#!/usr/bin/env python
# Copyright (c) 2015 Red Hat, Inc.
# This file is part of GlusterFS.
#
# This file is licensed to you under your choice of the GNU Lesser
# General Public License, version 3 or any later version (LGPLv3 or
# later), or the GNU General Public License, version 2 (GPLv2), in all
# cases as published by the Free Software Foundation.
import sys
from errno import ENOENT
import time
from multiprocessing import Process
import os
import xml.etree.cElementTree as etree
from argparse import ArgumentParser, RawDescriptionHelpFormatter, Action
import logging
import shutil
from utils import execute, is_host_local, mkdirp, fail
from utils import setup_logger, human_time, handle_rm_error
from utils import get_changelog_rollover_time, cache_output
import conf
PROG_DESCRIPTION = """
GlusterFS Incremental API
"""
ParseError = etree.ParseError if hasattr(etree, 'ParseError') else SyntaxError
logger = logging.getLogger()
node_outfiles = []
class StoreAbsPath(Action):
def __init__(self, option_strings, dest, nargs=None, **kwargs):
super(StoreAbsPath, self).__init__(option_strings, dest, **kwargs)
def __call__(self, parser, namespace, values, option_string=None):
setattr(namespace, self.dest, os.path.abspath(values))
def get_pem_key_path(session, volume):
return os.path.join(conf.get_opt("session_dir"),
session,
volume,
"%s_%s_secret.pem" % (session, volume))
def node_cmd(host, host_uuid, task, cmd, args, opts):
"""
Runs command via ssh if host is not local
"""
localdir = is_host_local(host_uuid)
pem_key_path = get_pem_key_path(args.session, args.volume)
if not localdir:
# prefix with ssh command if not local node
cmd = ["ssh",
"-i", pem_key_path,
"root@%s" % host] + cmd
execute(cmd, exit_msg="%s - %s failed" % (host, task), logger=logger)
if opts.get("copy_outfile", False):
cmd_copy = ["scp",
"-i", pem_key_path,
"root@%s:/%s" % (host, opts.get("node_outfile")),
os.path.dirname(opts.get("node_outfile"))]
execute(cmd_copy, exit_msg="%s - Copy command failed" % host,
logger=logger)
def run_cmd_nodes(task, args, **kwargs):
global node_outfiles
nodes = get_nodes(args.volume)
pool = []
for num, node in enumerate(nodes):
host, brick = node[1].split(":")
host_uuid = node[0]
cmd = []
opts = {}
node_outfile = os.path.join(conf.get_opt("working_dir"),
args.session, args.volume,
"tmp_output_%s" % num)
if task == "pre":
# If Full backup is requested or start time is zero, use brickfind
change_detector = conf.get_change_detector("changelog")
if args.full:
change_detector = conf.get_change_detector("brickfind")
node_outfiles.append(node_outfile)
cmd = [change_detector,
args.session,
args.volume,
brick,
node_outfile,
str(kwargs.get("start")),
"--output-prefix",
args.output_prefix] + \
(["--debug"] if args.debug else []) + \
(["--only-namespace-changes"] if args.only_namespace_changes
else [])
opts["node_outfile"] = node_outfile
opts["copy_outfile"] = True
elif task == "cleanup":
# After pre run, cleanup the working directory and other temp files
# Remove the copied node_outfile in main node
try:
os.remove(node_outfile)
except (OSError, IOError):
logger.warn("Failed to cleanup temporary file %s" %
node_outfile)
pass
cmd = [conf.get_opt("nodeagent"),
"cleanup",
args.session,
args.volume] + (["--debug"] if args.debug else [])
elif task == "create":
# When glusterfind create, create session directory in
# each brick nodes
cmd = [conf.get_opt("nodeagent"),
"create",
args.session,
args.volume,
brick,
kwargs.get("time_to_update")] + \
(["--debug"] if args.debug else []) + \
(["--reset-session-time"] if args.reset_session_time
else [])
elif task == "post":
# Rename pre status file to actual status file in each node
cmd = [conf.get_opt("nodeagent"),
"post",
args.session,
args.volume,
brick] + \
(["--debug"] if args.debug else [])
elif task == "delete":
# When glusterfind delete, cleanup all the session files/dirs
# from each node.
cmd = [conf.get_opt("nodeagent"),
"delete",
args.session,
args.volume] + \
(["--debug"] if args.debug else [])
if cmd:
p = Process(target=node_cmd,
args=(host, host_uuid, task, cmd, args, opts))
p.start()
pool.append(p)
for num, p in enumerate(pool):
p.join()
if p.exitcode != 0:
logger.warn("Command %s failed in %s" % (task, nodes[num][1]))
if task in ["create", "delete"]:
fail("Command %s failed in %s" % (task, nodes[num][1]))
elif task == "pre" and args.disable_partial:
sys.exit(1)
@cache_output
def get_nodes(volume):
"""
Get the gluster volume info xml output and parse to get
the brick details.
"""
cmd = ["gluster", 'volume', 'info', volume, "--xml"]
_, data, _ = execute(cmd,
exit_msg="Failed to Run Gluster Volume Info",
logger=logger)
tree = etree.fromstring(data)
nodes = []
volume_el = tree.find('volInfo/volumes/volume')
try:
for b in volume_el.findall('bricks/brick'):
nodes.append((b.find('hostUuid').text,
b.find('name').text))
except (ParseError, AttributeError, ValueError) as e:
fail("Failed to parse Volume Info: %s" % e, logger=logger)
return nodes
def _get_args():
parser = ArgumentParser(formatter_class=RawDescriptionHelpFormatter,
description=PROG_DESCRIPTION)
subparsers = parser.add_subparsers(dest="mode")
# create [--debug] [--force]
parser_create = subparsers.add_parser('create')
parser_create.add_argument("session", help="Session Name")
parser_create.add_argument("volume", help="Volume Name")
parser_create.add_argument("--debug", help="Debug", action="store_true")
parser_create.add_argument("--force", help="Force option to recreate "
"the session", action="store_true")
parser_create.add_argument("--reset-session-time",
help="Reset Session Time to Current Time",
action="store_true")
# delete [--debug]
parser_delete = subparsers.add_parser('delete')
parser_delete.add_argument("session", help="Session Name")
parser_delete.add_argument("volume", help="Volume Name")
parser_delete.add_argument("--debug", help="Debug", action="store_true")
# list [--session ] [--volume ]
parser_list = subparsers.add_parser('list')
parser_list.add_argument("--session", help="Session Name", default="")
parser_list.add_argument("--volume", help="Volume Name", default="")
parser_list.add_argument("--debug", help="Debug", action="store_true")
# pre
# [--output-prefix ] [--full]
parser_pre = subparsers.add_parser('pre')
parser_pre.add_argument("session", help="Session Name")
parser_pre.add_argument("volume", help="Volume Name")
parser_pre.add_argument("outfile", help="Output File", action=StoreAbsPath)
parser_pre.add_argument("--debug", help="Debug", action="store_true")
parser_pre.add_argument("--full", help="Full find", action="store_true")
parser_pre.add_argument("--disable-partial", help="Disable Partial find, "
"Fail when one node fails", action="store_true")
parser_pre.add_argument("--output-prefix", help="File prefix in output",
default=".")
parser_pre.add_argument("--regenerate-outfile",
help="Regenerate outfile, discard the outfile "
"generated from last pre command",
action="store_true")
# post
parser_post = subparsers.add_parser('post')
parser_post.add_argument("session", help="Session Name")
parser_post.add_argument("volume", help="Volume Name")
parser_post.add_argument("--debug", help="Debug", action="store_true")
return parser.parse_args()
def ssh_setup(args):
pem_key_path = get_pem_key_path(args.session, args.volume)
if not os.path.exists(pem_key_path):
# Generate ssh-key
cmd = ["ssh-keygen",
"-N",
"",
"-f",
pem_key_path]
execute(cmd,
exit_msg="Unable to generate ssh key %s"
% pem_key_path,
logger=logger)
logger.info("Ssh key generated %s" % pem_key_path)
try:
shutil.copyfile(pem_key_path + ".pub",
os.path.join(conf.get_opt("session_dir"),
".keys",
"%s_%s_secret.pem.pub" % (args.session,
args.volume)))
except (IOError, OSError) as e:
fail("Failed to copy public key to %s: %s"
% (os.path.join(conf.get_opt("session_dir"), ".keys"), e),
logger=logger)
# Copy pub file to all nodes
cmd = ["gluster",
"system::",
"copy",
"file",
"/glusterfind/.keys/%s.pub" % os.path.basename(pem_key_path)]
execute(cmd, exit_msg="Failed to distribute ssh keys", logger=logger)
logger.info("Distributed ssh key to all nodes of Volume")
# Add to authorized_keys file in each node
cmd = ["gluster",
"system::",
"execute",
"add_secret_pub",
"root",
"/glusterfind/.keys/%s.pub" % os.path.basename(pem_key_path)]
execute(cmd,
exit_msg="Failed to add ssh keys to authorized_keys file",
logger=logger)
logger.info("Ssh key added to authorized_keys of Volume nodes")
def mode_create(session_dir, args):
logger.debug("Init is called - Session: %s, Volume: %s"
% (args.session, args.volume))
execute(["gluster", "volume", "info", args.volume],
exit_msg="Unable to get volume details",
logger=logger)
mkdirp(session_dir, exit_on_err=True, logger=logger)
mkdirp(os.path.join(session_dir, args.volume), exit_on_err=True,
logger=logger)
status_file = os.path.join(session_dir, args.volume, "status")
if os.path.exists(status_file) and not args.force:
fail("Session %s already created" % args.session, logger=logger)
if not os.path.exists(status_file) or args.force:
ssh_setup(args)
execute(["gluster", "volume", "set",
args.volume, "build-pgfid", "on"],
exit_msg="Failed to set volume option build-pgfid on",
logger=logger)
logger.info("Volume option set %s, build-pgfid on" % args.volume)
execute(["gluster", "volume", "set",
args.volume, "changelog.changelog", "on"],
exit_msg="Failed to set volume option "
"changelog.changelog on", logger=logger)
logger.info("Volume option set %s, changelog.changelog on"
% args.volume)
# Add Rollover time to current time to make sure changelogs
# will be available if we use this time as start time
time_to_update = int(time.time()) + get_changelog_rollover_time(
args.volume)
run_cmd_nodes("create", args, time_to_update=str(time_to_update))
if not os.path.exists(status_file) or args.reset_session_time:
with open(status_file, "w", buffering=0) as f:
f.write(str(time_to_update))
sys.exit(0)
def mode_pre(session_dir, args):
"""
Read from Session file and write to session.pre file
"""
endtime_to_update = int(time.time()) - get_changelog_rollover_time(
args.volume)
status_file = os.path.join(session_dir, args.volume, "status")
status_file_pre = status_file + ".pre"
mkdirp(os.path.dirname(args.outfile), exit_on_err=True, logger=logger)
# If Pre status file exists and running pre command again
if os.path.exists(status_file_pre) and not args.regenerate_outfile:
fail("Post command is not run after last pre, "
"use --regenerate-outfile")
start = 0
try:
with open(status_file) as f:
start = int(f.read().strip())
except ValueError:
pass
except (OSError, IOError) as e:
fail("Error Opening Session file %s: %s"
% (status_file, e), logger=logger)
logger.debug("Pre is called - Session: %s, Volume: %s, "
"Start time: %s, End time: %s"
% (args.session, args.volume, start, endtime_to_update))
run_cmd_nodes("pre", args, start=start)
# Merger
cmd = ["sort", "-u"] + node_outfiles + ["-o", args.outfile]
execute(cmd,
exit_msg="Failed to merge output files "
"collected from nodes", logger=logger)
run_cmd_nodes("cleanup", args)
with open(status_file_pre, "w", buffering=0) as f:
f.write(str(endtime_to_update))
sys.stdout.write("Generated output file %s\n" % args.outfile)
def mode_post(session_dir, args):
"""
If pre session file exists, overwrite session file
If pre session file does not exists, return ERROR
"""
status_file = os.path.join(session_dir, args.volume, "status")
logger.debug("Post is called - Session: %s, Volume: %s"
% (args.session, args.volume))
status_file_pre = status_file + ".pre"
if os.path.exists(status_file_pre):
run_cmd_nodes("post", args)
os.rename(status_file_pre, status_file)
sys.exit(0)
else:
fail("Pre script is not run", logger=logger)
def mode_delete(session_dir, args):
run_cmd_nodes("delete", args)
shutil.rmtree(os.path.join(session_dir, args.volume),
onerror=handle_rm_error)
def mode_list(session_dir, args):
"""
List available sessions to stdout, if session name is set
only list that session.
"""
if args.session:
if not os.path.exists(os.path.join(session_dir, args.session)):
fail("Invalid Session", logger=logger)
sessions = [args.session]
else:
sessions = []
for d in os.listdir(session_dir):
if d != ".keys":
sessions.append(d)
output = []
for session in sessions:
# Session Volume Last Processed
volnames = os.listdir(os.path.join(session_dir, session))
for volname in volnames:
if args.volume and args.volume != volname:
continue
status_file = os.path.join(session_dir, session, volname, "status")
last_processed = None
try:
with open(status_file) as f:
last_processed = f.read().strip()
except (OSError, IOError) as e:
if e.errno == ENOENT:
pass
else:
raise
output.append((session, volname, last_processed))
if output:
sys.stdout.write("%s %s %s\n" % ("SESSION".ljust(25),
"VOLUME".ljust(25),
"SESSION TIME".ljust(25)))
sys.stdout.write("-"*75)
sys.stdout.write("\n")
for session, volname, last_processed in output:
sess_time = 'Session Corrupted'
if last_processed:
try:
sess_time = human_time(last_processed)
except TypeError:
sess_time = 'Session Corrupted'
sys.stdout.write("%s %s %s\n" % (session.ljust(25),
volname.ljust(25),
sess_time.ljust(25)))
if not output and (args.session or args.volume):
fail("Invalid Session", logger=logger)
def main():
args = _get_args()
mkdirp(conf.get_opt("session_dir"), exit_on_err=True)
if args.mode == "list":
session_dir = conf.get_opt("session_dir")
else:
session_dir = os.path.join(conf.get_opt("session_dir"),
args.session)
if not os.path.exists(session_dir) and args.mode not in ["create", "list"]:
fail("Invalid session %s" % args.session)
mkdirp(os.path.join(conf.get_opt("log_dir"), args.session, args.volume),
exit_on_err=True)
log_file = os.path.join(conf.get_opt("log_dir"),
args.session,
args.volume,
"cli.log")
setup_logger(logger, log_file, args.debug)
# globals() will have all the functions already defined.
# mode_ will be the function name to be called
globals()["mode_" + args.mode](session_dir, args)