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) 2000-2003 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)
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 Job=%s\n"),
131 jcr->JobId, 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 Jmsg1(jcr, M_FATAL, 0, _("Unimplemented save level %d\n"), jcr->JobLevel);
297 if (!send_run_before_and_after_commands(jcr)) {
302 * Send verify command/level to File daemon
304 bnet_fsend(fd, verifycmd, level);
305 if (!response(jcr, fd, OKverify, "Verify", DISPLAY_ERROR)) {
310 * Now get data back from File daemon and
311 * compare it to the catalog or store it in the
312 * catalog depending on the run type.
314 /* Compare to catalog */
315 switch (jcr->JobLevel) {
316 case L_VERIFY_CATALOG:
317 Dmsg0(10, "Verify level=catalog\n");
318 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
319 jcr->SDJobStatus = JS_Terminated;
320 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
323 case L_VERIFY_VOLUME_TO_CATALOG:
324 Dmsg0(10, "Verify level=volume\n");
325 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
328 case L_VERIFY_DISK_TO_CATALOG:
329 Dmsg0(10, "Verify level=disk_to_catalog\n");
330 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
331 jcr->SDJobStatus = JS_Terminated;
332 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
337 Dmsg0(10, "Verify level=init\n");
338 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
339 jcr->SDJobStatus = JS_Terminated;
340 get_attributes_and_put_in_catalog(jcr);
344 Jmsg1(jcr, M_FATAL, 0, _("Unimplemented verify level %d\n"), jcr->JobLevel);
348 stat = wait_for_job_termination(jcr);
349 verify_cleanup(jcr, stat);
353 verify_cleanup(jcr, JS_ErrorTerminated);
358 * Release resources allocated during backup.
361 static void verify_cleanup(JCR *jcr, int TermCode)
363 char sdt[50], edt[50];
365 char term_code[100], fd_term_msg[100], sd_term_msg[100];
371 // Dmsg1(100, "Enter verify_cleanup() TermCod=%d\n", TermCode);
373 JobId = jcr->jr.JobId;
374 set_jcr_job_status(jcr, TermCode);
376 update_job_end_record(jcr);
378 msg_type = M_INFO; /* by default INFO message */
381 term_msg = _("Verify OK");
383 case JS_ErrorTerminated:
384 term_msg = _("*** Verify Error ***");
385 msg_type = M_ERROR; /* Generate error message */
388 term_msg = _("Verify Canceled");
391 term_msg = _("Verify Differences");
394 term_msg = term_code;
395 bsnprintf(term_code, sizeof(term_code),
396 _("Inappropriate term code: %d %c\n"), TermCode, TermCode);
399 bstrftime(sdt, sizeof(sdt), jcr->jr.StartTime);
400 bstrftime(edt, sizeof(edt), jcr->jr.EndTime);
401 if (jcr->job->verify_job) {
402 Name = jcr->job->verify_job->hdr.name;
407 jobstatus_to_ascii(jcr->FDJobStatus, fd_term_msg, sizeof(fd_term_msg));
408 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
409 jobstatus_to_ascii(jcr->SDJobStatus, sd_term_msg, sizeof(sd_term_msg));
410 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n\
420 Files Examined: %s\n\
421 Non-fatal FD errors: %d\n\
422 FD termination status: %s\n\
423 SD termination status: %s\n\
424 Termination: %s\n\n"),
428 jcr->fileset->hdr.name,
429 level_to_str(jcr->JobLevel),
430 jcr->client->hdr.name,
431 jcr->verify_jr->JobId,
435 edit_uint64_with_commas(jcr->JobFiles, ec1),
441 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n\
451 Files Examined: %s\n\
452 Non-fatal FD errors: %d\n\
453 FD termination status: %s\n\
454 Termination: %s\n\n"),
458 jcr->fileset->hdr.name,
459 level_to_str(jcr->JobLevel),
460 jcr->client->hdr.name,
461 jcr->verify_jr->JobId,
465 edit_uint64_with_commas(jcr->JobFiles, ec1),
470 Dmsg0(100, "Leave verify_cleanup()\n");
472 free_memory(jcr->fname);
478 * This routine is called only during a Verify
480 int get_attributes_and_compare_to_catalog(JCR *jcr, JobId_t JobId)
485 struct stat statf; /* file stat */
486 struct stat statc; /* catalog stat */
487 int stat = JS_Terminated;
489 POOLMEM *fname = get_pool_memory(PM_MESSAGE);
493 memset(&fdbr, 0, sizeof(FILE_DBR));
494 fd = jcr->file_bsock;
498 Dmsg0(20, "bdird: waiting to receive file attributes\n");
500 * Get Attributes and Signature from File daemon
504 * Options or SIG (MD5/SHA1)
509 while ((n=bget_dirmsg(fd)) >= 0 && !job_canceled(jcr)) {
512 char Opts_SIG[MAXSTRING]; /* Verify Opts or MD5/SHA1 signature */
514 fname = check_pool_memory_size(fname, fd->msglen);
515 jcr->fname = check_pool_memory_size(jcr->fname, fd->msglen);
516 Dmsg1(200, "Atts+SIG=%s\n", fd->msg);
517 if ((len = sscanf(fd->msg, "%ld %d %100s", &file_index, &stream,
519 Jmsg3(jcr, M_FATAL, 0, _("bird<filed: bad attributes, expected 3 fields got %d\n\
520 mslen=%d msg=%s\n"), len, fd->msglen, fd->msg);
524 * We read the Options or Signature into fname
525 * to prevent overrun, now copy it to proper location.
527 bstrncpy(Opts_SIG, fname, sizeof(Opts_SIG));
529 skip_nonspaces(&p); /* skip FileIndex */
531 skip_nonspaces(&p); /* skip Stream */
533 skip_nonspaces(&p); /* skip Opts_SIG */
534 p++; /* skip space */
537 *fn++ = *p++; /* copy filename */
539 *fn = *p++; /* term filename and point to attribs */
542 * Got attributes stream, decode it
544 if (stream == STREAM_UNIX_ATTRIBUTES || stream == STREAM_UNIX_ATTRIBUTES_EX) {
545 int32_t LinkFIf, LinkFIc;
546 Dmsg2(400, "file_index=%d attr=%s\n", file_index, attr);
548 jcr->FileIndex = file_index; /* remember attribute file_index */
549 decode_stat(attr, &statf, &LinkFIf); /* decode file stat packet */
551 jcr->fn_printed = FALSE;
552 pm_strcpy(&jcr->fname, fname); /* move filename into JCR */
554 Dmsg2(040, "dird<filed: stream=%d %s\n", stream, jcr->fname);
555 Dmsg1(020, "dird<filed: attr=%s\n", attr);
558 * Find equivalent record in the database
561 if (!db_get_file_attributes_record(jcr, jcr->db, jcr->fname,
562 jcr->verify_jr, &fdbr)) {
563 Jmsg(jcr, M_INFO, 0, _("New file: %s\n"), jcr->fname);
564 Dmsg1(020, _("File not in catalog: %s\n"), jcr->fname);
565 stat = JS_Differences;
569 * mark file record as visited by stuffing the
570 * current JobId, which is unique, into the MarkId field.
572 db_mark_file_record(jcr, jcr->db, fdbr.FileId, jcr->JobId);
575 Dmsg3(400, "Found %s in catalog. inx=%d Opts=%s\n", jcr->fname,
576 file_index, Opts_SIG);
577 decode_stat(fdbr.LStat, &statc, &LinkFIc); /* decode catalog stat */
579 * Loop over options supplied by user and verify the
580 * fields he requests.
582 for (p=Opts_SIG; *p; p++) {
583 char ed1[30], ed2[30];
585 case 'i': /* compare INODEs */
586 if (statc.st_ino != statf.st_ino) {
588 Jmsg(jcr, M_INFO, 0, _(" st_ino differ. Cat: %s File: %s\n"),
589 edit_uint64((uint64_t)statc.st_ino, ed1),
590 edit_uint64((uint64_t)statf.st_ino, ed2));
591 stat = JS_Differences;
594 case 'p': /* permissions bits */
595 if (statc.st_mode != statf.st_mode) {
597 Jmsg(jcr, M_INFO, 0, _(" st_mode differ. Cat: %x File: %x\n"),
598 (uint32_t)statc.st_mode, (uint32_t)statf.st_mode);
599 stat = JS_Differences;
602 case 'n': /* number of links */
603 if (statc.st_nlink != statf.st_nlink) {
605 Jmsg(jcr, M_INFO, 0, _(" st_nlink differ. Cat: %d File: %d\n"),
606 (uint32_t)statc.st_nlink, (uint32_t)statf.st_nlink);
607 stat = JS_Differences;
610 case 'u': /* user id */
611 if (statc.st_uid != statf.st_uid) {
613 Jmsg(jcr, M_INFO, 0, _(" st_uid differ. Cat: %u File: %u\n"),
614 (uint32_t)statc.st_uid, (uint32_t)statf.st_uid);
615 stat = JS_Differences;
618 case 'g': /* group id */
619 if (statc.st_gid != statf.st_gid) {
621 Jmsg(jcr, M_INFO, 0, _(" st_gid differ. Cat: %u File: %u\n"),
622 (uint32_t)statc.st_gid, (uint32_t)statf.st_gid);
623 stat = JS_Differences;
627 if (statc.st_size != statf.st_size) {
629 Jmsg(jcr, M_INFO, 0, _(" st_size differ. Cat: %s File: %s\n"),
630 edit_uint64((uint64_t)statc.st_size, ed1),
631 edit_uint64((uint64_t)statf.st_size, ed2));
632 stat = JS_Differences;
635 case 'a': /* access time */
636 if (statc.st_atime != statf.st_atime) {
638 Jmsg(jcr, M_INFO, 0, _(" st_atime differs\n"));
639 stat = JS_Differences;
643 if (statc.st_mtime != statf.st_mtime) {
645 Jmsg(jcr, M_INFO, 0, _(" st_mtime differs\n"));
646 stat = JS_Differences;
649 case 'c': /* ctime */
650 if (statc.st_ctime != statf.st_ctime) {
652 Jmsg(jcr, M_INFO, 0, _(" st_ctime differs\n"));
653 stat = JS_Differences;
656 case 'd': /* file size decrease */
657 if (statc.st_size > statf.st_size) {
659 Jmsg(jcr, M_INFO, 0, _(" st_size decrease. Cat: %s File: %s\n"),
660 edit_uint64((uint64_t)statc.st_size, ed1),
661 edit_uint64((uint64_t)statf.st_size, ed2));
662 stat = JS_Differences;
665 case '5': /* compare MD5 */
666 Dmsg1(500, "set Do_MD5 for %s\n", jcr->fname);
669 case '1': /* compare SHA1 */
679 * Got SIG Signature from Storage daemon
680 * It came across in the Opts_SIG field.
682 } else if (stream == STREAM_MD5_SIGNATURE || stream == STREAM_SHA1_SIGNATURE) {
683 Dmsg2(400, "stream=SIG inx=%d SIG=%s\n", file_index, Opts_SIG);
685 * When ever we get a signature is MUST have been
686 * preceded by an attributes record, which sets attr_file_index
688 if (jcr->FileIndex != (uint32_t)file_index) {
689 Jmsg2(jcr, M_FATAL, 0, _("MD5/SHA1 index %d not same as attributes %d\n"),
690 file_index, jcr->FileIndex);
694 db_escape_string(buf, Opts_SIG, strlen(Opts_SIG));
695 if (strcmp(buf, fdbr.SIG) != 0) {
697 if (debug_level >= 10) {
698 Jmsg(jcr, M_INFO, 0, _(" %s not same. File=%s Cat=%s\n"),
699 stream==STREAM_MD5_SIGNATURE?"MD5":"SHA1", buf, fdbr.SIG);
701 Jmsg(jcr, M_INFO, 0, _(" %s differs.\n"),
702 stream==STREAM_MD5_SIGNATURE?"MD5":"SHA1");
704 stat = JS_Differences;
709 jcr->JobFiles = file_index;
711 if (is_bnet_error(fd)) {
712 Jmsg2(jcr, M_FATAL, 0, _("bdird<filed: bad attributes from filed n=%d : %s\n"),
717 /* Now find all the files that are missing -- i.e. all files in
718 * the database where the MarkedId != current JobId
720 jcr->fn_printed = FALSE;
722 "SELECT Path.Path,Filename.Name FROM File,Path,Filename "
723 "WHERE File.JobId=%d "
724 "AND File.MarkedId!=%d AND File.PathId=Path.PathId "
725 "AND File.FilenameId=Filename.FilenameId",
727 /* missing_handler is called for each file found */
728 db_sql_query(jcr->db, buf, missing_handler, (void *)jcr);
729 if (jcr->fn_printed) {
730 stat = JS_Differences;
732 free_pool_memory(fname);
733 set_jcr_job_status(jcr, stat);
737 free_pool_memory(fname);
738 set_jcr_job_status(jcr, JS_ErrorTerminated);
743 * We are called here for each record that matches the above
744 * SQL query -- that is for each file contained in the Catalog
745 * that was not marked earlier. This means that the file in
746 * question is a missing file (in the Catalog but on on Disk).
748 static int missing_handler(void *ctx, int num_fields, char **row)
750 JCR *jcr = (JCR *)ctx;
752 if (!jcr->fn_printed) {
753 Jmsg(jcr, M_INFO, 0, "\n");
754 Jmsg(jcr, M_INFO, 0, _("The following files are missing:\n"));
755 jcr->fn_printed = TRUE;
757 Jmsg(jcr, M_INFO, 0, " %s%s\n", row[0]?row[0]:"", row[1]?row[1]:"");
763 * Print filename for verify
765 static void prt_fname(JCR *jcr)
767 if (!jcr->fn_printed) {
768 Jmsg(jcr, M_INFO, 0, _("File: %s\n"), jcr->fname);
769 jcr->fn_printed = TRUE;