'], 'E', text)
if command == 'Q' or command == 'q':
level = 'top'
elif level == 'rule':
text = ("S(show all rules of type)/D(show rules covering dir)/"
"C(show rules covering dir contents)/"
"A(add rule to rules store)/"
"R(remove rule from rules store/"
"E(export rules from store to file)/"
"I(import rules from file to store)/Q(quit rule menu)")
command = self.get_menu_entry(['S', 'C', 'A', 'R', 'E', 'I', 'Q'], 'D', text)
if command == 'Q' or command == 'q':
level = 'top'
else:
command = None
return command
def get_file_contents(self, path):
# get 20 lines and hope that's enough for the user to evaluate
# fixme the number of lines should be configurable
fileexamin = RemoteFileExaminer(path, self.cenv.host, 20, self.timeout, quiet=True)
contents = fileexamin.run()
return contents
def get_basedir_from_path(self, path):
for location in clouseau.retention.utils.config.conf[self.locations]:
if path == location or path.startswith(location + os.path.sep):
return location
# fixme is this really the right fallback? check it
return '/'
def entry_is_not_ignored(self, path, entrytype, do_check):
'''
see if the given entry is in NOT in the ingored lists and return
True if so, False otherwise
we only do this check if the do_check argment is set to 'check';
otherwise we default to True
'''
if do_check != 'check':
return True
basedir = self.get_basedir_from_path(path)
if self.audit_type == 'logs' and entrytype == 'file':
path = LocalLogsAuditor.normalize(path)
if entrytype == 'file':
checker = clouseau.retention.utils.ignores.file_is_ignored
dirs = False
else:
checker = clouseau.retention.utils.ignores.dir_is_ignored
dirs = True
for ignored in [self.ignores.global_ignored,
self.ignored_also]:
if dirs:
result = checker(path, ignored)
else:
result = checker(path, basedir, ignored)
if result:
return False
for ignored in [self.ignores.perhost_ignored,
self.ignored_from_rulestore]:
if self.cenv.host in ignored:
if dirs:
result = checker(path, ignored[self.cenv.host])
else:
result = checker(path, basedir, ignored[self.cenv.host])
if result:
return False
return True
def get_entries_from_wildcard(self, file_expr):
'''
get entries from cwdir that match the
expression
'''
# fixme that dang batchno, what a bad idea it was
self.dircontents.get(self.cenv.host, self.cenv.cwdir, 1)
# one wildcard only, them's the breaks
if '*' in file_expr:
start, end = file_expr.split('*', 1)
return [c for c in self.dircontents.entries_dict
if (c.startswith(start) and
c.endswith(end) and
len(c) >= len(start) + len(end))]
elif file_expr in self.dircontents.entries_dict:
return [file_expr]
else:
return []
def do_mark(self):
readline.set_completer(self.cmpl.dir_entries_completion)
file_expr = raw_input("file or dirname expression (empty to quit): ")
file_expr = file_expr.strip()
if file_expr == '':
return True
if file_expr[-1] == os.path.sep:
file_expr = file_expr[:-1]
if '*' in file_expr:
entries_todo = self.get_entries_from_wildcard(file_expr)
else:
entries_todo = [file_expr]
self.dircontents.get(self.cenv.host, self.cenv.cwdir, self.batchno)
if not self.dircontents.entries:
print 'failed to get directory contents for', self.cenv.cwdir
print 'marking dirs/files regardless'
for entry in entries_todo:
if entry not in self.dircontents.entries_dict:
print 'skipping %s, not in current dir listing' % entry
print self.dircontents.entries_dict
continue
filetype = clouseau.retention.utils.ruleutils.entrytype_to_text(
self.dircontents.entries_dict[entry]['type'])
if filetype == 'link':
print 'No need to mark', file_expr, 'links are always skipped'
continue
elif filetype != 'dir' and filetype != 'file':
print 'Not a dir or regular file, no need to mark, skipping'
continue
status = Status.text_to_status('good')
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, file_expr, filetype, status, self.cenv.host)
return True
def do_add_rule(self):
# fixme need different completer here I think, that
# completes relative to self.cwdir
readline.set_completer(None)
path = raw_input("path or wildcard expr in rule (empty to quit): ")
path = path.strip()
if path == '':
return True
default = Status.text_to_status('good')
self.cmpl.set_choices_completion(Status.STATUSES + ['Q'], default)
while True:
statuses_text = Status.get_statuses_prompt(", ")
status = raw_input(statuses_text + " Q(quit)) [%s]: " %
default)
status = status.strip()
if status == "":
status = default
if status[0].upper() in Status.STATUSES:
status = status[0].upper()
break
elif status == 'q' or status == 'Q':
return None
else:
print "Unknown status type"
continue
# fixme should check that any wildcard is only one and only
# in the last component... someday
if path[0] != os.path.sep:
path = os.path.join(self.cenv.cwdir, path)
if path[-1] == os.path.sep:
path = path[:-1]
filetype = clouseau.retention.utils.ruleutils.text_to_entrytype('dir')
else:
filetype = clouseau.retention.utils.ruleutils.text_to_entrytype('file')
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, path, filetype, status, self.cenv.host)
# update the ignores list since we have a new rule
results = clouseau.retention.utils.ignores.get_ignored_from_rulestore(
self.cdb, [self.cenv.host])
if self.cenv.host in results:
self.ignored_from_rulestore[self.cenv.host] = results[self.cenv.host]
return True
def do_show_rules_with_status(self):
default = Status.text_to_status('problem')
self.cmpl.set_choices_completion(['A'] + Status.STATUSES + ['Q'], default)
while True:
statuses_text = Status.get_statuses_prompt(", ")
status = raw_input("status type A(all), " + statuses_text +
", Q(quit)) [%s]: " % default)
status = status.strip()
if status == "":
status = default
if status == 'q' or status == 'Q':
return None
elif status[0].upper() not in ['A'] + Status.STATUSES:
print "Unknown status type"
continue
readline.set_completer(None)
prefix = raw_input("starting with prefix? [/]: ")
prefix = prefix.strip()
if prefix == "":
prefix = "/"
if status == 'a' or status == 'A':
clouseau.retention.utils.ruleutils.show_rules(
self.cdb, self.cenv.host, prefix=prefix)
return True
elif status[0].upper() in Status.STATUSES:
clouseau.retention.utils.ruleutils.show_rules(
self.cdb, self.cenv.host, status[0].upper(), prefix=prefix)
return True
def do_remove_rule(self):
# fixme need different completer here I think, that
# completes relative to self.cwdir
readline.set_completer(None)
path = raw_input("path or wildcard expr in rule (empty to quit): ")
path = path.strip()
if path == '':
return True
elif path[0] != os.path.sep:
path = os.path.join(self.cenv.cwdir, path)
if path[-1] == os.path.sep:
path = path[:-1]
clouseau.retention.utils.ruleutils.do_remove_rule(self.cdb, path, self.cenv.host)
# update the ignores list since we removed a rule
results = clouseau.retention.utils.ignores.get_ignored_from_rulestore(
self.cdb, [self.cenv.host])
if self.cenv.host in results:
self.ignored_from_rulestore[self.cenv.host] = results[self.cenv.host]
return True
def get_rules_path(self):
readline.set_completer(None)
rules_path = raw_input("full path to rules file (empty to quit): ")
rules_path = rules_path.strip()
if rules_path == '':
return rules_path
if not clouseau.retention.utils.cliutils.check_rules_path(rules_path):
print "bad rules file path specified, aborting"
return ''
return rules_path
def do_rule(self, command):
if command == 'A' or command == 'a':
result = self.do_add_rule()
elif command == 'S' or command == 's':
result = self.do_show_rules_with_status()
elif command == 'D' or command == 'd':
self.dircontents.get(self.cenv.host, self.cenv.cwdir, self.batchno)
clouseau.retention.utils.ruleutils.get_rules_for_path(self.cdb, self.cenv.cwdir,
self.cenv.host)
result = True
elif command == 'C' or command == 'c':
self.dircontents.get(self.cenv.host, self.cenv.cwdir, self.batchno)
clouseau.retention.utils.ruleutils.get_rules_for_entries(
self.cdb, self.cenv.cwdir,
self.dircontents.entries_dict,
self.cenv.host)
result = True
elif command == 'R' or command == 'r':
result = self.do_remove_rule()
elif command == 'I' or command == 'i':
rules_path = self.get_rules_path()
if rules_path != '':
clouseau.retention.utils.ruleutils.import_rules(self.cdb, rules_path,
self.cenv.host)
result = True
elif command == 'E' or command == 'e':
rules_path = self.get_rules_path()
if rules_path != '':
clouseau.retention.utils.ruleutils.export_rules(self.cdb, rules_path,
self.cenv.host)
result = True
elif command == 'Q' or command == 'q':
print "quitting this level"
result = None
else:
clouseau.retention.utils.cliutils.show_help('rule')
result = True
return result
def do_file_contents(self):
# fixme need a different completer here... meh
readline.set_completer(None)
filename = raw_input("filename (empty to quit): ")
filename = filename.strip()
if filename == '':
return
if filename[0] != os.path.sep:
filename = os.path.join(self.cenv.cwdir, filename)
contents = self.get_file_contents(filename)
if contents is not None:
print contents
else:
print "failed to get contents of file"
def do_filter(self):
default = 'C'
self.cmpl.set_choices_completion(['A', 'D', 'F', 'C', 'Q'], default)
while True:
filtertype = raw_input("filter A(all), D(directories only),"
" F(files only),"
" C(Entries checked (not ignored),"
" Q(quit)) [?]: ")
filtertype = filtertype.strip()
if filtertype == "":
filtertype = default
if filtertype == 'a' or filtertype == 'A':
self.filtertype = 'all'
return True
elif filtertype == 'D' or filtertype == 'd':
self.filtertype = 'dir'
return True
elif filtertype == 'F' or filtertype == 'f':
self.filtertype = 'file'
return True
elif filtertype == 'C' or filtertype == 'c':
self.filtertype = 'check'
return True
elif filtertype == 'q' or filtertype == 'Q':
return None
else:
print "Unknown filter type"
continue
def do_dir_descend(self, command):
while True:
# prompt user for dir to descend
readline.set_completer(self.cmpl.dir_completion)
self.cenv.set_prompt()
directory = raw_input(self.cenv.prompt + ' ' + "directory name (empty to quit): ")
directory = directory.strip()
if directory == '':
return command
if directory[-1] == os.path.sep:
directory = directory[:-1]
if (directory[0] == '/' and
not directory.startswith(self.cenv.cwdir + os.path.sep)):
print 'New directory is not a subdirectory of',
print self.cenv.cwdir, "skipping"
else:
self.cenv.cwdir = os.path.join(self.cenv.cwdir,
directory)
self.dircontents.clear()
self.cenv.set_prompt()
print 'Now at', self.cenv.cwdir
return True
def do_examine(self, command):
if command == 'D' or command == 'd':
return self.do_dir_descend(command)
elif command == 'U' or command == 'u':
if self.cenv.cwdir != self.basedir:
self.cenv.cwdir = os.path.dirname(self.cenv.cwdir)
self.dircontents.clear()
self.cenv.set_prompt()
print 'Now at', self.cenv.cwdir
else:
print 'Already at top', self.cenv.cwdir
result = True
elif command == 'E' or command == 'e':
self.dircontents.show(self.cenv.host, self.cenv.cwdir, 1,
self.filtertype, self.entry_is_not_ignored)
result = True
elif command == 'C' or command == 'c':
self.do_file_contents()
result = True
elif command == 'F' or command == 'f':
self.do_filter()
result = True
elif command == 'R' or command == 'r':
continuing = True
while continuing:
command = self.show_menu('rule')
continuing = self.do_command(command, 'rule', self.cenv.cwdir)
result = True
elif command == 'M' or command == 'm':
result = self.do_mark()
elif command == 'Q' or command == 'q' or command == '':
print "quitting this level"
result = None
else:
clouseau.retention.utils.cliutils.show_help('examine')
result = True
return result
def do_top(self, command, dir_path):
result = True
if command == 'S' or command == 's':
continuing = True
while continuing:
command = self.show_menu('status')
continuing = self.do_command(command, 'status', dir_path)
elif command == 'E' or command == 'e':
self.dircontents.show(self.cenv.host, self.cenv.cwdir, 1, self.filtertype,
self.entry_is_not_ignored)
continuing = True
while continuing:
# fixme this should let the user page through batches,
# not use '1' every time
command = self.show_menu('examine')
continuing = self.do_command(command, 'examine',
self.cenv.cwdir)
elif command == 'F' or command == 'f':
self.do_filter()
elif command == 'I' or command == 'i':
# do nothing
result = command
elif command == 'R' or command == 'r':
continuing = True
while continuing:
command = self.show_menu('rule')
continuing = self.do_command(command, 'rule', self.cenv.cwdir)
elif command == 'Q' or command == 'q':
result = None
else:
clouseau.retention.utils.cliutils.show_help('top')
return result
def do_command(self, command, level, dir_path):
result = None
if self.basedir is None:
self.basedir = dir_path
if self.cenv.cwdir is None:
self.cenv.cwdir = dir_path
if command is None:
return None
if level == 'top':
result = self.do_top(command, dir_path)
elif level == 'status':
if command in Status.STATUSES:
# this option is invoked on a directory so
# type is dir every time
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, dir_path,
clouseau.retention.utils.ruleutils.text_to_entrytype('dir'),
command, self.cenv.host)
return None
elif command == 'Q' or command == 'q':
return None
else:
clouseau.retention.utils.cliutils.show_help(level)
result = True
elif level == 'examine':
result = self.do_examine(command)
elif level == 'rule':
result = self.do_rule(command)
return result
class RemoteFilesAuditor(object):
'''
audit files across a set of remote hosts,
in a specified set of directories
'''
def __init__(self, hosts_expr, audit_type,
confdir=None,
prettyprint=False,
show_content=False, dirsizes=False, summary_report=False,
depth=2, to_check=None, ignore_also=None,
timeout=60, maxfiles=None,
store_filepath=None,
verbose=False):
'''
hosts_expr: list or grain-based or wildcard expr for hosts
to be audited
audit_type: type of audit e.g. 'logs', 'homes'
confdir: directory where the yaml config files are stored
prettyprint: nicely format the output display
show_content: show the first line or so from problematic files
dirsizes: show only directories which have too many files to
audit properly, don't report on files at all
summary_report: do a summary of results instead of detailed
this means different thiings depending on the audit
type
depth: the auditor will give up if a directory has too any files
it (saves it form dying on someone's 25gb homedir).
this option tells it how far down the tree to go from
the top dir of the audit, before starting to count.
e.g. do we count in /home/ariel or separately in
/home/ariel/* or in /home/ariel/*/*, etc.
to_check: comma-separated list of dirs (must end in '/') and/or
files that will be checked; if this is None then
all dirs/files will be checked
ignore_also: comma-separated list of dirs (must end in '/') and/or
files that will be skipped in addition to the ones
in the config, rules, etc.
timeout: salt timeout for running remote commands
maxfiles: how many files in a directory tree is too many to audit
(at which point we warn about that and move on)
store_filepath: full path to rule store (sqlite3 db)
verbose: show informative messages during processing
'''
self.hosts_expr = hosts_expr
self.audit_type = audit_type
self.confdir = confdir
self.locations = audit_type + "_locations"
self.prettyprint = prettyprint
self.show_sample_content = show_content
self.dirsizes = dirsizes
self.show_summary = summary_report
self.depth = depth + 1 # actually count of path separators in dirname
self.to_check = to_check
self.ignore_also = ignore_also
self.timeout = timeout
self.store_filepath = store_filepath
self.verbose = verbose
self.max_files = maxfiles
self.set_up_max_files(maxfiles)
self.magic = clouseau.retention.utils.magic.magic_open(
clouseau.retention.utils.magic.MAGIC_NONE)
self.magic.load()
self.summary = None
self.display_from_dict = FileInfo.display_from_dict
self.runner = None
if self.hosts_expr == "127.0.0.1" or self.hosts_expr == "localhost":
# run locally
self.localaudit = True
self.expanded_hosts = []
self.cdb = None
else:
self.localaudit = False
clouseau.retention.utils.config.set_up_conf(confdir)
client = LocalClientPlus()
hosts, expr_type = Runner.get_hosts_expr_type(self.hosts_expr)
self.expanded_hosts = client.cmd_expandminions(
hosts, "test.ping", expr_form=expr_type)
self.cdb = RuleStore(self.store_filepath)
self.cdb.store_db_init(self.expanded_hosts)
self.set_up_and_export_rule_store()
def get_audit_args(self):
audit_args = [self.confdir,
self.show_sample_content,
self.dirsizes,
self.depth - 1,
self.to_check,
self.ignore_also,
self.max_files]
return audit_args
def set_up_runner(self):
self.runner = Runner(self.confdir,
self.store_filepath,
self.hosts_expr,
self.expanded_hosts,
self.audit_type,
self.get_audit_args(),
self.show_sample_content,
self.to_check,
self.timeout,
self.verbose)
def set_up_max_files(self, maxfiles):
'''
more than this many files in a subdir we won't process,
we'll just try to name top offenders
if we've been asked only to report dir trees that are
too large in this manner, we can set defaults mich
higher, since we don't stat files, open them to guess
their filetype, etc; processing then goes much quicker
'''
if maxfiles is None:
if self.dirsizes:
self.max_files = 1000
else:
self.max_files = 100
else:
self.max_files = maxfiles
def set_up_and_export_rule_store(self):
hosts = self.cdb.store_db_list_all_hosts()
destdir = os.path.join(os.path.dirname(self.store_filepath),
"data_retention.d")
if not os.path.isdir(destdir):
os.makedirs(destdir, 0755)
for host in hosts:
all_destpath = os.path.join(destdir, host + "_store.yaml")
clouseau.retention.utils.ruleutils.export_rules(self.cdb, all_destpath, host)
good_destpath = os.path.join(destdir, host + "_store_good.yaml")
clouseau.retention.utils.ruleutils.export_rules(self.cdb, good_destpath, host,
Status.text_to_status('good'))
def normalize(self, fname):
'''
subclasses may want to do something different, see
LogsAuditor for an example
'''
return fname
@staticmethod
def get_dirname_from_warning(warning):
'''
some audit output lines warn about directory trees
having too many files to audit; grab the dirname
out of such a line and return it
'''
start = "WARNING: directory "
if warning.startswith(start):
# WARNING: directory %s has more than %d files
rindex = warning.rfind(" has more than")
if not rindex:
return None
else:
return warning[len(start):rindex]
start = "WARNING: too many files to audit in directory "
if warning.startswith(start):
return warning[len(start):]
return None
def add_stats(self, item, summary):
'''
gather stats on how many files/dirs
may be problematic; summary is where the results
are collected, item is the item to include in
the summary if needed
'''
dirname = os.path.dirname(item['path'])
if dirname not in summary:
summary[dirname] = {
'binary': {'old': 0, 'maybe_old': 0, 'nonroot': 0},
'text': {'old': 0, 'maybe_old': 0, 'nonroot': 0}
}
if item['binary'] is True:
group = 'binary'
else:
group = 'text'
if item['old'] == 'T':
summary[dirname][group]['old'] += 1
elif item['old'] == '-':
summary[dirname][group]['maybe_old'] += 1
if item['owner'] != 0:
summary[dirname][group]['nonroot'] += 1
return summary
def display_host_summary(self):
if self.summary is not None:
paths = sorted(self.summary.keys())
for path in paths:
for group in self.summary[path]:
if (self.summary[path][group]['old'] > 0 or
self.summary[path][group]['maybe_old'] > 0 or
self.summary[path][group]['nonroot'] > 0):
print ("in directory %s, (%s), %d old,"
" %d maybe old, %d with non root owner"
% (path, group, self.summary[path][group]['old'],
self.summary[path][group]['maybe_old'],
self.summary[path][group]['nonroot']))
def display_summary(self, result):
for host in result:
self.summary = {}
print "host:", host
if result[host]:
try:
lines = result[host].split('\n')
for line in lines:
if display_summary_line(line):
continue
else:
try:
item = json.loads(
line, object_hook=JsonHelper.decode_dict)
if item['empty'] is not True:
self.add_stats(item, self.summary)
except:
print "WARNING: failed to json load from host",
print host, "this line:", line
self.display_host_summary()
except:
print "WARNING: failed to process output from host"
else:
if self.verbose:
print "WARNING: no output from host", host
def display_remote_host(self, result):
try:
lines = result.split('\n')
files = []
for line in lines:
if line == "":
continue
elif line.startswith("WARNING:") or line.startswith("INFO:"):
print line
else:
files.append(json.loads(line, object_hook=JsonHelper.decode_dict))
if files == []:
return
path_justify = max([len(finfo['path']) for finfo in files]) + 2
for finfo in files:
self.display_from_dict(finfo, self.show_sample_content, path_justify)
except:
print "WARNING: failed to load json from host"
def get_local_auditor(self):
return LocalFilesAuditor(self.audit_type, self.confdir,
self.show_sample_content, self.dirsizes,
self.depth, self.to_check, self.ignore_also,
self.max_files)
def audit_hosts(self):
# do local audit instead
if self.localaudit:
localauditor = self.get_local_auditor()
result = localauditor.do_local_audit()
self.display_remote_host(result)
return
# proceed to regular remote audit
self.set_up_runner()
result = self.runner.run_remotely()
if result is None:
print "WARNING: failed to get output from audit script on any host"
elif self.show_summary:
self.display_summary(result)
else:
for host in result:
print "host:", host
if result[host]:
self.display_remote_host(result[host])
else:
if self.verbose:
print "no output from host", host
# add some results to rule store
self.update_status_rules_from_report(result)
return result
def update_status_rules_from_report(self, report):
hostlist = report.keys()
for host in hostlist:
try:
problem_rules = clouseau.retention.utils.ruleutils.get_rules(
self.cdb, host, Status.text_to_status('problem'))
except:
print 'WARNING: problem retrieving problem rules for host', host
problem_rules = None
if problem_rules is not None:
existing_problems = [rule['path'] for rule in problem_rules]
else:
existing_problems = []
dirs_problem, dirs_skipped = get_dirs_toexamine(report[host])
if dirs_problem is not None:
dirs_problem = list(set(dirs_problem))
for dirname in dirs_problem:
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, dirname,
clouseau.retention.utils.ruleutils.text_to_entrytype('dir'),
Status.text_to_status('problem'), host)
if dirs_skipped is not None:
dirs_skipped = list(set(dirs_skipped))
for dirname in dirs_skipped:
if dirname in dirs_problem or dirname in existing_problems:
# problem report overrides 'too many to audit'
continue
clouseau.retention.utils.ruleutils.do_add_rule(
self.cdb, dirname,
clouseau.retention.utils.ruleutils.text_to_entrytype('dir'),
Status.text_to_status('unreviewed'), host)