1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
|
#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# Copyright (c) 2014-2015, NORDUnet A/S.
# See LICENSE for licensing information.
#
# Copy entries indicated by file 'fetched' to all secondary merge nodes.
# See catlfish/doc/merge.txt for more about the merge process.
#
import sys
import select
import requests
import errno
import logging
from time import sleep
from base64 import b64encode, b64decode
from os import stat
from multiprocessing import Process, Pipe, active_children
from certtools import timing_point, build_merkle_tree, write_file, \
create_ssl_context
from mergetools import chunks, backup_sendlog, get_logorder, \
get_verifiedsize, get_missingentriesforbackup, \
hexencode, setverifiedsize, sendentries_merge, verifyroot, \
get_nfetched, parse_args, perm, waitforfile, flock_ex_or_fail, \
Status, loginit
def backup_loop(nodename, nodeaddress, own_key, paths, verifiedsize, chunk):
for trynumber in range(5, 0, -1):
sendlogresult = \
backup_sendlog(nodename, nodeaddress, own_key, paths,
{"start": verifiedsize, "hashes": chunk})
if sendlogresult == None:
if trynumber == 1:
return None
select.select([], [], [], 10.0)
logging.info("tries left: %d", trynumber)
continue
return sendlogresult
sendlog_discover_chunksize = 100000
def sendlog_helper(entries, verifiedsize, nodename, nodeaddress, own_key, paths,
statusupdates):
logging.info("sending log")
for chunk in chunks(entries, 1000):
sendlogresult = backup_loop(nodename, nodeaddress, own_key, paths, verifiedsize, chunk)
if sendlogresult == None:
sys.exit(1)
if sendlogresult["result"] != "ok":
logging.error("backup_sendlog: %s", sendlogresult)
sys.exit(1)
verifiedsize += len(chunk)
statusupdates.status("PROG sending log: %d" % verifiedsize)
logging.info("log sent")
def fill_in_missing_entries(nodename, nodeaddress, own_key, paths, chainsdb,
timing, statusupdates):
missingentries = get_missingentriesforbackup(nodename, nodeaddress,
own_key, paths)
timing_point(timing, "get missing")
while missingentries:
logging.info("about to send %d missing entries", len(missingentries))
fetched_entries = 0
with requests.sessions.Session() as session:
for missingentry_chunk in chunks(missingentries, 100):
missingentry_hashes = [b64decode(missingentry) for missingentry in missingentry_chunk]
hashes_and_entries = [(ehash, chainsdb.get(ehash)) for ehash in missingentry_hashes]
sendentryresult = sendentries_merge(nodename, nodeaddress,
own_key, paths,
hashes_and_entries, session)
if sendentryresult["result"] != "ok":
logging.error("sendentries_merge: %s", sendentryresult)
sys.exit(1)
fetched_entries += len(missingentry_hashes)
statusupdates.status("PROG sending missing entries: %d" %
fetched_entries)
timing_point(timing, "send missing")
missingentries = get_missingentriesforbackup(nodename, nodeaddress,
own_key, paths)
timing_point(timing, "get missing")
def check_root(logorder, nodename, nodeaddress, own_key, paths, tree_size, timing):
tree = build_merkle_tree(logorder[:tree_size])
root_hash = tree[-1][0]
timing_point(timing, "build tree")
verifyrootresult = verifyroot(nodename, nodeaddress, own_key, paths,
tree_size)
if verifyrootresult["result"] != "ok":
logging.error("verifyroot: %s", verifyrootresult)
sys.exit(1)
secondary_root_hash = b64decode(verifyrootresult["root_hash"])
if root_hash != secondary_root_hash:
logging.error("secondary root hash was %s while expected was %s",
hexencode(secondary_root_hash), hexencode(root_hash))
sys.exit(1)
timing_point(timing, "verifyroot")
return root_hash
def do_it(backupargs):
secondary, localconfig, chainsdb, logorder, s, timing = backupargs
maxwindow = localconfig.get("maxwindow", 1000)
paths = localconfig["paths"]
nodename = secondary["name"]
nodeaddress = "https://%s/" % secondary["address"]
own_key = (localconfig["nodename"],
"%s/%s-private.pem" % (paths["privatekeys"],
localconfig["nodename"]))
tree_size = len(logorder)
logging.info("backing up to node %s", nodename)
verifiedsize = get_verifiedsize(nodename, nodeaddress, own_key, paths)
timing_point(timing, "get verified size")
logging.info("verified size %d", verifiedsize)
if verifiedsize == tree_size:
root_hash = check_root(logorder, nodename, nodeaddress, own_key, paths, tree_size, timing)
else:
while verifiedsize < tree_size:
uptopos = min(verifiedsize + maxwindow, tree_size)
entries = [b64encode(entry) for entry in logorder[verifiedsize:uptopos]]
sendlog_helper(entries, verifiedsize, nodename, nodeaddress, own_key, paths, s)
timing_point(timing, "sendlog")
fill_in_missing_entries(nodename, nodeaddress, own_key, paths, chainsdb, timing, s)
root_hash = check_root(logorder, nodename, nodeaddress, own_key, paths, uptopos, timing)
verifiedsize = uptopos
setverifiedsize(nodename, nodeaddress, own_key, paths, verifiedsize)
return root_hash
def worker(pipe, backupargs):
root_hash = do_it(backupargs)
pipe.send(root_hash)
return 0
def start_worker(backupargs):
_, _, _, nodename, _, _, _, _, _, _, _ = backupargs
parent_conn, child_conn = Pipe()
p = Process(target=worker,
args=(child_conn, backupargs),
name='backup_%s' % nodename)
p.start()
return p, parent_conn
def update_backupfile(mergedb, nodename, tree_size, root_hash):
backuppath = mergedb + "/verified." + nodename
backupdata = {"tree_size": tree_size,
"sha256_root_hash": hexencode(root_hash)}
logging.debug("writing to %s: %s", backuppath, backupdata)
write_file(backuppath, backupdata)
def merge_backup(args, config, localconfig, secondaries):
failures = 0
paths = localconfig["paths"]
mergedb = paths["mergedb"]
chainsdb = perm(localconfig.get("dbbackend", "filedb"), mergedb + "/chains")
logorderfile = mergedb + "/logorder"
currentsizefile = mergedb + "/fetched"
statusfile = mergedb + "/merge_backup.status"
s = Status(statusfile)
timing = timing_point()
nfetched = get_nfetched(currentsizefile, logorderfile)
timing_point(timing, "get nfetched")
logorder = get_logorder(logorderfile, nfetched)
tree_size = len(logorder)
timing_point(timing, "get logorder")
procs = {}
for secondary in secondaries:
if secondary["name"] == config["primarymergenode"]:
continue
nodename = secondary["name"]
timing = timing_point()
backupargs = (secondary, localconfig, chainsdb, logorder, s, timing)
if args.mergeinterval:
pipe_mine, pipe_theirs = Pipe()
p = Process(target=worker,
args=(pipe_theirs, backupargs),
name='backup_%s' % nodename)
p.start()
procs[p] = (nodename, pipe_mine)
else:
root_hash = do_it(backupargs)
update_backupfile(mergedb, nodename, tree_size, root_hash)
if args.mergeinterval:
while True:
for p in list(procs):
if not p.is_alive():
p.join()
nodename, pipe = procs[p]
if p.exitcode == 0:
root_hash = pipe.recv()
update_backupfile(mergedb, nodename, tree_size, root_hash)
else:
logging.warning("%s failure: %d", nodename, p.exitcode)
failures += 1
del procs[p]
if not procs:
break
sleep(1)
if args.timing:
logging.debug("timing: merge_backup: %s", timing["deltatimes"])
return failures
def main():
"""
Wait until 'fetched' exists and read it.
Read 'logorder' up until what's indicated by 'fetched' and build the
tree.
Distribute entries to all secondaries, in parallel if `--mergeinterval'.
Write tree size and tree head to 'backup.<secondary>' files as each
secondary is verified to have the entries.
If `--mergeinterval', wait until 'fetched' is updated and read it
and start over from the point where 'logorder' is read.
"""
args, config, localconfig = parse_args()
paths = localconfig["paths"]
mergedb = paths["mergedb"]
lockfile = mergedb + "/.merge_backup.lock"
fetched_path = mergedb + "/fetched"
loginit(args, "merge_backup.log")
if not flock_ex_or_fail(lockfile):
logging.critical("unable to take lock %s", lockfile)
return 1
all_secondaries = \
[n for n in config.get('mergenodes', []) if \
n['name'] != config['primarymergenode']]
if len(args.node) == 0:
nodes = all_secondaries
else:
nodes = [n for n in all_secondaries if n["name"] in args.node]
create_ssl_context(cafile=paths["https_cacertfile"])
fetched_statinfo = waitforfile(fetched_path)
retval = 0
while True:
failures = merge_backup(args, config, localconfig, nodes)
if not args.mergeinterval:
break
fetched_statinfo_old = fetched_statinfo
while fetched_statinfo == fetched_statinfo_old:
sleep(max(3, args.mergeinterval / 10))
if failures > 0:
break
fetched_statinfo = stat(fetched_path)
return retval
if __name__ == '__main__':
sys.exit(main())
|
