2 Bacula® - The Network Backup Solution
4 Copyright (C) 2000-2007 Free Software Foundation Europe e.V.
6 The main author of Bacula is Kern Sibbald, with contributions from
7 many others, a complete list can be found in the file AUTHORS.
8 This program is Free Software; you can redistribute it and/or
9 modify it under the terms of version two of the GNU General Public
10 License as published by the Free Software Foundation plus additions
11 that are listed in the file LICENSE.
13 This program is distributed in the hope that it will be useful, but
14 WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
23 Bacula® is a registered trademark of John Walker.
24 The licensor of Bacula is the Free Software Foundation Europe
25 (FSFE), Fiduciary Program, Sumatrastrasse 25, 8006 Zürich,
26 Switzerland, email:ftf@fsfeurope.org.
30 * Bacula Director -- verify.c -- responsible for running file verification
32 * Kern Sibbald, October MM
34 * Basic tasks done here:
36 * Open connection with File daemon and pass him commands
38 * When the File daemon sends the attributes, compare them to
47 #include "findlib/find.h"
49 /* Commands sent to File daemon */
50 static char verifycmd[] = "verify level=%s\n";
51 static char storaddr[] = "storage address=%s port=%d ssl=0\n";
53 /* Responses received from File daemon */
54 static char OKverify[] = "2000 OK verify\n";
55 static char OKstore[] = "2000 OK storage\n";
56 static char OKbootstrap[] = "2000 OK bootstrap\n";
58 /* Forward referenced functions */
59 static void prt_fname(JCR *jcr);
60 static int missing_handler(void *ctx, int num_fields, char **row);
64 * Called here before the job is run to do the job
67 bool do_verify_init(JCR *jcr)
74 * Do a verification of the specified files against the Catlaog
76 * Returns: false on failure
79 bool do_verify(JCR *jcr)
86 JobId_t verify_jobid = 0;
89 free_wstorage(jcr); /* we don't write */
91 memset(&jcr->previous_jr, 0, sizeof(jcr->previous_jr));
94 * Find JobId of last job that ran. Note, we do this when
95 * the job actually starts running, not at schedule time,
96 * so that we find the last job that terminated before
97 * this job runs rather than before it is scheduled. This
98 * permits scheduling a Backup and Verify at the same time,
99 * but with the Verify at a lower priority.
101 * For VERIFY_CATALOG we want the JobId of the last INIT.
102 * For VERIFY_VOLUME_TO_CATALOG, we want the JobId of the
105 if (jcr->JobLevel == L_VERIFY_CATALOG ||
106 jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
107 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG) {
108 memcpy(&jr, &jcr->jr, sizeof(jr));
109 if (jcr->verify_job &&
110 (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
111 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG)) {
112 Name = jcr->verify_job->name();
116 Dmsg1(100, "find last jobid for: %s\n", NPRT(Name));
117 if (!db_find_last_jobid(jcr, jcr->db, Name, &jr)) {
118 if (jcr->JobLevel == L_VERIFY_CATALOG) {
119 Jmsg(jcr, M_FATAL, 0, _(
120 "Unable to find JobId of previous InitCatalog Job.\n"
121 "Please run a Verify with Level=InitCatalog before\n"
122 "running the current Job.\n"));
124 Jmsg(jcr, M_FATAL, 0, _(
125 "Unable to find JobId of previous Job for this client.\n"));
129 verify_jobid = jr.JobId;
130 Dmsg1(100, "Last full jobid=%d\n", verify_jobid);
133 * Now get the job record for the previous backup that interests
134 * us. We use the verify_jobid that we found above.
136 if (jcr->JobLevel == L_VERIFY_CATALOG ||
137 jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
138 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG) {
139 jcr->previous_jr.JobId = verify_jobid;
140 if (!db_get_job_record(jcr, jcr->db, &jcr->previous_jr)) {
141 Jmsg(jcr, M_FATAL, 0, _("Could not get job record for previous Job. ERR=%s"),
142 db_strerror(jcr->db));
145 if (jcr->previous_jr.JobStatus != 'T') {
146 Jmsg(jcr, M_FATAL, 0, _("Last Job %d did not terminate normally. JobStatus=%c\n"),
147 verify_jobid, jcr->previous_jr.JobStatus);
150 Jmsg(jcr, M_INFO, 0, _("Verifying against JobId=%d Job=%s\n"),
151 jcr->previous_jr.JobId, jcr->previous_jr.Job);
155 * If we are verifying a Volume, we need the Storage
156 * daemon, so open a connection, otherwise, just
157 * create a dummy authorization key (passed to
158 * File daemon but not used).
160 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
161 if (!create_restore_bootstrap_file(jcr)) {
165 jcr->sd_auth_key = bstrdup("dummy"); /* dummy Storage daemon key */
168 if (jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG && jcr->verify_job) {
169 jcr->fileset = jcr->verify_job->fileset;
171 Dmsg2(100, "ClientId=%u JobLevel=%c\n", jcr->previous_jr.ClientId, jcr->JobLevel);
173 if (!db_update_job_start_record(jcr, jcr->db, &jcr->jr)) {
174 Jmsg(jcr, M_FATAL, 0, "%s", db_strerror(jcr->db));
178 /* Print Job Start message */
179 Jmsg(jcr, M_INFO, 0, _("Start Verify JobId=%s Level=%s Job=%s\n"),
180 edit_uint64(jcr->JobId, ed1), level_to_str(jcr->JobLevel), jcr->Job);
182 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
184 * Start conversation with Storage daemon
186 set_jcr_job_status(jcr, JS_Blocked);
187 if (!connect_to_storage_daemon(jcr, 10, SDConnectTimeout, 1)) {
191 * Now start a job with the Storage daemon
193 if (!start_storage_daemon_job(jcr, jcr->rstorage, NULL)) {
196 if (!bnet_fsend(jcr->store_bsock, "run")) {
200 * Now start a Storage daemon message thread
202 if (!start_storage_daemon_message_thread(jcr)) {
205 Dmsg0(50, "Storage daemon connection OK\n");
209 * OK, now connect to the File daemon
210 * and ask him for the files.
212 set_jcr_job_status(jcr, JS_Blocked);
213 if (!connect_to_file_daemon(jcr, 10, FDConnectTimeout, 1)) {
217 set_jcr_job_status(jcr, JS_Running);
218 fd = jcr->file_bsock;
221 Dmsg0(30, ">filed: Send include list\n");
222 if (!send_include_list(jcr)) {
226 Dmsg0(30, ">filed: Send exclude list\n");
227 if (!send_exclude_list(jcr)) {
232 * Send Level command to File daemon, as well
233 * as the Storage address if appropriate.
235 switch (jcr->JobLevel) {
239 case L_VERIFY_CATALOG:
242 case L_VERIFY_VOLUME_TO_CATALOG:
244 * send Storage daemon address to the File daemon
246 if (jcr->rstore->SDDport == 0) {
247 jcr->rstore->SDDport = jcr->rstore->SDport;
249 bnet_fsend(fd, storaddr, jcr->rstore->address, jcr->rstore->SDDport);
250 if (!response(jcr, fd, OKstore, "Storage", DISPLAY_ERROR)) {
255 * Send the bootstrap file -- what Volumes/files to restore
257 if (!send_bootstrap_file(jcr, fd) ||
258 !response(jcr, fd, OKbootstrap, "Bootstrap", DISPLAY_ERROR)) {
262 if (!jcr->RestoreBootstrap) {
263 Jmsg0(jcr, M_FATAL, 0, _("Deprecated feature ... use bootstrap.\n"));
272 case L_VERIFY_DISK_TO_CATALOG:
273 level="disk_to_catalog";
276 Jmsg2(jcr, M_FATAL, 0, _("Unimplemented Verify level %d(%c)\n"), jcr->JobLevel,
281 if (!send_runscripts_commands(jcr)) {
286 * Send verify command/level to File daemon
288 bnet_fsend(fd, verifycmd, level);
289 if (!response(jcr, fd, OKverify, "Verify", DISPLAY_ERROR)) {
294 * Now get data back from File daemon and
295 * compare it to the catalog or store it in the
296 * catalog depending on the run type.
298 /* Compare to catalog */
299 switch (jcr->JobLevel) {
300 case L_VERIFY_CATALOG:
301 Dmsg0(10, "Verify level=catalog\n");
302 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
303 jcr->SDJobStatus = JS_Terminated;
304 get_attributes_and_compare_to_catalog(jcr, jcr->previous_jr.JobId);
307 case L_VERIFY_VOLUME_TO_CATALOG:
308 Dmsg0(10, "Verify level=volume\n");
309 get_attributes_and_compare_to_catalog(jcr, jcr->previous_jr.JobId);
312 case L_VERIFY_DISK_TO_CATALOG:
313 Dmsg0(10, "Verify level=disk_to_catalog\n");
314 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
315 jcr->SDJobStatus = JS_Terminated;
316 get_attributes_and_compare_to_catalog(jcr, jcr->previous_jr.JobId);
321 Dmsg0(10, "Verify level=init\n");
322 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
323 jcr->SDJobStatus = JS_Terminated;
324 get_attributes_and_put_in_catalog(jcr);
328 Jmsg1(jcr, M_FATAL, 0, _("Unimplemented verify level %d\n"), jcr->JobLevel);
332 stat = wait_for_job_termination(jcr);
333 if (stat == JS_Terminated) {
334 verify_cleanup(jcr, stat);
342 * Release resources allocated during backup.
345 void verify_cleanup(JCR *jcr, int TermCode)
347 char sdt[50], edt[50];
348 char ec1[30], ec2[30];
349 char term_code[100], fd_term_msg[100], sd_term_msg[100];
350 const char *term_msg;
355 // Dmsg1(100, "Enter verify_cleanup() TermCod=%d\n", TermCode);
357 Dmsg3(900, "JobLevel=%c Expected=%u JobFiles=%u\n", jcr->JobLevel,
358 jcr->ExpectedFiles, jcr->JobFiles);
359 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG &&
360 jcr->ExpectedFiles != jcr->JobFiles) {
361 TermCode = JS_ErrorTerminated;
364 /* If no files were expected, there can be no error */
365 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG &&
366 jcr->ExpectedFiles == 0) {
367 TermCode = JS_Terminated;
370 JobId = jcr->jr.JobId;
372 update_job_end(jcr, TermCode);
374 if (jcr->unlink_bsr && jcr->RestoreBootstrap) {
375 unlink(jcr->RestoreBootstrap);
376 jcr->unlink_bsr = false;
379 msg_type = M_INFO; /* by default INFO message */
382 term_msg = _("Verify OK");
385 case JS_ErrorTerminated:
386 term_msg = _("*** Verify Error ***");
387 msg_type = M_ERROR; /* Generate error message */
390 term_msg = _("Verify warnings");
393 term_msg = _("Verify Canceled");
396 term_msg = _("Verify Differences");
399 term_msg = term_code;
400 bsnprintf(term_code, sizeof(term_code),
401 _("Inappropriate term code: %d %c\n"), TermCode, TermCode);
404 bstrftimes(sdt, sizeof(sdt), jcr->jr.StartTime);
405 bstrftimes(edt, sizeof(edt), jcr->jr.EndTime);
406 if (jcr->verify_job) {
407 Name = jcr->verify_job->hdr.name;
412 jobstatus_to_ascii(jcr->FDJobStatus, fd_term_msg, sizeof(fd_term_msg));
413 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
414 jobstatus_to_ascii(jcr->SDJobStatus, sd_term_msg, sizeof(sd_term_msg));
415 Jmsg(jcr, msg_type, 0, _("Bacula %s %s (%s): %s\n"
416 " Build OS: %s %s %s\n"
420 " Verify Level: %s\n"
422 " Verify JobId: %d\n"
426 " Files Expected: %s\n"
427 " Files Examined: %s\n"
428 " Non-fatal FD errors: %d\n"
429 " FD termination status: %s\n"
430 " SD termination status: %s\n"
431 " Termination: %s\n\n"),
432 my_name, VERSION, LSMDATE, edt,
433 HOST_OS, DISTNAME, DISTVER,
436 jcr->fileset->hdr.name,
437 level_to_str(jcr->JobLevel),
438 jcr->client->hdr.name,
439 jcr->previous_jr.JobId,
443 edit_uint64_with_commas(jcr->ExpectedFiles, ec1),
444 edit_uint64_with_commas(jcr->JobFiles, ec2),
450 Jmsg(jcr, msg_type, 0, _("Bacula %s %s (%s): %s\n"
455 " Verify Level: %s\n"
457 " Verify JobId: %d\n"
461 " Files Examined: %s\n"
462 " Non-fatal FD errors: %d\n"
463 " FD termination status: %s\n"
464 " Termination: %s\n\n"),
465 my_name, VERSION, LSMDATE, edt,
466 HOST_OS, DISTNAME, DISTVER,
469 jcr->fileset->hdr.name,
470 level_to_str(jcr->JobLevel),
471 jcr->client->hdr.name,
472 jcr->previous_jr.JobId,
476 edit_uint64_with_commas(jcr->JobFiles, ec1),
481 Dmsg0(100, "Leave verify_cleanup()\n");
485 * This routine is called only during a Verify
487 int get_attributes_and_compare_to_catalog(JCR *jcr, JobId_t JobId)
492 struct stat statf; /* file stat */
493 struct stat statc; /* catalog stat */
494 int stat = JS_Terminated;
496 POOLMEM *fname = get_pool_memory(PM_MESSAGE);
497 int do_Digest = CRYPTO_DIGEST_NONE;
498 int32_t file_index = 0;
500 memset(&fdbr, 0, sizeof(FILE_DBR));
501 fd = jcr->file_bsock;
505 Dmsg0(20, "bdird: waiting to receive file attributes\n");
507 * Get Attributes and Signature from File daemon
511 * Options or Digest (MD5/SHA1)
516 while ((n=bget_dirmsg(fd)) >= 0 && !job_canceled(jcr)) {
519 char Opts_Digest[MAXSTRING]; /* Verify Opts or MD5/SHA1 digest */
521 if (job_canceled(jcr)) {
524 fname = check_pool_memory_size(fname, fd->msglen);
525 jcr->fname = check_pool_memory_size(jcr->fname, fd->msglen);
526 Dmsg1(200, "Atts+Digest=%s\n", fd->msg);
527 if ((len = sscanf(fd->msg, "%ld %d %100s", &file_index, &stream,
529 Jmsg3(jcr, M_FATAL, 0, _("bird<filed: bad attributes, expected 3 fields got %d\n"
530 " mslen=%d msg=%s\n"), len, fd->msglen, fd->msg);
534 * We read the Options or Signature into fname
535 * to prevent overrun, now copy it to proper location.
537 bstrncpy(Opts_Digest, fname, sizeof(Opts_Digest));
539 skip_nonspaces(&p); /* skip FileIndex */
541 skip_nonspaces(&p); /* skip Stream */
543 skip_nonspaces(&p); /* skip Opts_Digest */
544 p++; /* skip space */
547 *fn++ = *p++; /* copy filename */
549 *fn = *p++; /* term filename and point to attribs */
552 * Got attributes stream, decode it
554 if (stream == STREAM_UNIX_ATTRIBUTES || stream == STREAM_UNIX_ATTRIBUTES_EX) {
555 int32_t LinkFIf, LinkFIc;
556 Dmsg2(400, "file_index=%d attr=%s\n", file_index, attr);
558 jcr->FileIndex = file_index; /* remember attribute file_index */
559 decode_stat(attr, &statf, &LinkFIf); /* decode file stat packet */
560 do_Digest = CRYPTO_DIGEST_NONE;
561 jcr->fn_printed = false;
562 pm_strcpy(jcr->fname, fname); /* move filename into JCR */
564 Dmsg2(040, "dird<filed: stream=%d %s\n", stream, jcr->fname);
565 Dmsg1(020, "dird<filed: attr=%s\n", attr);
568 * Find equivalent record in the database
571 if (!db_get_file_attributes_record(jcr, jcr->db, jcr->fname,
572 &jcr->previous_jr, &fdbr)) {
573 Jmsg(jcr, M_INFO, 0, _("New file: %s\n"), jcr->fname);
574 Dmsg1(020, _("File not in catalog: %s\n"), jcr->fname);
575 stat = JS_Differences;
579 * mark file record as visited by stuffing the
580 * current JobId, which is unique, into the MarkId field.
582 db_mark_file_record(jcr, jcr->db, fdbr.FileId, jcr->JobId);
585 Dmsg3(400, "Found %s in catalog. inx=%d Opts=%s\n", jcr->fname,
586 file_index, Opts_Digest);
587 decode_stat(fdbr.LStat, &statc, &LinkFIc); /* decode catalog stat */
589 * Loop over options supplied by user and verify the
590 * fields he requests.
592 for (p=Opts_Digest; *p; p++) {
593 char ed1[30], ed2[30];
595 case 'i': /* compare INODEs */
596 if (statc.st_ino != statf.st_ino) {
598 Jmsg(jcr, M_INFO, 0, _(" st_ino differ. Cat: %s File: %s\n"),
599 edit_uint64((uint64_t)statc.st_ino, ed1),
600 edit_uint64((uint64_t)statf.st_ino, ed2));
601 stat = JS_Differences;
604 case 'p': /* permissions bits */
605 if (statc.st_mode != statf.st_mode) {
607 Jmsg(jcr, M_INFO, 0, _(" st_mode differ. Cat: %x File: %x\n"),
608 (uint32_t)statc.st_mode, (uint32_t)statf.st_mode);
609 stat = JS_Differences;
612 case 'n': /* number of links */
613 if (statc.st_nlink != statf.st_nlink) {
615 Jmsg(jcr, M_INFO, 0, _(" st_nlink differ. Cat: %d File: %d\n"),
616 (uint32_t)statc.st_nlink, (uint32_t)statf.st_nlink);
617 stat = JS_Differences;
620 case 'u': /* user id */
621 if (statc.st_uid != statf.st_uid) {
623 Jmsg(jcr, M_INFO, 0, _(" st_uid differ. Cat: %u File: %u\n"),
624 (uint32_t)statc.st_uid, (uint32_t)statf.st_uid);
625 stat = JS_Differences;
628 case 'g': /* group id */
629 if (statc.st_gid != statf.st_gid) {
631 Jmsg(jcr, M_INFO, 0, _(" st_gid differ. Cat: %u File: %u\n"),
632 (uint32_t)statc.st_gid, (uint32_t)statf.st_gid);
633 stat = JS_Differences;
637 if (statc.st_size != statf.st_size) {
639 Jmsg(jcr, M_INFO, 0, _(" st_size differ. Cat: %s File: %s\n"),
640 edit_uint64((uint64_t)statc.st_size, ed1),
641 edit_uint64((uint64_t)statf.st_size, ed2));
642 stat = JS_Differences;
645 case 'a': /* access time */
646 if (statc.st_atime != statf.st_atime) {
648 Jmsg(jcr, M_INFO, 0, _(" st_atime differs\n"));
649 stat = JS_Differences;
653 if (statc.st_mtime != statf.st_mtime) {
655 Jmsg(jcr, M_INFO, 0, _(" st_mtime differs\n"));
656 stat = JS_Differences;
659 case 'c': /* ctime */
660 if (statc.st_ctime != statf.st_ctime) {
662 Jmsg(jcr, M_INFO, 0, _(" st_ctime differs\n"));
663 stat = JS_Differences;
666 case 'd': /* file size decrease */
667 if (statc.st_size > statf.st_size) {
669 Jmsg(jcr, M_INFO, 0, _(" st_size decrease. Cat: %s File: %s\n"),
670 edit_uint64((uint64_t)statc.st_size, ed1),
671 edit_uint64((uint64_t)statf.st_size, ed2));
672 stat = JS_Differences;
675 case '5': /* compare MD5 */
676 Dmsg1(500, "set Do_MD5 for %s\n", jcr->fname);
677 do_Digest = CRYPTO_DIGEST_MD5;
679 case '1': /* compare SHA1 */
680 do_Digest = CRYPTO_DIGEST_SHA1;
689 * Got Digest Signature from Storage daemon
690 * It came across in the Opts_Digest field.
692 } else if (crypto_digest_stream_type(stream) != CRYPTO_DIGEST_NONE) {
693 Dmsg2(400, "stream=Digest inx=%d Digest=%s\n", file_index, Opts_Digest);
695 * When ever we get a digest is MUST have been
696 * preceded by an attributes record, which sets attr_file_index
698 if (jcr->FileIndex != (uint32_t)file_index) {
699 Jmsg2(jcr, M_FATAL, 0, _("MD5/SHA1 index %d not same as attributes %d\n"),
700 file_index, jcr->FileIndex);
703 if (do_Digest != CRYPTO_DIGEST_NONE) {
704 db_escape_string(buf, Opts_Digest, strlen(Opts_Digest));
705 if (strcmp(buf, fdbr.Digest) != 0) {
707 if (debug_level >= 10) {
708 Jmsg(jcr, M_INFO, 0, _(" %s not same. File=%s Cat=%s\n"),
709 stream_to_ascii(stream), buf, fdbr.Digest);
711 Jmsg(jcr, M_INFO, 0, _(" %s differs.\n"),
712 stream_to_ascii(stream));
714 stat = JS_Differences;
716 do_Digest = CRYPTO_DIGEST_NONE;
719 jcr->JobFiles = file_index;
721 if (is_bnet_error(fd)) {
723 Jmsg2(jcr, M_FATAL, 0, _("bdird<filed: bad attributes from filed n=%d : %s\n"),
728 /* Now find all the files that are missing -- i.e. all files in
729 * the database where the MarkId != current JobId
731 jcr->fn_printed = false;
732 bsnprintf(buf, sizeof(buf),
733 "SELECT Path.Path,Filename.Name FROM File,Path,Filename "
734 "WHERE File.JobId=%d "
735 "AND File.MarkId!=%d AND File.PathId=Path.PathId "
736 "AND File.FilenameId=Filename.FilenameId",
738 /* missing_handler is called for each file found */
739 db_sql_query(jcr->db, buf, missing_handler, (void *)jcr);
740 if (jcr->fn_printed) {
741 stat = JS_Differences;
743 free_pool_memory(fname);
744 set_jcr_job_status(jcr, stat);
745 return stat == JS_Terminated;
749 * We are called here for each record that matches the above
750 * SQL query -- that is for each file contained in the Catalog
751 * that was not marked earlier. This means that the file in
752 * question is a missing file (in the Catalog but not on Disk).
754 static int missing_handler(void *ctx, int num_fields, char **row)
756 JCR *jcr = (JCR *)ctx;
758 if (job_canceled(jcr)) {
761 if (!jcr->fn_printed) {
762 Jmsg(jcr, M_INFO, 0, "\n");
763 Jmsg(jcr, M_INFO, 0, _("The following files are in the Catalog but not on disk:\n"));
764 jcr->fn_printed = true;
766 Jmsg(jcr, M_INFO, 0, " %s%s\n", row[0]?row[0]:"", row[1]?row[1]:"");
772 * Print filename for verify
774 static void prt_fname(JCR *jcr)
776 if (!jcr->fn_printed) {
777 Jmsg(jcr, M_INFO, 0, _("File: %s\n"), jcr->fname);
778 jcr->fn_printed = TRUE;