3 * Bacula Director -- verify.c -- responsible for running file verification
5 * Kern Sibbald, October MM
7 * Basic tasks done here:
9 * Open connection with File daemon and pass him commands
11 * When the File daemon sends the attributes, compare them to
18 Copyright (C) 2000-2004 Kern Sibbald and John Walker
20 This program is free software; you can redistribute it and/or
21 modify it under the terms of the GNU General Public License as
22 published by the Free Software Foundation; either version 2 of
23 the License, or (at your option) any later version.
25 This program is distributed in the hope that it will be useful,
26 but WITHOUT ANY WARRANTY; without even the implied warranty of
27 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
28 General Public License for more details.
30 You should have received a copy of the GNU General Public
31 License along with this program; if not, write to the Free
32 Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
39 #include "findlib/find.h"
41 /* Imported Global Variables */
42 extern int debug_level;
44 /* Commands sent to File daemon */
45 static char verifycmd[] = "verify level=%s\n";
46 static char storaddr[] = "storage address=%s port=%d ssl=0\n";
47 static char sessioncmd[] = "session %s %ld %ld %ld %ld %ld %ld\n";
49 /* Responses received from File daemon */
50 static char OKverify[] = "2000 OK verify\n";
51 static char OKstore[] = "2000 OK storage\n";
52 static char OKsession[] = "2000 OK session\n";
54 /* Forward referenced functions */
55 static void verify_cleanup(JCR *jcr, int TermCode);
56 static void prt_fname(JCR *jcr);
57 static int missing_handler(void *ctx, int num_fields, char **row);
60 * Do a verification of the specified files against the Catlaog
62 * Returns: 0 on failure
65 int do_verify(JCR *jcr)
67 const char *level, *Name;
69 JOB_DBR jr, verify_jr;
70 JobId_t verify_jobid = 0;
73 memset(&verify_jr, 0, sizeof(verify_jr));
74 if (!jcr->verify_jr) {
75 jcr->verify_jr = &verify_jr;
77 if (!get_or_create_client_record(jcr)) {
81 Dmsg1(9, "bdird: created client %s record\n", jcr->client->hdr.name);
84 * Find JobId of last job that ran. E.g.
85 * for VERIFY_CATALOG we want the JobId of the last INIT.
86 * for VERIFY_VOLUME_TO_CATALOG, we want the JobId of the
89 if (jcr->JobLevel == L_VERIFY_CATALOG ||
90 jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
91 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG) {
92 memcpy(&jr, &jcr->jr, sizeof(jr));
93 if (jcr->job->verify_job &&
94 (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
95 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG)) {
96 Name = jcr->job->verify_job->hdr.name;
100 Dmsg1(100, "find last jobid for: %s\n", NPRT(Name));
101 if (!db_find_last_jobid(jcr, jcr->db, Name, &jr)) {
102 if (jcr->JobLevel == L_VERIFY_CATALOG) {
103 Jmsg(jcr, M_FATAL, 0, _(
104 "Unable to find JobId of previous InitCatalog Job.\n"
105 "Please run a Verify with Level=InitCatalog before\n"
106 "running the current Job.\n"));
108 Jmsg(jcr, M_FATAL, 0, _(
109 "Unable to find JobId of previous Job for this client.\n"));
113 verify_jobid = jr.JobId;
114 Dmsg1(100, "Last full jobid=%d\n", verify_jobid);
117 jcr->jr.JobId = jcr->JobId;
118 jcr->jr.StartTime = jcr->start_time;
119 jcr->jr.Level = jcr->JobLevel;
120 if (!db_update_job_start_record(jcr, jcr->db, &jcr->jr)) {
121 Jmsg(jcr, M_FATAL, 0, "%s", db_strerror(jcr->db));
126 jcr->fname = get_pool_memory(PM_FNAME);
129 /* Print Job Start message */
130 Jmsg(jcr, M_INFO, 0, _("Start Verify JobId=%d Level=%s Job=%s\n"),
131 jcr->JobId, level_to_str(jcr->JobLevel), jcr->Job);
134 * Now get the job record for the previous backup that interests
135 * us. We use the verify_jobid that we found above.
137 if (jcr->JobLevel == L_VERIFY_CATALOG ||
138 jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
139 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG) {
140 verify_jr.JobId = verify_jobid;
141 if (!db_get_job_record(jcr, jcr->db, &verify_jr)) {
142 Jmsg(jcr, M_FATAL, 0, _("Could not get job record for previous Job. ERR=%s"),
143 db_strerror(jcr->db));
146 if (verify_jr.JobStatus != 'T') {
147 Jmsg(jcr, M_FATAL, 0, _("Last Job %d did not terminate normally. JobStatus=%c\n"),
148 verify_jobid, verify_jr.JobStatus);
151 Jmsg(jcr, M_INFO, 0, _("Verifying against JobId=%d Job=%s\n"),
152 verify_jr.JobId, verify_jr.Job);
156 * If we are verifying a Volume, we need the Storage
157 * daemon, so open a connection, otherwise, just
158 * create a dummy authorization key (passed to
159 * File daemon but not used).
161 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
162 RBSR *bsr = new_bsr();
164 bsr->JobId = verify_jr.JobId;
165 ua = new_ua_context(jcr);
166 complete_bsr(ua, bsr);
167 bsr->fi = new_findex();
169 bsr->fi->findex2 = verify_jr.JobFiles;
170 if (!write_bsr_file(ua, bsr)) {
177 if (jcr->RestoreBootstrap) {
178 free(jcr->RestoreBootstrap);
180 POOLMEM *fname = get_pool_memory(PM_MESSAGE);
181 Mmsg(&fname, "%s/restore.bsr", working_directory);
182 jcr->RestoreBootstrap = bstrdup(fname);
183 free_pool_memory(fname);
186 * Start conversation with Storage daemon
188 set_jcr_job_status(jcr, JS_Blocked);
189 if (!connect_to_storage_daemon(jcr, 10, SDConnectTimeout, 1)) {
193 * Now start a job with the Storage daemon
195 if (!start_storage_daemon_job(jcr)) {
199 * Now start a Storage daemon message thread
201 if (!start_storage_daemon_message_thread(jcr)) {
204 Dmsg0(50, "Storage daemon connection OK\n");
206 jcr->sd_auth_key = bstrdup("dummy"); /* dummy Storage daemon key */
209 if (jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG && jcr->job->verify_job) {
210 jcr->fileset = jcr->job->verify_job->fileset;
212 Dmsg2(100, "ClientId=%u JobLevel=%c\n", verify_jr.ClientId, jcr->JobLevel);
215 * OK, now connect to the File daemon
216 * and ask him for the files.
218 set_jcr_job_status(jcr, JS_Blocked);
219 if (!connect_to_file_daemon(jcr, 10, FDConnectTimeout, 1)) {
223 set_jcr_job_status(jcr, JS_Running);
224 fd = jcr->file_bsock;
227 Dmsg0(30, ">filed: Send include list\n");
228 if (!send_include_list(jcr)) {
232 Dmsg0(30, ">filed: Send exclude list\n");
233 if (!send_exclude_list(jcr)) {
238 * Send Level command to File daemon, as well
239 * as the Storage address if appropriate.
241 switch (jcr->JobLevel) {
245 case L_VERIFY_CATALOG:
248 case L_VERIFY_VOLUME_TO_CATALOG:
250 * send Storage daemon address to the File daemon
252 if (jcr->store->SDDport == 0) {
253 jcr->store->SDDport = jcr->store->SDport;
255 bnet_fsend(fd, storaddr, jcr->store->address, jcr->store->SDDport);
256 if (!response(jcr, fd, OKstore, "Storage", DISPLAY_ERROR)) {
261 * Send the bootstrap file -- what Volumes/files to restore
263 if (!send_bootstrap_file(jcr)) {
268 * The following code is deprecated
270 if (!jcr->RestoreBootstrap) {
272 * Pass the VolSessionId, VolSessionTime, Start and
273 * end File and Blocks on the session command.
275 bnet_fsend(fd, sessioncmd,
277 jr.VolSessionId, jr.VolSessionTime,
278 jr.StartFile, jr.EndFile, jr.StartBlock,
280 if (!response(jcr, fd, OKsession, "Session", DISPLAY_ERROR)) {
289 case L_VERIFY_DISK_TO_CATALOG:
290 level="disk_to_catalog";
293 Jmsg2(jcr, M_FATAL, 0, _("Unimplemented save level %d(%c)\n"), jcr->JobLevel,
298 if (!send_run_before_and_after_commands(jcr)) {
303 * Send verify command/level to File daemon
305 bnet_fsend(fd, verifycmd, level);
306 if (!response(jcr, fd, OKverify, "Verify", DISPLAY_ERROR)) {
311 * Now get data back from File daemon and
312 * compare it to the catalog or store it in the
313 * catalog depending on the run type.
315 /* Compare to catalog */
316 switch (jcr->JobLevel) {
317 case L_VERIFY_CATALOG:
318 Dmsg0(10, "Verify level=catalog\n");
319 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
320 jcr->SDJobStatus = JS_Terminated;
321 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
324 case L_VERIFY_VOLUME_TO_CATALOG:
325 Dmsg0(10, "Verify level=volume\n");
326 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
329 case L_VERIFY_DISK_TO_CATALOG:
330 Dmsg0(10, "Verify level=disk_to_catalog\n");
331 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
332 jcr->SDJobStatus = JS_Terminated;
333 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
338 Dmsg0(10, "Verify level=init\n");
339 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
340 jcr->SDJobStatus = JS_Terminated;
341 get_attributes_and_put_in_catalog(jcr);
345 Jmsg1(jcr, M_FATAL, 0, _("Unimplemented verify level %d\n"), jcr->JobLevel);
349 stat = wait_for_job_termination(jcr);
350 verify_cleanup(jcr, stat);
354 verify_cleanup(jcr, JS_ErrorTerminated);
359 * Release resources allocated during backup.
362 static void verify_cleanup(JCR *jcr, int TermCode)
364 char sdt[50], edt[50];
366 char term_code[100], fd_term_msg[100], sd_term_msg[100];
367 const char *term_msg;
372 // Dmsg1(100, "Enter verify_cleanup() TermCod=%d\n", TermCode);
374 JobId = jcr->jr.JobId;
375 set_jcr_job_status(jcr, TermCode);
377 update_job_end_record(jcr);
379 msg_type = M_INFO; /* by default INFO message */
382 term_msg = _("Verify OK");
384 case JS_ErrorTerminated:
385 term_msg = _("*** Verify Error ***");
386 msg_type = M_ERROR; /* Generate error message */
389 term_msg = _("Verify Canceled");
392 term_msg = _("Verify Differences");
395 term_msg = term_code;
396 bsnprintf(term_code, sizeof(term_code),
397 _("Inappropriate term code: %d %c\n"), TermCode, TermCode);
400 bstrftime(sdt, sizeof(sdt), jcr->jr.StartTime);
401 bstrftime(edt, sizeof(edt), jcr->jr.EndTime);
402 if (jcr->job->verify_job) {
403 Name = jcr->job->verify_job->hdr.name;
408 jobstatus_to_ascii(jcr->FDJobStatus, fd_term_msg, sizeof(fd_term_msg));
409 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
410 jobstatus_to_ascii(jcr->SDJobStatus, sd_term_msg, sizeof(sd_term_msg));
411 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n\
421 Files Examined: %s\n\
422 Non-fatal FD errors: %d\n\
423 FD termination status: %s\n\
424 SD termination status: %s\n\
425 Termination: %s\n\n"),
429 jcr->fileset->hdr.name,
430 level_to_str(jcr->JobLevel),
431 jcr->client->hdr.name,
432 jcr->verify_jr->JobId,
436 edit_uint64_with_commas(jcr->JobFiles, ec1),
442 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n\
452 Files Examined: %s\n\
453 Non-fatal FD errors: %d\n\
454 FD termination status: %s\n\
455 Termination: %s\n\n"),
459 jcr->fileset->hdr.name,
460 level_to_str(jcr->JobLevel),
461 jcr->client->hdr.name,
462 jcr->verify_jr->JobId,
466 edit_uint64_with_commas(jcr->JobFiles, ec1),
471 Dmsg0(100, "Leave verify_cleanup()\n");
473 free_memory(jcr->fname);
479 * This routine is called only during a Verify
481 int get_attributes_and_compare_to_catalog(JCR *jcr, JobId_t JobId)
486 struct stat statf; /* file stat */
487 struct stat statc; /* catalog stat */
488 int stat = JS_Terminated;
490 POOLMEM *fname = get_pool_memory(PM_MESSAGE);
494 memset(&fdbr, 0, sizeof(FILE_DBR));
495 fd = jcr->file_bsock;
499 Dmsg0(20, "bdird: waiting to receive file attributes\n");
501 * Get Attributes and Signature from File daemon
505 * Options or SIG (MD5/SHA1)
510 while ((n=bget_dirmsg(fd)) >= 0 && !job_canceled(jcr)) {
513 char Opts_SIG[MAXSTRING]; /* Verify Opts or MD5/SHA1 signature */
515 fname = check_pool_memory_size(fname, fd->msglen);
516 jcr->fname = check_pool_memory_size(jcr->fname, fd->msglen);
517 Dmsg1(200, "Atts+SIG=%s\n", fd->msg);
518 if ((len = sscanf(fd->msg, "%ld %d %100s", &file_index, &stream,
520 Jmsg3(jcr, M_FATAL, 0, _("bird<filed: bad attributes, expected 3 fields got %d\n\
521 mslen=%d msg=%s\n"), len, fd->msglen, fd->msg);
525 * We read the Options or Signature into fname
526 * to prevent overrun, now copy it to proper location.
528 bstrncpy(Opts_SIG, fname, sizeof(Opts_SIG));
530 skip_nonspaces(&p); /* skip FileIndex */
532 skip_nonspaces(&p); /* skip Stream */
534 skip_nonspaces(&p); /* skip Opts_SIG */
535 p++; /* skip space */
538 *fn++ = *p++; /* copy filename */
540 *fn = *p++; /* term filename and point to attribs */
543 * Got attributes stream, decode it
545 if (stream == STREAM_UNIX_ATTRIBUTES || stream == STREAM_UNIX_ATTRIBUTES_EX) {
546 int32_t LinkFIf, LinkFIc;
547 Dmsg2(400, "file_index=%d attr=%s\n", file_index, attr);
549 jcr->FileIndex = file_index; /* remember attribute file_index */
550 decode_stat(attr, &statf, &LinkFIf); /* decode file stat packet */
552 jcr->fn_printed = FALSE;
553 pm_strcpy(&jcr->fname, fname); /* move filename into JCR */
555 Dmsg2(040, "dird<filed: stream=%d %s\n", stream, jcr->fname);
556 Dmsg1(020, "dird<filed: attr=%s\n", attr);
559 * Find equivalent record in the database
562 if (!db_get_file_attributes_record(jcr, jcr->db, jcr->fname,
563 jcr->verify_jr, &fdbr)) {
564 Jmsg(jcr, M_INFO, 0, _("New file: %s\n"), jcr->fname);
565 Dmsg1(020, _("File not in catalog: %s\n"), jcr->fname);
566 stat = JS_Differences;
570 * mark file record as visited by stuffing the
571 * current JobId, which is unique, into the MarkId field.
573 db_mark_file_record(jcr, jcr->db, fdbr.FileId, jcr->JobId);
576 Dmsg3(400, "Found %s in catalog. inx=%d Opts=%s\n", jcr->fname,
577 file_index, Opts_SIG);
578 decode_stat(fdbr.LStat, &statc, &LinkFIc); /* decode catalog stat */
580 * Loop over options supplied by user and verify the
581 * fields he requests.
583 for (p=Opts_SIG; *p; p++) {
584 char ed1[30], ed2[30];
586 case 'i': /* compare INODEs */
587 if (statc.st_ino != statf.st_ino) {
589 Jmsg(jcr, M_INFO, 0, _(" st_ino differ. Cat: %s File: %s\n"),
590 edit_uint64((uint64_t)statc.st_ino, ed1),
591 edit_uint64((uint64_t)statf.st_ino, ed2));
592 stat = JS_Differences;
595 case 'p': /* permissions bits */
596 if (statc.st_mode != statf.st_mode) {
598 Jmsg(jcr, M_INFO, 0, _(" st_mode differ. Cat: %x File: %x\n"),
599 (uint32_t)statc.st_mode, (uint32_t)statf.st_mode);
600 stat = JS_Differences;
603 case 'n': /* number of links */
604 if (statc.st_nlink != statf.st_nlink) {
606 Jmsg(jcr, M_INFO, 0, _(" st_nlink differ. Cat: %d File: %d\n"),
607 (uint32_t)statc.st_nlink, (uint32_t)statf.st_nlink);
608 stat = JS_Differences;
611 case 'u': /* user id */
612 if (statc.st_uid != statf.st_uid) {
614 Jmsg(jcr, M_INFO, 0, _(" st_uid differ. Cat: %u File: %u\n"),
615 (uint32_t)statc.st_uid, (uint32_t)statf.st_uid);
616 stat = JS_Differences;
619 case 'g': /* group id */
620 if (statc.st_gid != statf.st_gid) {
622 Jmsg(jcr, M_INFO, 0, _(" st_gid differ. Cat: %u File: %u\n"),
623 (uint32_t)statc.st_gid, (uint32_t)statf.st_gid);
624 stat = JS_Differences;
628 if (statc.st_size != statf.st_size) {
630 Jmsg(jcr, M_INFO, 0, _(" st_size differ. Cat: %s File: %s\n"),
631 edit_uint64((uint64_t)statc.st_size, ed1),
632 edit_uint64((uint64_t)statf.st_size, ed2));
633 stat = JS_Differences;
636 case 'a': /* access time */
637 if (statc.st_atime != statf.st_atime) {
639 Jmsg(jcr, M_INFO, 0, _(" st_atime differs\n"));
640 stat = JS_Differences;
644 if (statc.st_mtime != statf.st_mtime) {
646 Jmsg(jcr, M_INFO, 0, _(" st_mtime differs\n"));
647 stat = JS_Differences;
650 case 'c': /* ctime */
651 if (statc.st_ctime != statf.st_ctime) {
653 Jmsg(jcr, M_INFO, 0, _(" st_ctime differs\n"));
654 stat = JS_Differences;
657 case 'd': /* file size decrease */
658 if (statc.st_size > statf.st_size) {
660 Jmsg(jcr, M_INFO, 0, _(" st_size decrease. Cat: %s File: %s\n"),
661 edit_uint64((uint64_t)statc.st_size, ed1),
662 edit_uint64((uint64_t)statf.st_size, ed2));
663 stat = JS_Differences;
666 case '5': /* compare MD5 */
667 Dmsg1(500, "set Do_MD5 for %s\n", jcr->fname);
670 case '1': /* compare SHA1 */
680 * Got SIG Signature from Storage daemon
681 * It came across in the Opts_SIG field.
683 } else if (stream == STREAM_MD5_SIGNATURE || stream == STREAM_SHA1_SIGNATURE) {
684 Dmsg2(400, "stream=SIG inx=%d SIG=%s\n", file_index, Opts_SIG);
686 * When ever we get a signature is MUST have been
687 * preceded by an attributes record, which sets attr_file_index
689 if (jcr->FileIndex != (uint32_t)file_index) {
690 Jmsg2(jcr, M_FATAL, 0, _("MD5/SHA1 index %d not same as attributes %d\n"),
691 file_index, jcr->FileIndex);
695 db_escape_string(buf, Opts_SIG, strlen(Opts_SIG));
696 if (strcmp(buf, fdbr.SIG) != 0) {
698 if (debug_level >= 10) {
699 Jmsg(jcr, M_INFO, 0, _(" %s not same. File=%s Cat=%s\n"),
700 stream==STREAM_MD5_SIGNATURE?"MD5":"SHA1", buf, fdbr.SIG);
702 Jmsg(jcr, M_INFO, 0, _(" %s differs.\n"),
703 stream==STREAM_MD5_SIGNATURE?"MD5":"SHA1");
705 stat = JS_Differences;
710 jcr->JobFiles = file_index;
712 if (is_bnet_error(fd)) {
713 Jmsg2(jcr, M_FATAL, 0, _("bdird<filed: bad attributes from filed n=%d : %s\n"),
718 /* Now find all the files that are missing -- i.e. all files in
719 * the database where the MarkedId != current JobId
721 jcr->fn_printed = FALSE;
723 "SELECT Path.Path,Filename.Name FROM File,Path,Filename "
724 "WHERE File.JobId=%d "
725 "AND File.MarkedId!=%d AND File.PathId=Path.PathId "
726 "AND File.FilenameId=Filename.FilenameId",
728 /* missing_handler is called for each file found */
729 db_sql_query(jcr->db, buf, missing_handler, (void *)jcr);
730 if (jcr->fn_printed) {
731 stat = JS_Differences;
733 free_pool_memory(fname);
734 set_jcr_job_status(jcr, stat);
738 free_pool_memory(fname);
739 set_jcr_job_status(jcr, JS_ErrorTerminated);
744 * We are called here for each record that matches the above
745 * SQL query -- that is for each file contained in the Catalog
746 * that was not marked earlier. This means that the file in
747 * question is a missing file (in the Catalog but on on Disk).
749 static int missing_handler(void *ctx, int num_fields, char **row)
751 JCR *jcr = (JCR *)ctx;
753 if (!jcr->fn_printed) {
754 Jmsg(jcr, M_INFO, 0, "\n");
755 Jmsg(jcr, M_INFO, 0, _("The following files are missing:\n"));
756 jcr->fn_printed = TRUE;
758 Jmsg(jcr, M_INFO, 0, " %s%s\n", row[0]?row[0]:"", row[1]?row[1]:"");
764 * Print filename for verify
766 static void prt_fname(JCR *jcr)
768 if (!jcr->fn_printed) {
769 Jmsg(jcr, M_INFO, 0, _("File: %s\n"), jcr->fname);
770 jcr->fn_printed = TRUE;