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
17 Copyright (C) 2000-2005 Kern Sibbald
19 This program is free software; you can redistribute it and/or
20 modify it under the terms of the GNU General Public License
21 version 2 as ammended with additional clauses defined in the
22 file LICENSE in the main source directory.
24 This program is distributed in the hope that it will be useful,
25 but WITHOUT ANY WARRANTY; without even the implied warranty of
26 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
27 the file LICENSE for additional details.
34 #include "findlib/find.h"
36 /* Imported Global Variables */
37 extern int debug_level;
39 /* Commands sent to File daemon */
40 static char verifycmd[] = "verify level=%s\n";
41 static char storaddr[] = "storage address=%s port=%d ssl=0\n";
43 /* Responses received from File daemon */
44 static char OKverify[] = "2000 OK verify\n";
45 static char OKstore[] = "2000 OK storage\n";
47 /* Forward referenced functions */
48 static void prt_fname(JCR *jcr);
49 static int missing_handler(void *ctx, int num_fields, char **row);
53 * Called here before the job is run to do the job
56 bool do_verify_init(JCR *jcr)
59 JobId_t verify_jobid = 0;
62 memset(&jcr->target_jr, 0, sizeof(jcr->target_jr));
64 Dmsg1(9, "bdird: created client %s record\n", jcr->client->hdr.name);
67 * Find JobId of last job that ran. E.g.
68 * for VERIFY_CATALOG we want the JobId of the last INIT.
69 * for VERIFY_VOLUME_TO_CATALOG, we want the JobId of the
72 if (jcr->JobLevel == L_VERIFY_CATALOG ||
73 jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
74 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG) {
75 memcpy(&jr, &jcr->jr, sizeof(jr));
76 if (jcr->verify_job &&
77 (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
78 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG)) {
79 Name = jcr->verify_job->hdr.name;
83 Dmsg1(100, "find last jobid for: %s\n", NPRT(Name));
84 if (!db_find_last_jobid(jcr, jcr->db, Name, &jr)) {
85 if (jcr->JobLevel == L_VERIFY_CATALOG) {
86 Jmsg(jcr, M_FATAL, 0, _(
87 "Unable to find JobId of previous InitCatalog Job.\n"
88 "Please run a Verify with Level=InitCatalog before\n"
89 "running the current Job.\n"));
91 Jmsg(jcr, M_FATAL, 0, _(
92 "Unable to find JobId of previous Job for this client.\n"));
96 verify_jobid = jr.JobId;
97 Dmsg1(100, "Last full jobid=%d\n", verify_jobid);
100 * Now get the job record for the previous backup that interests
101 * us. We use the verify_jobid that we found above.
103 if (jcr->JobLevel == L_VERIFY_CATALOG ||
104 jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
105 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG) {
106 jcr->target_jr.JobId = verify_jobid;
107 if (!db_get_job_record(jcr, jcr->db, &jcr->target_jr)) {
108 Jmsg(jcr, M_FATAL, 0, _("Could not get job record for previous Job. ERR=%s"),
109 db_strerror(jcr->db));
112 if (jcr->target_jr.JobStatus != 'T') {
113 Jmsg(jcr, M_FATAL, 0, _("Last Job %d did not terminate normally. JobStatus=%c\n"),
114 verify_jobid, jcr->target_jr.JobStatus);
117 Jmsg(jcr, M_INFO, 0, _("Verifying against JobId=%d Job=%s\n"),
118 jcr->target_jr.JobId, jcr->target_jr.Job);
122 * If we are verifying a Volume, we need the Storage
123 * daemon, so open a connection, otherwise, just
124 * create a dummy authorization key (passed to
125 * File daemon but not used).
127 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
128 RBSR *bsr = new_bsr();
130 bsr->JobId = jcr->target_jr.JobId;
131 ua = new_ua_context(jcr);
132 complete_bsr(ua, bsr);
133 bsr->fi = new_findex();
135 bsr->fi->findex2 = jcr->target_jr.JobFiles;
136 jcr->ExpectedFiles = write_bsr_file(ua, bsr);
137 if (jcr->ExpectedFiles == 0) {
142 if (jcr->RestoreBootstrap) {
143 free(jcr->RestoreBootstrap);
145 POOLMEM *fname = get_pool_memory(PM_MESSAGE);
146 make_unique_restore_filename(ua, &fname);
147 jcr->RestoreBootstrap = bstrdup(fname);
150 free_pool_memory(fname);
151 jcr->needs_sd = true;
154 jcr->sd_auth_key = bstrdup("dummy"); /* dummy Storage daemon key */
157 if (jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG && jcr->verify_job) {
158 jcr->fileset = jcr->verify_job->fileset;
160 Dmsg2(100, "ClientId=%u JobLevel=%c\n", jcr->target_jr.ClientId, jcr->JobLevel);
166 * Do a verification of the specified files against the Catlaog
168 * Returns: false on failure
171 bool do_verify(JCR *jcr)
177 if (!db_update_job_start_record(jcr, jcr->db, &jcr->jr)) {
178 Jmsg(jcr, M_FATAL, 0, "%s", db_strerror(jcr->db));
182 /* Print Job Start message */
183 Jmsg(jcr, M_INFO, 0, _("Start Verify JobId=%d Level=%s Job=%s\n"),
184 jcr->JobId, level_to_str(jcr->JobLevel), jcr->Job);
186 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
188 * Start conversation with Storage daemon
190 set_jcr_job_status(jcr, JS_Blocked);
191 if (!connect_to_storage_daemon(jcr, 10, SDConnectTimeout, 1)) {
195 * Now start a job with the Storage daemon
197 if (!start_storage_daemon_job(jcr, jcr->storage, SD_READ)) {
201 * Now start a Storage daemon message thread
203 if (!start_storage_daemon_message_thread(jcr)) {
206 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->store->SDDport == 0) {
247 jcr->store->SDDport = jcr->store->SDport;
249 bnet_fsend(fd, storaddr, jcr->store->address, jcr->store->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)) {
261 if (!jcr->RestoreBootstrap) {
262 Jmsg0(jcr, M_FATAL, 0, _("Deprecated feature ... use bootstrap.\n"));
271 case L_VERIFY_DISK_TO_CATALOG:
272 level="disk_to_catalog";
275 Jmsg2(jcr, M_FATAL, 0, _("Unimplemented Verify level %d(%c)\n"), jcr->JobLevel,
280 if (!send_run_before_and_after_commands(jcr)) {
285 * Send verify command/level to File daemon
287 bnet_fsend(fd, verifycmd, level);
288 if (!response(jcr, fd, OKverify, "Verify", DISPLAY_ERROR)) {
293 * Now get data back from File daemon and
294 * compare it to the catalog or store it in the
295 * catalog depending on the run type.
297 /* Compare to catalog */
298 switch (jcr->JobLevel) {
299 case L_VERIFY_CATALOG:
300 Dmsg0(10, "Verify level=catalog\n");
301 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
302 jcr->SDJobStatus = JS_Terminated;
303 get_attributes_and_compare_to_catalog(jcr, jcr->target_jr.JobId);
306 case L_VERIFY_VOLUME_TO_CATALOG:
307 Dmsg0(10, "Verify level=volume\n");
308 get_attributes_and_compare_to_catalog(jcr, jcr->target_jr.JobId);
311 case L_VERIFY_DISK_TO_CATALOG:
312 Dmsg0(10, "Verify level=disk_to_catalog\n");
313 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
314 jcr->SDJobStatus = JS_Terminated;
315 get_attributes_and_compare_to_catalog(jcr, jcr->target_jr.JobId);
320 Dmsg0(10, "Verify level=init\n");
321 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
322 jcr->SDJobStatus = JS_Terminated;
323 get_attributes_and_put_in_catalog(jcr);
327 Jmsg1(jcr, M_FATAL, 0, _("Unimplemented verify level %d\n"), jcr->JobLevel);
331 stat = wait_for_job_termination(jcr);
332 if (stat == JS_Terminated) {
333 verify_cleanup(jcr, stat);
341 * Release resources allocated during backup.
344 void verify_cleanup(JCR *jcr, int TermCode)
346 char sdt[50], edt[50];
347 char ec1[30], ec2[30];
348 char term_code[100], fd_term_msg[100], sd_term_msg[100];
349 const char *term_msg;
354 // Dmsg1(100, "Enter verify_cleanup() TermCod=%d\n", TermCode);
355 dequeue_messages(jcr); /* display any queued messages */
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 JobId = jcr->jr.JobId;
365 set_jcr_job_status(jcr, TermCode);
367 update_job_end_record(jcr);
368 if (jcr->unlink_bsr && jcr->RestoreBootstrap) {
369 unlink(jcr->RestoreBootstrap);
370 jcr->unlink_bsr = false;
373 msg_type = M_INFO; /* by default INFO message */
376 term_msg = _("Verify OK");
379 case JS_ErrorTerminated:
380 term_msg = _("*** Verify Error ***");
381 msg_type = M_ERROR; /* Generate error message */
384 term_msg = _("Verify warnings");
387 term_msg = _("Verify Canceled");
390 term_msg = _("Verify Differences");
393 term_msg = term_code;
394 bsnprintf(term_code, sizeof(term_code),
395 _("Inappropriate term code: %d %c\n"), TermCode, TermCode);
398 bstrftimes(sdt, sizeof(sdt), jcr->jr.StartTime);
399 bstrftimes(edt, sizeof(edt), jcr->jr.EndTime);
400 if (jcr->verify_job) {
401 Name = jcr->verify_job->hdr.name;
406 jobstatus_to_ascii(jcr->FDJobStatus, fd_term_msg, sizeof(fd_term_msg));
407 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
408 jobstatus_to_ascii(jcr->SDJobStatus, sd_term_msg, sizeof(sd_term_msg));
409 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n"
413 " Verify Level: %s\n"
415 " Verify JobId: %d\n"
419 " Files Expected: %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->target_jr.JobId,
435 edit_uint64_with_commas(jcr->ExpectedFiles, ec1),
436 edit_uint64_with_commas(jcr->JobFiles, ec2),
442 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n"
446 " Verify Level: %s\n"
448 " Verify JobId: %d\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->target_jr.JobId,
466 edit_uint64_with_commas(jcr->JobFiles, ec1),
471 Dmsg0(100, "Leave verify_cleanup()\n");
475 * This routine is called only during a Verify
477 int get_attributes_and_compare_to_catalog(JCR *jcr, JobId_t JobId)
482 struct stat statf; /* file stat */
483 struct stat statc; /* catalog stat */
484 int stat = JS_Terminated;
486 POOLMEM *fname = get_pool_memory(PM_MESSAGE);
488 int32_t file_index = 0;
490 memset(&fdbr, 0, sizeof(FILE_DBR));
491 fd = jcr->file_bsock;
495 Dmsg0(20, "bdird: waiting to receive file attributes\n");
497 * Get Attributes and Signature from File daemon
501 * Options or SIG (MD5/SHA1)
506 while ((n=bget_dirmsg(fd)) >= 0 && !job_canceled(jcr)) {
509 char Opts_SIG[MAXSTRING]; /* Verify Opts or MD5/SHA1 signature */
511 fname = check_pool_memory_size(fname, fd->msglen);
512 jcr->fname = check_pool_memory_size(jcr->fname, fd->msglen);
513 Dmsg1(200, "Atts+SIG=%s\n", fd->msg);
514 if ((len = sscanf(fd->msg, "%ld %d %100s", &file_index, &stream,
516 Jmsg3(jcr, M_FATAL, 0, _("bird<filed: bad attributes, expected 3 fields got %d\n"
517 " mslen=%d msg=%s\n"), len, fd->msglen, fd->msg);
521 * We read the Options or Signature into fname
522 * to prevent overrun, now copy it to proper location.
524 bstrncpy(Opts_SIG, fname, sizeof(Opts_SIG));
526 skip_nonspaces(&p); /* skip FileIndex */
528 skip_nonspaces(&p); /* skip Stream */
530 skip_nonspaces(&p); /* skip Opts_SIG */
531 p++; /* skip space */
534 *fn++ = *p++; /* copy filename */
536 *fn = *p++; /* term filename and point to attribs */
539 * Got attributes stream, decode it
541 if (stream == STREAM_UNIX_ATTRIBUTES || stream == STREAM_UNIX_ATTRIBUTES_EX) {
542 int32_t LinkFIf, LinkFIc;
543 Dmsg2(400, "file_index=%d attr=%s\n", file_index, attr);
545 jcr->FileIndex = file_index; /* remember attribute file_index */
546 decode_stat(attr, &statf, &LinkFIf); /* decode file stat packet */
548 jcr->fn_printed = false;
549 pm_strcpy(jcr->fname, fname); /* move filename into JCR */
551 Dmsg2(040, "dird<filed: stream=%d %s\n", stream, jcr->fname);
552 Dmsg1(020, "dird<filed: attr=%s\n", attr);
555 * Find equivalent record in the database
558 if (!db_get_file_attributes_record(jcr, jcr->db, jcr->fname,
559 &jcr->target_jr, &fdbr)) {
560 Jmsg(jcr, M_INFO, 0, _("New file: %s\n"), jcr->fname);
561 Dmsg1(020, _("File not in catalog: %s\n"), jcr->fname);
562 stat = JS_Differences;
566 * mark file record as visited by stuffing the
567 * current JobId, which is unique, into the MarkId field.
569 db_mark_file_record(jcr, jcr->db, fdbr.FileId, jcr->JobId);
572 Dmsg3(400, "Found %s in catalog. inx=%d Opts=%s\n", jcr->fname,
573 file_index, Opts_SIG);
574 decode_stat(fdbr.LStat, &statc, &LinkFIc); /* decode catalog stat */
576 * Loop over options supplied by user and verify the
577 * fields he requests.
579 for (p=Opts_SIG; *p; p++) {
580 char ed1[30], ed2[30];
582 case 'i': /* compare INODEs */
583 if (statc.st_ino != statf.st_ino) {
585 Jmsg(jcr, M_INFO, 0, _(" st_ino differ. Cat: %s File: %s\n"),
586 edit_uint64((uint64_t)statc.st_ino, ed1),
587 edit_uint64((uint64_t)statf.st_ino, ed2));
588 stat = JS_Differences;
591 case 'p': /* permissions bits */
592 if (statc.st_mode != statf.st_mode) {
594 Jmsg(jcr, M_INFO, 0, _(" st_mode differ. Cat: %x File: %x\n"),
595 (uint32_t)statc.st_mode, (uint32_t)statf.st_mode);
596 stat = JS_Differences;
599 case 'n': /* number of links */
600 if (statc.st_nlink != statf.st_nlink) {
602 Jmsg(jcr, M_INFO, 0, _(" st_nlink differ. Cat: %d File: %d\n"),
603 (uint32_t)statc.st_nlink, (uint32_t)statf.st_nlink);
604 stat = JS_Differences;
607 case 'u': /* user id */
608 if (statc.st_uid != statf.st_uid) {
610 Jmsg(jcr, M_INFO, 0, _(" st_uid differ. Cat: %u File: %u\n"),
611 (uint32_t)statc.st_uid, (uint32_t)statf.st_uid);
612 stat = JS_Differences;
615 case 'g': /* group id */
616 if (statc.st_gid != statf.st_gid) {
618 Jmsg(jcr, M_INFO, 0, _(" st_gid differ. Cat: %u File: %u\n"),
619 (uint32_t)statc.st_gid, (uint32_t)statf.st_gid);
620 stat = JS_Differences;
624 if (statc.st_size != statf.st_size) {
626 Jmsg(jcr, M_INFO, 0, _(" st_size differ. Cat: %s File: %s\n"),
627 edit_uint64((uint64_t)statc.st_size, ed1),
628 edit_uint64((uint64_t)statf.st_size, ed2));
629 stat = JS_Differences;
632 case 'a': /* access time */
633 if (statc.st_atime != statf.st_atime) {
635 Jmsg(jcr, M_INFO, 0, _(" st_atime differs\n"));
636 stat = JS_Differences;
640 if (statc.st_mtime != statf.st_mtime) {
642 Jmsg(jcr, M_INFO, 0, _(" st_mtime differs\n"));
643 stat = JS_Differences;
646 case 'c': /* ctime */
647 if (statc.st_ctime != statf.st_ctime) {
649 Jmsg(jcr, M_INFO, 0, _(" st_ctime differs\n"));
650 stat = JS_Differences;
653 case 'd': /* file size decrease */
654 if (statc.st_size > statf.st_size) {
656 Jmsg(jcr, M_INFO, 0, _(" st_size decrease. Cat: %s File: %s\n"),
657 edit_uint64((uint64_t)statc.st_size, ed1),
658 edit_uint64((uint64_t)statf.st_size, ed2));
659 stat = JS_Differences;
662 case '5': /* compare MD5 */
663 Dmsg1(500, "set Do_MD5 for %s\n", jcr->fname);
666 case '1': /* compare SHA1 */
676 * Got SIG Signature from Storage daemon
677 * It came across in the Opts_SIG field.
679 } else if (stream == STREAM_MD5_SIGNATURE || stream == STREAM_SHA1_SIGNATURE) {
680 Dmsg2(400, "stream=SIG inx=%d SIG=%s\n", file_index, Opts_SIG);
682 * When ever we get a signature is MUST have been
683 * preceded by an attributes record, which sets attr_file_index
685 if (jcr->FileIndex != (uint32_t)file_index) {
686 Jmsg2(jcr, M_FATAL, 0, _("MD5/SHA1 index %d not same as attributes %d\n"),
687 file_index, jcr->FileIndex);
691 db_escape_string(buf, Opts_SIG, strlen(Opts_SIG));
692 if (strcmp(buf, fdbr.SIG) != 0) {
694 if (debug_level >= 10) {
695 Jmsg(jcr, M_INFO, 0, _(" %s not same. File=%s Cat=%s\n"),
696 stream==STREAM_MD5_SIGNATURE?"MD5":"SHA1", buf, fdbr.SIG);
698 Jmsg(jcr, M_INFO, 0, _(" %s differs.\n"),
699 stream==STREAM_MD5_SIGNATURE?"MD5":"SHA1");
701 stat = JS_Differences;
706 jcr->JobFiles = file_index;
708 if (is_bnet_error(fd)) {
710 Jmsg2(jcr, M_FATAL, 0, _("bdird<filed: bad attributes from filed n=%d : %s\n"),
715 /* Now find all the files that are missing -- i.e. all files in
716 * the database where the MarkedId != current JobId
718 jcr->fn_printed = false;
719 bsnprintf(buf, sizeof(buf),
720 "SELECT Path.Path,Filename.Name FROM File,Path,Filename "
721 "WHERE File.JobId=%d "
722 "AND File.MarkedId!=%d AND File.PathId=Path.PathId "
723 "AND File.FilenameId=Filename.FilenameId",
725 /* missing_handler is called for each file found */
726 db_sql_query(jcr->db, buf, missing_handler, (void *)jcr);
727 if (jcr->fn_printed) {
728 stat = JS_Differences;
730 free_pool_memory(fname);
731 set_jcr_job_status(jcr, stat);
732 return stat == JS_Terminated;
736 * We are called here for each record that matches the above
737 * SQL query -- that is for each file contained in the Catalog
738 * that was not marked earlier. This means that the file in
739 * question is a missing file (in the Catalog but not on Disk).
741 static int missing_handler(void *ctx, int num_fields, char **row)
743 JCR *jcr = (JCR *)ctx;
745 if (!jcr->fn_printed) {
746 Jmsg(jcr, M_INFO, 0, "\n");
747 Jmsg(jcr, M_INFO, 0, _("The following files are missing:\n"));
748 jcr->fn_printed = true;
750 Jmsg(jcr, M_INFO, 0, " %s%s\n", row[0]?row[0]:"", row[1]?row[1]:"");
756 * Print filename for verify
758 static void prt_fname(JCR *jcr)
760 if (!jcr->fn_printed) {
761 Jmsg(jcr, M_INFO, 0, _("File: %s\n"), jcr->fname);
762 jcr->fn_printed = TRUE;