-
Notifications
You must be signed in to change notification settings - Fork 0
/
store.cc
392 lines (320 loc) · 11 KB
/
store.cc
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
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
/* Cumulus: Efficient Filesystem Backup to the Cloud
* Copyright (C) 2008-2009 The Cumulus Developers
* See the AUTHORS file for a list of contributors.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*/
/* Backup data is stored in a collection of objects, which are grouped together
* into segments for storage purposes. This implementation of the object store
* represents segments as TAR files and objects as files within them. */
#include <assert.h>
#include <errno.h>
#include <stdio.h>
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/resource.h>
#include <sys/wait.h>
#include <unistd.h>
#include <fcntl.h>
#include <time.h>
#include <algorithm>
#include <list>
#include <map>
#include <set>
#include <string>
#include <iostream>
#include "hash.h"
#include "localdb.h"
#include "store.h"
#include "ref.h"
#include "util.h"
using std::max;
using std::list;
using std::map;
using std::pair;
using std::set;
using std::string;
/* Default filter program is bzip2 */
const char *filter_program = "bzip2 -c";
const char *filter_extension = ".bz2";
Tarfile::Tarfile(RemoteFile *file, const string &segment)
: size(0),
segment_name(segment)
{
assert(sizeof(struct tar_header) == TAR_BLOCK_SIZE);
this->file = file;
this->filter.reset(FileFilter::New(file->get_fd(), filter_program));
}
Tarfile::~Tarfile()
{
char buf[TAR_BLOCK_SIZE];
/* Append the EOF marker: two blocks filled with nulls. */
memset(buf, 0, sizeof(buf));
tar_write(buf, TAR_BLOCK_SIZE);
tar_write(buf, TAR_BLOCK_SIZE);
if (close(filter->get_wrapped_fd()) != 0)
fatal("Error closing Tarfile");
/* ...and wait for filter process to finish. */
if (filter->wait() != 0) {
fatal("Filter process error");
}
}
FileFilter::FileFilter(int raw, int wrapped, pid_t pid)
: fd_raw(raw), fd_wrapped(wrapped), pid(pid) { }
FileFilter *FileFilter::New(int fd, const char *program)
{
if (program == NULL || strlen(program) == 0) {
return new FileFilter(fd, fd, -1);
}
pid_t pid;
int wrapped_fd = spawn_filter(fd, program, &pid);
return new FileFilter(fd, wrapped_fd, pid);
}
int FileFilter::wait()
{
// No filter program was launched implies no need to wait.
if (pid == -1)
return 0;
// The raw file descriptor was held open to track the output file size, but
// is not needed any longer.
close(fd_raw);
int status;
if (waitpid(pid, &status, 0) < 0) {
fprintf(stderr, "Error waiting for filter process: %m\n");
return -1;
}
if (!WIFEXITED(status) || WEXITSTATUS(status) != 0) {
fprintf(stderr, "Filter process error: %d\n", WEXITSTATUS(status));
}
return status;
}
/* Launch a child process which can act as a filter (compress, encrypt, etc.)
* on the TAR output. The file descriptor to which output should be written
* must be specified; the return value is the file descriptor which will be
* attached to the standard input of the filter program. */
int FileFilter::spawn_filter(int fd_out, const char *program, pid_t *filter_pid)
{
int fds[2];
pid_t pid;
/* Create a pipe for communicating with the filter process. */
if (pipe(fds) < 0) {
fatal("Unable to create pipe for filter");
}
/* Create a child process which can exec() the filter program. */
pid = fork();
if (pid < 0)
fatal("Unable to fork filter process");
if (pid > 0) {
/* Parent process */
close(fds[0]);
cloexec(fds[1]);
if (filter_pid != NULL)
*filter_pid = pid;
} else {
/* Child process. Rearrange file descriptors. stdin is fds[0], stdout
* is fd_out, stderr is unchanged. */
close(fds[1]);
if (dup2(fds[0], 0) < 0)
exit(1);
close(fds[0]);
if (dup2(fd_out, 1) < 0)
exit(1);
close(fd_out);
/* Exec the filter program. */
execlp("/bin/sh", "/bin/sh", "-c", program, NULL);
/* Should not reach here except for error cases. */
fprintf(stderr, "Could not exec filter: %m\n");
exit(1);
}
return fds[1];
}
void Tarfile::tar_write(const char *data, size_t len)
{
size += len;
while (len > 0) {
int res = write(filter->get_wrapped_fd(), data, len);
if (res < 0) {
if (errno == EINTR)
continue;
fprintf(stderr, "Write error: %m\n");
fatal("Write error");
}
len -= res;
data += res;
}
}
void Tarfile::write_object(int id, const char *data, size_t len)
{
struct tar_header header;
memset(&header, 0, sizeof(header));
char buf[64];
sprintf(buf, "%08x", id);
string path = segment_name + "/" + buf;
assert(path.size() < 100);
memcpy(header.name, path.data(), path.size());
sprintf(header.mode, "%07o", 0600);
sprintf(header.uid, "%07o", 0);
sprintf(header.gid, "%07o", 0);
sprintf(header.size, "%011o", (int)len);
sprintf(header.mtime, "%011o", (int)time(NULL));
header.typeflag = '0';
strcpy(header.magic, "ustar ");
strcpy(header.uname, "root");
strcpy(header.gname, "root");
memset(header.chksum, ' ', sizeof(header.chksum));
int checksum = 0;
for (int i = 0; i < TAR_BLOCK_SIZE; i++) {
checksum += ((uint8_t *)&header)[i];
}
sprintf(header.chksum, "%06o", checksum);
tar_write((const char *)&header, TAR_BLOCK_SIZE);
if (len == 0)
return;
tar_write(data, len);
char padbuf[TAR_BLOCK_SIZE];
size_t blocks = (len + TAR_BLOCK_SIZE - 1) / TAR_BLOCK_SIZE;
size_t padding = blocks * TAR_BLOCK_SIZE - len;
memset(padbuf, 0, padding);
tar_write(padbuf, padding);
}
/* Estimate the size based on the size of the actual output file on disk.
* However, it might be the case that the filter program is buffering all its
* data, and might potentially not write a single byte until we have closed
* our end of the pipe. If we don't do so until we see data written, we have
* a problem. So, arbitrarily pick an upper bound on the compression ratio
* that the filter will achieve (128:1), and return a size estimate which is
* the larger of a) bytes actually seen written to disk, and b) input
* bytes/128. */
size_t Tarfile::size_estimate()
{
struct stat statbuf;
if (fstat(filter->get_raw_fd(), &statbuf) == 0)
return max((int64_t)statbuf.st_size, (int64_t)(size / 128));
/* Couldn't stat the file on disk, so just return the actual number of
* bytes, before compression. */
return size;
}
static const size_t SEGMENT_SIZE = 4 * 1024 * 1024;
/* Backup size summary: segment type -> (uncompressed size, compressed size) */
static map<string, pair<int64_t, int64_t> > group_sizes;
ObjectReference TarSegmentStore::write_object(const char *data, size_t len,
const std::string &group,
const std::string &checksum,
double age)
{
struct segment_info *segment;
// Find the segment into which the object should be written, looking up by
// group. If no segment exists yet, create one.
if (segments.find(group) == segments.end()) {
segment = new segment_info;
segment->name = generate_uuid();
segment->group = group;
segment->basename = segment->name + ".tar";
segment->basename += filter_extension;
segment->count = 0;
segment->data_size = 0;
segment->rf = remote->alloc_file(segment->basename,
group == "metadata" ? "segments0"
: "segments1");
segment->file = new Tarfile(segment->rf, segment->name);
segments[group] = segment;
} else {
segment = segments[group];
}
int id = segment->count;
char id_buf[64];
sprintf(id_buf, "%08x", id);
segment->file->write_object(id, data, len);
segment->count++;
segment->data_size += len;
group_sizes[group].first += len;
ObjectReference ref(segment->name, id_buf);
ref.set_range(0, len, true);
if (checksum.size() > 0)
ref.set_checksum(checksum);
if (db != NULL)
db->StoreObject(ref, age);
// If this segment meets or exceeds the size target, close it so that
// future objects will go into a new segment.
if (segment->file->size_estimate() >= SEGMENT_SIZE)
close_segment(group);
return ref;
}
void TarSegmentStore::sync()
{
while (!segments.empty())
close_segment(segments.begin()->first);
}
void TarSegmentStore::dump_stats()
{
printf("Data written:\n");
for (map<string, pair<int64_t, int64_t> >::iterator i = group_sizes.begin();
i != group_sizes.end(); ++i) {
printf(" %s: %lld (%lld compressed)\n", i->first.c_str(),
(long long)i->second.first, (long long)i->second.second);
}
}
void TarSegmentStore::close_segment(const string &group)
{
struct segment_info *segment = segments[group];
delete segment->file;
if (db != NULL) {
struct stat stat_buf;
int disk_size = 0;
if (stat(segment->rf->get_local_path().c_str(), &stat_buf) == 0) {
disk_size = stat_buf.st_size;
group_sizes[segment->group].second += disk_size;
}
string checksum
= Hash::hash_file(segment->rf->get_local_path().c_str());
db->SetSegmentMetadata(segment->name, segment->rf->get_remote_path(),
checksum, group, segment->data_size, disk_size);
}
segment->rf->send();
segments.erase(segments.find(group));
delete segment;
}
string TarSegmentStore::object_reference_to_segment(const string &object)
{
return object;
}
LbsObject::LbsObject()
: group(""), age(0.0), data(NULL), data_len(0), written(false)
{
}
LbsObject::~LbsObject()
{
}
void LbsObject::set_data(const char *d, size_t len, const char *checksum)
{
data = d;
data_len = len;
if (checksum != NULL) {
this->checksum = checksum;
} else {
Hash *hash = Hash::New();
hash->update(data, data_len);
this->checksum = hash->digest_str();
delete hash;
}
}
void LbsObject::write(TarSegmentStore *store)
{
assert(data != NULL);
assert(!written);
ref = store->write_object(data, data_len, group, checksum, age);
written = true;
}