Hi Felix,
I am looking at BareOS to do our exchange backups as well. Have you changed
your script after posting this? Is it still running as expected after all these
months?
On Friday, June 13, 2014 6:01:47 AM UTC-7, Felix Zachlod wrote:
> Hello!
>
> we where very unhappy with having Bareos to backup our whole Exchange DB each
> and every day. Furthermore Exchange is not able to recognize the backup taken
> by Bareos as full backup and will never flush the log files. This situation
> is very unsatisfactory. So I read a bit about VSS snapshotting and was
> finally able to set up a nearly perfect solution with the use of some
> scripting on the client.
>
> Here is what you will need (We assume the Exchange Database is in one Folder
> named ExchangeDB and the transaction log in the ExchangeLOG folder on the
> same (virtual) drive:
>
> FileSet {
> Name = "ExchangeFullDB"
> Ignore FileSet Changes = yes
> Enable VSS = no
> Include {
> Options { signature = MD5
> }
> ExcludeDirContaining = nobackup.txt
> File = "B:/ExchangeDB"
> File = "B:/ExchangeLOG"
> }
> }
>
> Notice that B: ist NOT the actual drive where the exchange db resides but a
> temporary drive letter where we will expose the VSS snapshot. Furthermore
> notice that Enable VSS is set to no as our client script will take care of
> taking a vss snapshot. The ExcludeDirContaining directive is important this
> will make us able to exclude the DB from all incremental (and differential)
> backups.
>
> Here is the Jobdef.
>
> JobDefs {
> Name = "ExchangeServer"
> FileSet = ExchangeFullDB
> Type = Backup
> Level = Incremental
> Messages = Standard
> Pool = Default
> Full Backup Pool = HDDFull
> Incremental Backup Pool = HDDIncr
> Differential Backup Pool = HDDDiff
> Priority = 10
> Schedule = MonthlyCycle2
> Write Bootstrap = "/var/lib/bacula/%c_%n.bsr"
> Allow Duplicate Jobs = no
> Prefer Mounted Volumes = no
> Client Run Before Job =
> "C:/Programme/Bareos/Scripts/pre_backup_ex.cmd %l"
> Client Run After Job =
> "C:/Programme/Bareos/Scripts/post_backup_ex.cmd"
> Reschedule On Error = yes
> Reschedule Interval = 1 hour
> Reschedule Times = 20
> }
>
> And here is the pre_backup_ex.cmd which will reside on the client:
>
> @echo off
> if "%1" == "" (
> echo Backup job level has to be passed as first parameter.
> exit /b 1
> )
> if /i %1 == full (
> IF EXIST E:\ExchangeDB\nobackup.txt del /F E:\ExchangeDB\nobackup.txt
> diskshadow -s "c:\Program
> Files\Bareos\Scripts\pre_backup_ex_diskshadow.txt"
> ) else (
> copy /y nul E:\ExchangeDB\nobackup.txt
> diskshadow -s "c:\Program
> Files\Bareos\Scripts\pre_backup_ex_diskshadow_nofull.txt"
> )
>
> Furthermore this is the post backup script:
>
> @echo off
> diskshadow -s "c:\Program Files\Bareos\Scripts\post_backup_ex_diskshadow.txt"
>
> pre_backup_diskshadow.txt:
>
> #DiskShadow script file
> delete shadows all
> set context persistent
> begin backup
> add volume E: alias ExVolumeShadow
> create
> expose %ExVolumeShadow% b:
> end backup
>
> pre_backup_diskshadow_nofull.txt
>
> #DiskShadow script file
> delete shadows all
> set context persistent
> begin backup
> add volume E: alias ExVolumeShadow
> create
> expose %ExVolumeShadow% b:
>
> Note that the only difference is we are not issuing and end backup command
> here. That cancels the VSS operation and Exchange does not flush logs.
>
> This produces an error message in the Eventlog which can be ignored but I
> don't know of a better solution right now.
>
> A drawback is that end backup in case of a full backup will be issued BEFERE
> the actual data is flowing from the snapshot to the backup medium. I
> currently don't know a sultion for this cause of the design of Bareos
> pre/post scripts and disshadow. One can call a script from within diskshadow
> to do the backup but that does not help us. I don't see a possibility to quit
> diskshadow without destroying the context thats why I need to call end backup
> in the pre script else Exchange will not flush logs. You COULD instead run
> all the client stuff asynchronous and copy the Exchange Data somewhere
> manually where you grad it later with Bareos but this seemed to complicated
> for me. Instead I installed a job repition in case of a failing backup- which
> is not 1000% safe but seems good enough to me.
>
> The last one is the rather boring post_backup_ex_diskshatow.txt:
>
> #DiskShadow script file
> delete shadows all
>
> Which should in case of a successful backup delete the shadow copy. Possibly
> one could modify the pre script to not remove the shadow that would lead in
> case of a failure of a persisting b drive which can be copied the next time
> the backup runs. But this all is up to you.
>
> I hope this helps other users too.
>
> regards, Felix
--
You received this message because you are subscribed to the Google Groups
"bareos-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
For more options, visit https://groups.google.com/d/optout.
