Your message dated Mon, 01 Dec 2014 20:59:38 +0100
with message-id <[email protected]>
and subject line Re: Bacula: Verify Differences shown on mountpoints using
"Verify Volume to Catalog"
has caused the Debian Bug report #692355,
regarding Bacula: Verify Differences shown on mountpoints using "Verify Volume
to Catalog"
to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [email protected]
immediately.)
--
692355: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=692355
Debian Bug Tracking System
Contact [email protected] with problems
--- Begin Message ---
Package: bacula
Version: 5.0.2-2.2+squeeze1
Severity: normal
Tags: patch
Debian squeeze
#cat /etc/debian_version
6.0.6
#dpkg -l bacula\* | grep ^ii | cut -c5-65
bacula-common 5.0.2-2.2+squeeze1
bacula-common-pgsql 5.0.2-2.2+squeeze1
bacula-console 5.0.2-2.2+squeeze1
bacula-director-common 5.0.2-2.2+squeeze1
bacula-director-pgsql 5.0.2-2.2+squeeze1
bacula-doc 5.0.2-1
bacula-fd 5.0.2-2.2+squeeze1
bacula-sd 5.0.2-2.2+squeeze1
bacula-sd-pgsql 5.0.2-2.2+squeeze1
03-Nov 00:04 vsrvbck01-dir JobId 35711: Verifying against JobId=35699
Job=srvxen02-Tape.2012-11-02_23.05.01_53
03-Nov 00:04 vsrvbck01-dir JobId 35711: Bootstrap records written to
/var/lib/bacula/vsrvbck01-dir.restore.34.bsr
03-Nov 00:04 vsrvbck01-dir JobId 35711: Start Verify JobId=35711
Level=VolumeToCatalog Job=srvxen02-Tape-Verify.2012-11-02_23.10.00_05
03-Nov 00:04 vsrvbck01-dir JobId 35711: Using Device "Drive0_SB5"
03-Nov 00:06 vsrvbck01-sd JobId 35711: Ready to read from volume "HP1054L3" on
device "Drive0_SB5" (/dev/nst3).
03-Nov 00:06 vsrvbck01-sd JobId 35711: Forward spacing Volume "HP1054L3" to
file:block 147:0.
03-Nov 00:28 vsrvbck01-dir JobId 35711: Warning: The following files are in the
Catalog but not on the Volume(s):
03-Nov 00:28 vsrvbck01-dir JobId 35711: /boot/
03-Nov 00:28 vsrvbck01-dir JobId 35711: Bacula vsrvbck01-dir 5.0.2 (28Apr10):
03-Nov-2012 00:28:50
Build OS: x86_64-pc-linux-gnu debian 6.0.6
JobId: 35711
Job: srvxen02-Tape-Verify.2012-11-02_23.10.00_05
FileSet: None
Verify Level: VolumeToCatalog
Client: srvxen02-fd
Verify JobId: 35699
Verify Job:
Start time: 03-Nov-2012 00:04:44
End time: 03-Nov-2012 00:28:50
Files Expected: 371,031
Files Examined: 371,031
Non-fatal FD errors: 0
FD termination status: OK
SD termination status: OK
Termination: Verify Differences
03-Nov 00:28 vsrvbck01-dir JobId 35711: Begin pruning Jobs older than 2 months .
03-Nov 00:28 vsrvbck01-dir JobId 35711: No Jobs found to prune.
03-Nov 00:28 vsrvbck01-dir JobId 35711: Begin pruning Jobs.
03-Nov 00:28 vsrvbck01-dir JobId 35711: No Files found to prune.
03-Nov 00:28 vsrvbck01-dir JobId 35711: End auto prune.
The problem is that the File table in the catalog contains two entries for
All of the inner mounted filesystems (/boot etc), one which would normally be
Used to recreate the mount point and another for the top level directory of the
filesystem itself. This confuses the VolumeToCatalog verification code.
It looks like it was broken by the fix for bacula_bug#1524.
http://bugs.bacula.org/view.php?id=1661
In L_VERIFY_VOLUME_TO_CATALOG, it needs to check every FileIndex to ensure
that the catalog is in the same order as the volume. That also correctly
handles duplicate names, which have different values for FileIndex.
http://bacula.git.sourceforge.net/git/gitweb.cgi?p=bacula/bacula;a=commit;h=626d681b782437c44b12a9e77627312260544062
>From 626d681b782437c44b12a9e77627312260544062 Mon Sep 17 00:00:00 2001
From: Martin Simmons <[email protected]<mailto:[email protected]>>
Date: Wed, 5 Jan 2011 10:19:22 +0100
Subject: [PATCH] Fix #1661 about verify differences with
VERIFY_VOLUME_TO_CATALOG
In L_VERIFY_VOLUME_TO_CATALOG, it needs to check every FileIndex to ensure that
the catalog is in the same order as the volume. That also correctly handles
duplicate names, which have different values for FileIndex.
This bug was introduced when fixing #1524
---
bacula/src/cats/sql_get.c | 10 +++++++++-
1 files changed, 9 insertions(+), 1 deletions(-)
diff --git a/bacula/src/cats/sql_get.c b/bacula/src/cats/sql_get.c index
e4ebfba..d3d21a2 100644
--- a/bacula/src/cats/sql_get.c
+++ b/bacula/src/cats/sql_get.c
@@ -105,7 +105,7 @@ int db_get_file_attributes_record(JCR *jcr, B_DB *mdb, char
*fname, JOB_DBR *jr,
* use includes the directory twice. In this case, Verify
* VolumeToCatalog fails because we have two copies in the catalog,
* and only the first one is marked (twice). So, when calling from Verify,
- * jr is not NULL and we know jr->FileIndex is the fileindex
+ * VolumeToCatalog jr is not NULL and we know jr->FileIndex is the fileindex
* of the version of the directory/file we actually want and do
* a more explicit SQL search.
*/
@@ -125,6 +125,14 @@ int db_get_file_record(JCR *jcr, B_DB *mdb, JOB_DBR *jr,
FILE_DBR *fdbr)
edit_int64(fdbr->PathId, ed1),
edit_int64(fdbr->FilenameId, ed2),
edit_int64(jr->ClientId,ed3));
+ } else if (jcr->getJobLevel() == L_VERIFY_VOLUME_TO_CATALOG) {
+ Mmsg(mdb->cmd,
+ "SELECT FileId, LStat, MD5 FROM File WHERE File.JobId=%s AND
File.PathId=%s AND "
+ "File.FilenameId=%s AND File.FileIndex=%u",
+ edit_int64(fdbr->JobId, ed1),
+ edit_int64(fdbr->PathId, ed2),
+ edit_int64(fdbr->FilenameId,ed3),
+ jr->FileIndex);
} else {
Mmsg(mdb->cmd,
"SELECT FileId, LStat, MD5 FROM File WHERE File.JobId=%s AND File.PathId=%s
AND "
--
1.7.4.1
--- End Message ---
--- Begin Message ---
The fix for this bug went in v5.2 of bacula, so I'm closing this bug.
--- End Message ---
