>>>>> On Thu, 01 Jul 2010 17:15:09 +0200, Andreas Koch said:
>
> On 07/01/2010 02:37 PM, Martin Simmons wrote:
> > Does the FileSet have just one Include section and one Options section?
> If
> > it is more complicated than that, please post it.
>
> The complete FileSet is:
>
> # List of files to be backed up
> FileSet {
> Name = "FullSet"
> Include {
> Options {
> signature = SHA1;
> verify=pins1;
> accurate=pins1;
> }
> File = /boot
> File = /
> File = /home/stud
> File = /home/wimi
> File = /opt
> File = /fs/images
> File = /fs/vmware
> File = /fs/backups
> }
>
> Exclude {
> File = /var/lib/bacula
> File = /tmp
> File = /proc
> File = /tmp
> File = /.journal
> File = /.fsck
> }
> }
>
>
> > Can you repeat it with a FileSet containing only a few files (maybe try
> > /usr/share/postgresql-8.4/man/man7)?
>
> Done. The error occurs only during the differential run, on the same file as
> before
>
> /usr/share/postgresql-8.4/man/man7/table.7.bz2
Good, at least it is consistent.
> That file is backed-up, though: (together with a dummy one I created to force
> the differential):
>
> *list files jobid=17
> +--------------------------------------------------+
> | filename |
> +--------------------------------------------------+
> | /usr/share/postgresql-8.4/man/man7/ |
> | /usr/share/postgresql-8.4/man/man7/table.7.bz2 |
> | /usr/share/postgresql-8.4/man/man7/I_AM_FAKE_AHK |
> +--------------------------------------------------+
>
> I assume that Bacula is being conservative here: Unless it can prove (by
> identical checksums) that a file has not changed, it will back-up the file?
Yes.
> > You could try runing the bacula-fd with debug level 100 and look for
> > messages
> > like
> >
> > add fname=<...> lstat=... chksum=...
> >
> > in the debug output of the Differential job (it could be large).
> >
> > What is the add fname message for a file that gives the warning?
>
> gundabad-fd: accurate.c:224-19 add
> fname=</usr/share/postgresql-8.4/man/man7/table.7.bz2> lstat=PwA Dky9 IGk C A
> A A 6 BAA I BMBmj+ BMBmj+ BMBmkF BU A C chksum=
>
> Note that `chksum=´ is empty.
>
> > What is the add fname message for a file that doesn't give the warning?
>
> gundabad-fd: accurate.c:224-19 add
> fname=</usr/share/postgresql-8.4/man/man7/alter_tablespace.7.bz2> lstat=PwA
> Dkxm IGk B A A A Kz BAA I BMBmj8 BMBmj8 BMBmkF A A C
> chksum=vJvtTQ9F2sguOypX9ThR7u5IJi8
>
> Here, chksum has a value.
OK, that is interesting. This missing chksum is causing the warning, so the
we need to find why is it missing.
> Also I did some more experimentation: The problem only occurs when I set
> Accurate = yes (as I have normally done). If I change that to no, no warnings
> are shown.
Yes, the warning is generated by the code for accurate backups, so setting
Accurate = no will prevent it.
> Please let me know if you need more info!
It would be interesting to check the MD5 column in the output of the SQL query
SELECT Path.Path,Filename.Name,File.MD5 FROM File,Filename,Path WHERE
File.JobId=NNNN
AND Filename.FilenameId=File.FilenameId
AND Path.PathId=File.PathId ORDER BY
Path.Path,Filename.Name;
when NNNN is the jobid of the Full backup (the column is called MD5, but it
can also contain the SHA1 value).
> Assuming it is a bug, where can I file a bug report?
See http://www.bacula.org/en/?page=bugs
__Martin
------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users
|
