Re: [Bacula-users] Problems with usage of Copy Jobs
Moin,
2010/7/22 c.kesch...@internet-mit-iq.de:
I don't see anything wront with it. Can you post
/etc/bacula/INCLUDES/schedules.conf ?
Sure, but I don't think that it would change anything, because the
described behaviour also happens if I comment out the schedule entry
and do a manually run job=DiskToTape yes on the console.
The schedule in question would be this one:
Schedule {
Name = Bandsicherung
# Run = Full mon at 10:42
}
For testing purpose, I've just tested the behaviour out again. The
current state is: all full backups are on tape, which means that run
job=DiskToTape yes does not do anything currently.
*run job=DiskToTape yes
Using Catalog MyCatalog
Job queued. JobId=54422
23-Jul 11:01 backup-dir JobId 54422: No JobIds found to copy.
So, after that, I've done one explicit full backup of one client:
*run job=balance.company.com level=full yes
Job queued. JobId=54423
23-Jul 11:01 backup-dir JobId 54423: Start Backup JobId 54423,
Job=balance.company.com.2010-07-23_11.01.54_18
23-Jul 11:01 backup-dir JobId 54423: Using Device FileStorageFull
23-Jul 11:01 backup-sd JobId 54423: Volume FileStorage-Full-1779
previously written, moving to end of data.
23-Jul 11:01 backup-sd JobId 54423: Ready to append to end of Volume
FileStorage-Full-1779 size=10912854848
[...]
23-Jul 11:06 backup-sd JobId 54423: Job write elapsed time = 00:04:15,
Transfer rate = 1.447 M Bytes/second
So, now I got exactly one full backup which should be considered for
the copy job. Now I've started the copy procedure manually:
*run job=DiskToTape yes
Job queued. JobId=54424
23-Jul 11:11 backup-dir JobId 54424: The following 1 JobId was chosen
to be copied: 54423
23-Jul 11:11 backup-dir JobId 54424: Copying using JobId=54423
Job=balance.company.com.2010-07-23_11.01.54_18
23-Jul 11:11 backup-dir JobId 54424: Start Copying JobId 54424,
Job=DiskToTape.2010-07-23_11.11.21_19
[...]
23-Jul 11:11 backup-dir JobId 54424: Bacula backup-dir 5.0.2
(28Apr10): 23-Jul-2010 11:11:47
Build OS: i686-pc-linux-gnu suse 11
Prev Backup JobId: 54423
Prev Backup Job:balance.company.com.2010-07-23_11.01.54_18
New Backup JobId: 54425
Current JobId: 54424
Current Job:DiskToTape.2010-07-23_11.11.21_19
Backup Level: Full
So, now I've got the current list of Terminated Jobs output:
54423 Full 52,624362.3 M OK 23-Jul-10 11:06 balance.company.com
54425 Incr 52,624369.0 M OK 23-Jul-10 11:11 balance.company.com
54424 Full 52,624369.0 M OK 23-Jul-10 11:11 DiskToTape
You see what I mean? Where does the jobid 54425 come from? :)
Le deagh dhùraghd,
Frank Altpeter
--
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@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Bacula 5.0.2 restore files from tape very slow
On Thu, 22 Jul 2010 14:37:22 -0600, May, John said: The point I was making with my original post was that the file restore of the 6.2TB of data is taking many times longer than it did to back it up. For instance, I started the backup last Friday 7/22 and it still isn't finished today, only about half-finished. If the original backup time was 37 hours, shouldn't the restore time be about the same or around the same time, not 10 times longer? Yes, if the hardware is working optimally. Restore doesn't use spooling though, so it can trigger shoe-shining more easily that backup. __Martin -John -Original Message- From: Martin Simmons [mailto:mar...@lispworks.com] Sent: Thursday, July 22, 2010 1:51 PM To: bacula-users@lists.sourceforge.net Subject: Re: [Bacula-users] Bacula 5.0.2 restore files from tape very slow On Thu, 22 Jul 2010 08:51:58 -0600, May, John said: Bacula says the durantion for the job was 36:49:03 to backup 6.2TB. So, the time it takes to backup the job affects the time it takes to restore? No, but both times are affected by the size of the job. But isn't that obvious? Maybe I misunderstood your question. __Martin -John -Original Message- From: Martin Simmons [mailto:mar...@lispworks.com] Sent: Thursday, July 22, 2010 5:50 AM To: bacula-users@lists.sourceforge.net Subject: Re: [Bacula-users] Bacula 5.0.2 restore files from tape very slow On Wed, 21 Jul 2010 13:59:34 -0600, May, John said: As far as I know it was running at about 70MB/s. It could have been shoe-shining a bit though, not really sure. 600GB at 70MB/s would only take 2.5 hours so it can't have been getting that at a sustained rate. Shoe-shining sounds like a real possibility. -John -Original Message- From: Martin Simmons [mailto:mar...@lispworks.com] Sent: Wednesday, July 21, 2010 1:26 PM To: bacula-users@lists.sourceforge.net Subject: Re: [Bacula-users] Bacula 5.0.2 restore files from tape very slow On Mon, 19 Jul 2010 07:50:41 -0600, May, John said: Monday morning, 3 days later and it is still restoring. It's finished about 600GB, which is less than half of what is on the two tapes. I'm pretty sure that only one job was written to the tape at the time of backup. Also, I thought 'bextract' would be faster than a normal restore, since I'm telling it restore everything on tape, and not just individual jobs. Is there anyway I can speed this up? Next up is an 8TB restore, and I dread to think how long that will take. Do you know how the tape drive was functioning during the restore? Was it continually repositioning the tape (shoe-shining)? That would happen if Bacula (or the target filesystem) was unable to keep up with the speed at which the drive was sending data. __Martin John -Original Message- From: May, John [mailto:john@fugrohorizons.com] Sent: Friday, July 16, 2010 7:52 AM To: bacula-users@lists.sourceforge.net Subject: Re: [Bacula-users] Bacula 5.0.2 restore files from tape very slow Hi John... Is it possible that the backup job you are trying to restore from was running at the same time as other backup jobs were running (eg: concurrent jobs enabled)? If yes, then what might be going on here is that several jobs from several clients were written to the tape at the same time, and the data from each job was interleaved, so the write to tape during the backup was fast, but now the trade-off is that restores will be slow since the tape drive needs to read parts of the files it wants to restore, skip the files from other servers/jobs, lather, rinse, repeat. It could also explain the initial fast restore of some data as it is possible that this particular backup job you are trying to restore was already in progress when the other started so it had full access to the storage and it was the only one writing to the tape. Something to consider. Well, this was an archive job and I ran it during the day when no other backups were running. Also, I have Bacula configured to only write a single job to each tape for the archive pool (Maximum Volume Jobs = 1). John -Original Message- From: Bill Arlofski [mailto:waa-bac...@revpol.com] Sent: Thursday, July 15, 2010 7:06 PM To: bacula-users@lists.sourceforge.net Subject: Re: [Bacula-users] Bacula 5.0.2 restore files from tape very slow On 07/15/10 14:07, May, John wrote: I am using Bacula 5.0.2 using MySQL on Centos 5.4 x64. I'm trying to restore some files from a backup I made a couple months ago. I the restore is about 1.5 TB in size and spans two LTO4 tapes. I can start the
Re: [Bacula-users] Installing Bacula on Red Hat Enterprise Linux Server 5.5 (Tikanga)
On Friday, 23 July 2010 02:13:51 +0200, Jeroen van Meeuwen wrote: Hi there Daniel! Hi, Jeroen! But when trying to run the startup script, I get the following errors: [r...@localhost scripts]# /etc/init.d/bacula start Starting the Bacula Storage daemon /usr/sbin/bacula-sd: error while loading shared libraries: libbacpy-5.0.1.so: cannot open shared object file: No such file or directory Starting the Bacula File daemon /usr/sbin/bacula-fd: error while loading shared libraries: libbacfind-5.0.1.so: cannot open shared object file: No such file or directory Starting the Bacula Director daemon /usr/sbin/bacula-dir: error while loading shared libraries: libbacfind-5.0.1.so: cannot open shared object file: No such file or directory But the files are: [r...@localhost scripts]# ll /usr/lib/libbacpy-5.0.1.so -rwxr-xr-- 1 root root 24083 jul 22 12:25 /usr/lib/libbacpy-5.0.1.so [r...@localhost scripts]# ll /usr/lib/libbacfind-5.0.1.so -rwxr-xr-- 1 root root 228325 jul 22 12:25 /usr/lib/libbacfind-5.0.1.so Have you ran ldconfig? It generates the available libraries cache for the dynamic linker, and the dynamic linker is relatively strict about not searching various locations throughout the filesystem. You can verify dynamic linking capabilities with ldd /path/to/binary. Also, you may want to consider Security Enhanced Linux. Verify with the sestatus command whether SELinux is enforcing the targeted policy, and if so, consider restoring the default context for the files freshly installed using restorecon, and if all else fails, please get back to us with ls -Z output ;-) I was also trying to install Bacula 5.0.1 using the same procedure as mentioned in previous email and I have the same problem. This procedure was successful in Debian with the difference that there I updated a previous version instead of installing from scratch. Like I said, I followed the steps mentioned in the other email (I did not run ldconfig manually) that had been successful in Debian. Using ldd with these two libraries, I get the following: [r...@localhost init.d]# ldd /usr/lib/libbacpy-5.0.1.so libstdc++.so.6 = /usr/lib64/libstdc++.so.6 (0x2b4eb5ee7000) libm.so.6 = /lib64/libm.so.6 (0x2b4eb61e7000) libc.so.6 = /lib64/libc.so.6 (0x2b4eb646a000) libgcc_s.so.1 = /lib64/libgcc_s.so.1 (0x2b4eb67c2000) /lib64/ld-linux-x86-64.so.2 (0x00320a40) [r...@localhost init.d]# ldd /usr/lib/libbacfind-5.0.1.so libstdc++.so.6 = /usr/lib64/libstdc++.so.6 (0x2b97a2131000) libm.so.6 = /lib64/libm.so.6 (0x2b97a2431000) libc.so.6 = /lib64/libc.so.6 (0x2b97a26b4000) libgcc_s.so.1 = /lib64/libgcc_s.so.1 (0x2b97a2a0c000) /lib64/ld-linux-x86-64.so.2 (0x00320a40) Comparing this output with the one I get with the same version of Bacula (5.0.1) installed with the same procedure in Debian, I'm seeing that in both files are additionally a reference to linux-vdso.so.1. Can this be the cause of the problem? Thanks for your reply. Regards, Daniel -- Fingerprint: BFB3 08D6 B4D1 31B2 72B9 29CE 6696 BF1B 14E6 1D37 Powered by Debian GNU/Linux Lenny - Linux user #188.598 signature.asc Description: Digital signature -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Problems with usage of Copy Jobs
That seems strange. I presume you also didn't get a mail for job 54425? I also
don't know what to make of this. You could check the volumes to see if
there was actually data written for the incremental job.
Moin,
2010/7/22 c.kesch...@internet-mit-iq.de:
I don't see anything wront with it. Can you post
/etc/bacula/INCLUDES/schedules.conf ?
Sure, but I don't think that it would change anything, because the
described behaviour also happens if I comment out the schedule entry
and do a manually run job=DiskToTape yes on the console.
The schedule in question would be this one:
Schedule {
Name = Bandsicherung
# Run = Full mon at 10:42
}
For testing purpose, I've just tested the behaviour out again. The
current state is: all full backups are on tape, which means that run
job=DiskToTape yes does not do anything currently.
*run job=DiskToTape yes
Using Catalog MyCatalog
Job queued. JobId=54422
23-Jul 11:01 backup-dir JobId 54422: No JobIds found to copy.
So, after that, I've done one explicit full backup of one client:
*run job=balance.company.com level=full yes
Job queued. JobId=54423
23-Jul 11:01 backup-dir JobId 54423: Start Backup JobId 54423,
Job=balance.company.com.2010-07-23_11.01.54_18
23-Jul 11:01 backup-dir JobId 54423: Using Device FileStorageFull
23-Jul 11:01 backup-sd JobId 54423: Volume FileStorage-Full-1779
previously written, moving to end of data.
23-Jul 11:01 backup-sd JobId 54423: Ready to append to end of Volume
FileStorage-Full-1779 size=10912854848
[...]
23-Jul 11:06 backup-sd JobId 54423: Job write elapsed time = 00:04:15,
Transfer rate = 1.447 M Bytes/second
So, now I got exactly one full backup which should be considered for
the copy job. Now I've started the copy procedure manually:
*run job=DiskToTape yes
Job queued. JobId=54424
23-Jul 11:11 backup-dir JobId 54424: The following 1 JobId was chosen
to be copied: 54423
23-Jul 11:11 backup-dir JobId 54424: Copying using JobId=54423
Job=balance.company.com.2010-07-23_11.01.54_18
23-Jul 11:11 backup-dir JobId 54424: Start Copying JobId 54424,
Job=DiskToTape.2010-07-23_11.11.21_19
[...]
23-Jul 11:11 backup-dir JobId 54424: Bacula backup-dir 5.0.2
(28Apr10): 23-Jul-2010 11:11:47
Build OS: i686-pc-linux-gnu suse 11
Prev Backup JobId: 54423
Prev Backup Job:balance.company.com.2010-07-23_11.01.54_18
New Backup JobId: 54425
Current JobId: 54424
Current Job:DiskToTape.2010-07-23_11.11.21_19
Backup Level: Full
So, now I've got the current list of Terminated Jobs output:
54423 Full 52,624362.3 M OK 23-Jul-10 11:06 balance.company.com
54425 Incr 52,624369.0 M OK 23-Jul-10 11:11 balance.company.com
54424 Full 52,624369.0 M OK 23-Jul-10 11:11 DiskToTape
You see what I mean? Where does the jobid 54425 come from? :)
Le deagh dhùraghd,
Frank Altpeter
--
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@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users--
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@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] Remote second backup
Hi, i've configured a bacula system in my local network. I need to do the backup locally for a fast backup and a fast retore, but i also need a remote backup for safety reason. I'm think of doing a remote copy of the volume that bacula creates. The problem is that the volume is very large, about 118 Gb at the moment. And i need to copy the data in a secure way (btw the data is already encrypted with bacula PKI function). I don't think that i can do any type of incremental backup with this volume, because it's one big file that changes every day. My idea is to use bacula with a specific job that use a different storage. This storage is a mount point for the remote computer filesystem. I can introduce here some sort of protection with a vpn or other solution, i also thik of using bacula tls capabilities but then i'm forced to use tls in the local network, and this is not necessary. I've two option: 1. Do a backup of the volume (the volume contains the backup that bacula do every day) 2. Re-do every backup that bacula already did, but this time in the remote location (i think this is not a good idea) At this point I don't know if i can do this with bacula ... any other ideas? I'm using bacula 2.4.4 with ssl support and i've an ADSL line with 20Mbits/sec down and 512 Kbits/sec up. On the remote end there is a computer with similar internet connection. Regards, Marco Zancanaro -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Problems with usage of Copy Jobs
Moin, 2010/7/23 c.kesch...@internet-mit-iq.de: That seems strange. I presume you also didn't get a mail for job 54425? Correct. I also don't know what to make of this. You could check the volumes to see if there was actually data written for the incremental job. I checked, and it seems that the job did not actually run, but only got entered (and displayed) twice, as the database tells me: mysql select JobId,Job,Name,Type,Level,JobFiles,JobBytes,ReadBytes from Job where JobId in ( 54423, 54424, 54425 ) limit 10; +---+--+---+--+---+--+---+---+ | JobId | Job | Name | Type | Level | JobFiles | JobBytes | ReadBytes | +---+--+---+--+---+--+---+---+ | 54423 | balance.company.com.2010-07-23_11.01.54_18 | balance.company.com | B| F |52624 | 362371099 | 899243991 | | 54424 | DiskToTape.2010-07-23_11.11.21_19| DiskToTape | c| F |0 | 0 | 0 | | 54425 | balance.company.com.2010-07-23_11.11.22_20 | balance.company.com | C| F |52624 | 369022937 | 0 | +---+--+---+--+---+--+---+---+ Quite strange, since jobid 54424 does not have any data according to the mysql entry. Le deagh dhùraghd, Frank Altpeter -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Remote second backup
Marco Zancanaro marco.za...@gmail.com kirjoitti viestissä news:aanlktilg5xz50f0nobugolb_qngdmsa1xvzbpnqjr...@mail.gmail.com... Hi, i've configured a bacula system in my local network. I need to do the backup locally for a fast backup and a fast retore, but i also need a remote backup for safety reason. I'm think of doing a remote copy of the volume that bacula creates. The problem is that the volume is very large, about 118 Gb at the moment. And i need to copy the data in a secure way (btw the data is already encrypted with bacula PKI function). I don't think that i can do any type of incremental backup with this volume, because it's one big file that changes every day. My idea is to use bacula with a specific job that use a different storage. This storage is a mount point for the remote computer filesystem. I can introduce here some sort of protection with a vpn or other solution, i also thik of using bacula tls capabilities but then i'm forced to use tls in the local network, and this is not necessary. I've two option: 1. Do a backup of the volume (the volume contains the backup that bacula do every day) 2. Re-do every backup that bacula already did, but this time in the remote location (i think this is not a good idea) At this point I don't know if i can do this with bacula ... any other ideas? I'm using bacula 2.4.4 with ssl support and i've an ADSL line with 20Mbits/sec down and 512 Kbits/sec up. On the remote end there is a computer with similar internet connection. So you effectively have 512kbit/s, or 64 kB/s symmetric bandwith. 100 GB over this connection would take over 400 hours, over 2 weeks. Provided the connection is reliable enough, so there would be no fatal timeout errors during this time frame, causing the system to stop running. And, if your connections are dedicated for this purpose, and the bandwith is not significantly needed for some other use. I think your only option is to take care of full backups locally, and thereafter physically move the media to remote location. Then, if your data does not change too much daily, you maybe could run incremental backups to remote storage. But backing up the volume would mean that you have to back up the whole volume (100 GB could be split to smaller chunks). So I think you'd better run separate jobs to remote storage. In 24 hours, 64 kB/s can transfer 5.5 GB. So, if you'll get maybe less than 2 GB new data daily, the incrementals might work this way (assuming you can spend an average of 50% of the bandwith for backups) Note that running incrementals more or less often does not help -it's a relationship between bandwith and the rate you'll get new data. Whether to do this with Bacula or using some other means is only a question after you know your bandwith is adequate. Also remember that a possible restore takes as much time. Since it's about a secondary location for improved security, needing this remote data is probably related to some sort of disaster. So I guess in that hopefully rare occasion you may be able to physically get the remote media on-site. -- TiN -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] Bacula Status report
Hello, This is a sort of mini-Bacula status report on the following: 1. Next release 2. New release cycle 3. New bugs tracking database 4. New Bacula server (www.bacula.org) 5. New Bacula source distribution server 1. Next release: Before the end of August, we will be releasing the next version of Bacula -- version 5.0.3 which is a bug fix update to 5.0.2. This release is almost ready and the most recent code is in the SF bacula git repository under Branch-5.0 2. New release cycle: The little code we currently have for the next major release is in the SF bacula git repository under Branch-5.1. We are considering to moving to a regular 6 month release cycle. The advantage of such a cycle is that it gets features out to you faster. The disadvantage is that it doesn't work so well in small projects like Bacula if there are not sufficient contributions. Such a release would consist of the following points: - A release every 6 months - The deadline is not absolute and could be extended to 9 months if there were insufficient new submissions. - There will be far fewer or no bug fix updates as they are not really needed if we can maintain a 6 month cycle. - Two months before the projected release we will decide if there are sufficient new features to release - The release count down will consist of 3 phases 1. We will add all new approved features The first 4 months after a release this phase will go into effect for the next release - 2. Only very small new features (a few lines) will be added Two months before the final release this phase will go into effect. Note, this phase can be delayed 3 months if insufficient new features are submitted 3. Only bug fixes This phase will go into effect one month before the release Under this scheme, we are currently in Phase 3 for the 5.0.3 release, and the next major release (5.2.0) would be made before mid-January 2011, and is currently under development in Branch-5.1 on Source Forge. I would appreciate comments on this proposed new deadline release cycle. 3. New bugs tracking database Sometime in early August (possibly slightly before) we will be moving the current Mantis based bug tracking system to a new RT based system hosted by Bacula Systems. The upside is that the RT system is far more powerful, flexible and adaptible, and most important of all, it allows email responses to bugs. The downside is that it is a bit more complicated (as are most things that have more features) and that it will require everyone to re-register for the new system. In addition, if you don't want to rely on just the community to furnish bug fixes, you will be able to subscribe to a bug fix service that is more professional and has a guaranteed response time (not to be mistaken for a guaranteed fix time). More on this when the service is ready for production. 4. New Bacula server The current Bacula Community server is as you probably know generously offered by UKFast. However, the hardware is starting to age, so they have gratiously provided us with a new machine that we will be putting in place in the next few weeks. We don't expect that you will notice any differences, but the hardware running www.bacula.org should be more stable. 5. New Bacula source distribution server You may or may not be aware that we have not always been pleased with the services offered by Source Forge. The uploading is complicated by lines dropping (I have *never* seen this else where), their user interface is horrible, we don't get good statistics, being US based, they block direct access to our code from a number of countries such as Cuba, ... So, probably in September or October we will be moving our Bacula project off of Source Forge to a new server provided by UKFast. There is still a *lot* of work to be done to make this work -- principally getting up a good and suitable interface for users -- more as this develops. As mentioned above, I would appreciate any comments you might have, particularly on the proposed new release cycle. Best regards, Kern -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Remote second backup
Ok, thanks for helping. I'll try to find a suitable solution based on you suggestion. Marco On Fri, Jul 23, 2010 at 16:45, Timo Neuvonen timo-n...@tee-en.net wrote: Marco Zancanaro marco.za...@gmail.com kirjoitti viestissä news:aanlktilg5xz50f0nobugolb_qngdmsa1xvzbpnqjr...@mail.gmail.com... Hi, i've configured a bacula system in my local network. I need to do the backup locally for a fast backup and a fast retore, but i also need a remote backup for safety reason. I'm think of doing a remote copy of the volume that bacula creates. The problem is that the volume is very large, about 118 Gb at the moment. And i need to copy the data in a secure way (btw the data is already encrypted with bacula PKI function). I don't think that i can do any type of incremental backup with this volume, because it's one big file that changes every day. My idea is to use bacula with a specific job that use a different storage. This storage is a mount point for the remote computer filesystem. I can introduce here some sort of protection with a vpn or other solution, i also thik of using bacula tls capabilities but then i'm forced to use tls in the local network, and this is not necessary. I've two option: 1. Do a backup of the volume (the volume contains the backup that bacula do every day) 2. Re-do every backup that bacula already did, but this time in the remote location (i think this is not a good idea) At this point I don't know if i can do this with bacula ... any other ideas? I'm using bacula 2.4.4 with ssl support and i've an ADSL line with 20Mbits/sec down and 512 Kbits/sec up. On the remote end there is a computer with similar internet connection. So you effectively have 512kbit/s, or 64 kB/s symmetric bandwith. 100 GB over this connection would take over 400 hours, over 2 weeks. Provided the connection is reliable enough, so there would be no fatal timeout errors during this time frame, causing the system to stop running. And, if your connections are dedicated for this purpose, and the bandwith is not significantly needed for some other use. I think your only option is to take care of full backups locally, and thereafter physically move the media to remote location. Then, if your data does not change too much daily, you maybe could run incremental backups to remote storage. But backing up the volume would mean that you have to back up the whole volume (100 GB could be split to smaller chunks). So I think you'd better run separate jobs to remote storage. In 24 hours, 64 kB/s can transfer 5.5 GB. So, if you'll get maybe less than 2 GB new data daily, the incrementals might work this way (assuming you can spend an average of 50% of the bandwith for backups) Note that running incrementals more or less often does not help -it's a relationship between bandwith and the rate you'll get new data. Whether to do this with Bacula or using some other means is only a question after you know your bandwith is adequate. Also remember that a possible restore takes as much time. Since it's about a secondary location for improved security, needing this remote data is probably related to some sort of disaster. So I guess in that hopefully rare occasion you may be able to physically get the remote media on-site. -- TiN -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Remote second backup
2010/7/23 Marco Zancanaro marco.za...@gmail.com: Hi, i've configured a bacula system in my local network. I need to do the backup locally for a fast backup and a fast retore, but i also need a remote backup for safety reason. I'm think of doing a remote copy of the volume that bacula creates. The problem is that the volume is very large, about 118 Gb at the moment. And i need to copy the data in a secure way (btw the data is already encrypted with bacula PKI function). I don't think that i can do any type of incremental backup with this volume, because it's one big file that changes every day. Is this a disk volume? If so I believe rsync would handle the volume correctly. Remember bacula's disk format is the same as tape so it only appends and does not alter previously written data. John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Bacula Status report
Thanks for all the work you do, this sounds great. Sounds a lot like the Fedora Linux release cycle the last time I looked at it. Good luck on moving to the new server and thanks to UKFast. Thomas On Friday 23 July 2010 11:54:03 Kern Sibbald wrote: Hello, This is a sort of mini-Bacula status report on the following: 1. Next release 2. New release cycle 3. New bugs tracking database 4. New Bacula server (www.bacula.org) 5. New Bacula source distribution server 1. Next release: Before the end of August, we will be releasing the next version of Bacula -- version 5.0.3 which is a bug fix update to 5.0.2. This release is almost ready and the most recent code is in the SF bacula git repository under Branch-5.0 2. New release cycle: The little code we currently have for the next major release is in the SF bacula git repository under Branch-5.1. We are considering to moving to a regular 6 month release cycle. The advantage of such a cycle is that it gets features out to you faster. The disadvantage is that it doesn't work so well in small projects like Bacula if there are not sufficient contributions. Such a release would consist of the following points: - A release every 6 months - The deadline is not absolute and could be extended to 9 months if there were insufficient new submissions. - There will be far fewer or no bug fix updates as they are not really needed if we can maintain a 6 month cycle. - Two months before the projected release we will decide if there are sufficient new features to release - The release count down will consist of 3 phases 1. We will add all new approved features The first 4 months after a release this phase will go into effect for the next release - 2. Only very small new features (a few lines) will be added Two months before the final release this phase will go into effect. Note, this phase can be delayed 3 months if insufficient new features are submitted 3. Only bug fixes This phase will go into effect one month before the release Under this scheme, we are currently in Phase 3 for the 5.0.3 release, and the next major release (5.2.0) would be made before mid-January 2011, and is currently under development in Branch-5.1 on Source Forge. I would appreciate comments on this proposed new deadline release cycle. 3. New bugs tracking database Sometime in early August (possibly slightly before) we will be moving the current Mantis based bug tracking system to a new RT based system hosted by Bacula Systems. The upside is that the RT system is far more powerful, flexible and adaptible, and most important of all, it allows email responses to bugs. The downside is that it is a bit more complicated (as are most things that have more features) and that it will require everyone to re-register for the new system. In addition, if you don't want to rely on just the community to furnish bug fixes, you will be able to subscribe to a bug fix service that is more professional and has a guaranteed response time (not to be mistaken for a guaranteed fix time). More on this when the service is ready for production. 4. New Bacula server The current Bacula Community server is as you probably know generously offered by UKFast. However, the hardware is starting to age, so they have gratiously provided us with a new machine that we will be putting in place in the next few weeks. We don't expect that you will notice any differences, but the hardware running www.bacula.org should be more stable. 5. New Bacula source distribution server You may or may not be aware that we have not always been pleased with the services offered by Source Forge. The uploading is complicated by lines dropping (I have *never* seen this else where), their user interface is horrible, we don't get good statistics, being US based, they block direct access to our code from a number of countries such as Cuba, ... So, probably in September or October we will be moving our Bacula project off of Source Forge to a new server provided by UKFast. There is still a *lot* of work to be done to make this work -- principally getting up a good and suitable interface for users -- more as this develops. As mentioned above, I would appreciate any comments you might have, particularly on the proposed new release cycle. Best regards, Kern --- --- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users -- == Thomas McMillan Grant Bennett
[Bacula-users] recycle volume
I have a volume that I am recycling back into the pool before the volume expiration date. How do I simply wipe this tape or start the write position to the beginning of the tape to overwrite the old data? Thanks, JJ -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] recycle volume
On 07/23/2010 06:45 PM, Jeremiah D. Jester wrote: I have a volume that I am recycling back into the pool before the volume expiration date. How do I simply wipe this tape or start the write position to the beginning of the tape to overwrite the old data? When I've to force recycling a tape, for first I delete it from the db (delete volume=yadayada...), then I umount the tape and out of bconsole I do: # mt -f /dev/nst0 weof Once the tape was erased, I re-login into bconsole and relabel the tape with the oldname (yadayada...). Not very elegant, but functional. Simone -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] recycle volume
I have a volume that I am recycling back into the pool before the volume expiration date. How do I simply wipe this tape or start the write position to the beginning of the tape to overwrite the old data? purge volume John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] recycle volume
Does this do the same thing as purge volume? Is one preferable over the other? *update volume=VOLNAME pool=Offsite slots volstatus=recycle Recycle=yes JJ -Original Message- From: John Drescher [mailto:dresche...@gmail.com] Sent: Friday, July 23, 2010 9:54 AM To: Jeremiah D. Jester Cc: bacula-users Subject: Re: [Bacula-users] recycle volume I have a volume that I am recycling back into the pool before the volume expiration date. How do I simply wipe this tape or start the write position to the beginning of the tape to overwrite the old data? purge volume John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] recycle volume
Does this do the same thing as purge volume? Is one preferable over the other? *update volume=VOLNAME pool=Offsite slots volstatus=recycle Recycle=yes Purge cleans up the database (removes db records for jobs and files that were on the volume) while update volume I am not sure. John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] recycle volume
2010/7/23 Simone Martina smart...@noc.skylogicnet.com: On 07/23/2010 06:45 PM, Jeremiah D. Jester wrote: I have a volume that I am recycling back into the pool before the volume expiration date. How do I simply wipe this tape or start the write position to the beginning of the tape to overwrite the old data? When I've to force recycling a tape, for first I delete it from the db (delete volume=yadayada...), then I umount the tape and out of bconsole I do: # mt -f /dev/nst0 weof Once the tape was erased, I re-login into bconsole and relabel the tape with the oldname (yadayada...). Not very elegant, but functional. purge volume should be sufficient. John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] Volumes that are not really full.
I have a volume (KL0445) reporting as Full though this seems suspect giving that the VolBytes column is so much smaller than the rest of my tapes. Included a snippet. Any reason for this or does this mean there is an error? Should I purge the volume and reuse? | 77 | KL0406 | Full | 1 | 226,718,908,416 | 227 | 15,552,000 | 1 |0 | 0 | LTO2 | 2010-06-23 03:41:12 | | 80 | KL0445 | Full | 1 | 129,024 |2 | 7,776,000 | 1 | 12 | 1 | LTO2 | 2010-07-14 16:05:03 | | 83 | KL0393 | Full | 1 | 617,731,623,936 | 618 | 15,552,000 | 1 |0 | 0 | LTO2 | 2010-06-23 07:13:18 | Thanks, JJ -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
[Bacula-users] Fwd: Volumes that are not really full.
I have a volume (KL0445) reporting as Full though this seems suspect giving that the VolBytes column is so much smaller than the rest of my tapes. Included a snippet. Any reason for this or does this mean there is an error? Should I purge the volume and reuse? When bacula hits a media error writing it assumes that the tape is full. John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Volumes that are not really full.
Does a media error mean what it says? Problem with the media. Do I need to basically test each volume w/ btape that I get an error with. Seems like a lot of work. Could be a bad tape, kernel problem, sas/scsi problem, tape drive ... Basically when the tape driver reports a write error to bacula, bacula marks the volume as full. John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Volumes that are not really full.
Shouldn't it label as Error and not Full? Is there a simple way to query all Errored volumes without grepping the log file? Thanks, JJ -Original Message- From: John Drescher [mailto:dresche...@gmail.com] Sent: Friday, July 23, 2010 11:06 AM To: Jeremiah D. Jester; bacula-users Subject: Re: [Bacula-users] Volumes that are not really full. Does a media error mean what it says? Problem with the media. Do I need to basically test each volume w/ btape that I get an error with. Seems like a lot of work. Could be a bad tape, kernel problem, sas/scsi problem, tape drive ... Basically when the tape driver reports a write error to bacula, bacula marks the volume as full. John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
Re: [Bacula-users] Volumes that are not really full.
Shouldn't it label as Error and not Full? No. Bacula does not have a platform unspecific way to tell if the error is because the tape reached its end or that the tape was not writable past some point. Error is only for volumes that bacula detects a mismatch between the # of volume files that the catalog says versus what is on the tape. Is there a simple way to query all Errored volumes without grepping the log file? Since bacula does not know there is no log of this. John -- 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@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
