[Support] Bacula-Server

[greyday]

Will this include LTO support?! If so that would be amazing, it's the only thing I keep a Windows VM for...

[VRx]

I haven't any, so i cannot test.

But with your little help we could install some drivers and libs if it is not working.

[greyday]

Sorry, I missed this reply. Planning to overhaul my server this next weekend so will try it out then and get back to you!

[trypowercycle]

I'm also interested in LTO support. I have a Quantum LTO5 library I'm gonna test out this weekend. Are there any drivers built in the base install?

[VRx]

On 1/22/2022 at 3:31 PM, trypowercycle said:

Are there any drivers built in the base install?

I'm sorry but no.
I try to build some new image drivers included for about 2 weeks (actually i have big reconfiguration in my lab environment)

Important!

Last week I build smaller image using debian-slim, but there is some problem on Unraid with this image:
1. "all texts has @@ at both the start and end" - https://github.com/vrx-666/bacula-server/issues/1

2. I noticed error when container is first time starting after base image change.
Today I published update resolving first issue, but this is simpla come back to standard debian image, so second problem may occur again with this update.
Simple resolution is to delete container, and install it again, with same paths, ports, ip, user, password. Nothing should disapear, any backups, settings and history.

I'm very sorry for my bad automated testing.

[b0m541]

Looking for a better solution for the following problem:

 

The JobDefs for DefaultJob is buried in really many resources, but it cannot be edited in Baculum.

I had to edit JobDefs in the bacula-dir.conf to be able to get rid of the default resources (schedules, jobs, pools, storage) coming with the container.

 

Is there a way to edit JobDefs in Baculum?

 

 

[VRx]

20 hours ago, b0m541 said:

Is there a way to edit JobDefs in Baculum?

 

Director (left menu) > Configure Director (upper menu) > JobDefs > Edit

 

[jkexbx]

Was anyone able to test this out with a LTO drive?

[b0m541]

Today I had this problem when Bacula was automatically started again after the daily tasks (incl. Mover):

 

2022-04-30 05:00:03,712 WARN received SIGTERM indicating exit request
2022-04-30 05:00:03,724 INFO waiting for baculum to die
2022-04-30 05:00:03,725 INFO stopped: baculum (terminated by SIGTERM)
==> Checking DB...
==> Starting...
==> .......Storage Daemon...
Starting Bacula Storage Daemon: bacula-sd
==> .......File Daemon...
Starting Bacula File Daemon: bacula-fd
==> .......Bacula Director...
Starting Bacula Director: bacula-dir
==> .......Bacula Web...
2022-04-30 07:07:46,348 INFO Set uid to user 0 succeeded
2022-04-30 07:07:46,383 INFO supervisord started with pid 1
2022-04-30 07:07:47,394 INFO spawned: 'baculum' with pid 149
2022-04-30 07:07:48,113 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:07:49,122 INFO spawned: 'baculum' with pid 152
2022-04-30 07:07:49,740 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:07:51,773 INFO spawned: 'baculum' with pid 155
2022-04-30 07:07:52,777 INFO success: baculum entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2022-04-30 07:07:53,132 INFO exited: baculum (exit status 0; expected)
2022-04-30 07:07:53,137 INFO spawned: 'baculum' with pid 158
2022-04-30 07:07:53,636 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:07:54,650 INFO spawned: 'baculum' with pid 161
2022-04-30 07:07:55,632 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:07:57,641 INFO spawned: 'baculum' with pid 164
2022-04-30 07:07:57,981 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:08:00,995 INFO spawned: 'baculum' with pid 167
2022-04-30 07:08:02,016 INFO success: baculum entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2022-04-30 07:08:02,291 INFO exited: baculum (exit status 0; expected)
2022-04-30 07:08:03,299 INFO spawned: 'baculum' with pid 170
2022-04-30 07:08:03,990 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:08:05,004 INFO spawned: 'baculum' with pid 173
2022-04-30 07:08:05,733 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:08:07,740 INFO spawned: 'baculum' with pid 176
2022-04-30 07:08:08,374 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:08:11,384 INFO spawned: 'baculum' with pid 179
2022-04-30 07:08:11,864 INFO exited: baculum (exit status 0; not expected)
2022-04-30 07:08:12,866 INFO gave up: baculum entered FATAL state, too many start retries too quickly

 

The baculum web UI wouldn't work, but the backup tasks ran fine.

 

[VRx]

On 4/30/2022 at 8:44 PM, b0m541 said:

Today I had this problem when Bacula was automatically started again after the daily tasks (incl. Mover

@b0m541Did Your mover touch Baculum config path or Apache log path?

[b0m541]

On 5/5/2022 at 6:55 PM, VRx said:

@b0m541Did Your mover touch Baculum config path or Apache log path?

Nope.

[VRx]

Open Your's Unraid shell and run command:
 

docker exec $(docker ps | grep "pwa666/bacula-server" | awk '{print $1}') cat /var/log/apache2/error.log

copy and paste here output

[VRx]

End of development for Bacula 9.6 images versions!
Older versions of the client work well with the newest server version. The last update of the bacula9.6 code was in December 2020.

[VRx]

I tried to do few tests on unraid version 6.10 but there are so many bugs on various applications with this version that I went back to version 6.9.2.
Some problematic applications are necessary for me for everyday use of the server. In this situation I would have to spend a lot of time diagnosing problems and correcting them, but this is not my responsibility, apparently other developers are not interested in their applications.
After initial testing it looks like Bacula-Server should work after UnRAID update

[b0m541]

The Problem with Baculum not properly starting still happens now and then. I looked in the logs the way you proposed and I just find a lot of these:

 

[Thu May 12 07:05:02.667379 2022] [mpm_prefork:notice] [pid 148] AH00163: Apache/2.4.53 (Debian) configured -- resuming normal operations
[Thu May 12 07:05:02.667730 2022] [core:notice] [pid 148] AH00094: Command line: '/usr/sbin/apache2 -D FOREGROUND'
AH00558: apache2: Could not reliably determine the server's fully qualified domain name, using 172.17.0.18. Set the 'ServerName' directive globally to suppress this message
[Fri May 13 00:22:16.190020 2022] [core:warn] [pid 149] AH00098: pid file /var/run/apache2/apache2.pid overwritten -- Unclean shutdown of previous Apache run?

 

[b0m541]

I migrated from 6.9.x to 6.10.x - so far Bacula/Baculum works happily.

[b0m541]

I have a request for the next version of this container: please include vchanger

I tried to build it from the sources myself within the container, but there is (understandably) no compiler.

 

https://sourceforge.net/projects/vchanger/files/?source=navbar
 

Why would we want this in the container? There is a very good but also quite lenghty explanation in the source archive.

TL;DR for backups to the unRAID array this is not a necessary tool, as the array can be upscaled transparently from the perspective of Bacula.

But if you have another storage tier for backups that is not on the array, but e.g. on external USB storage, you run into a nasty limitation of Bacula:

Bacula is not able to handle more than 1 logical device per job. If you write all jobs to the same device it will fill up at some point and you have no way to upscale it, you can only recycle volumes at that point.

 

vchanger makes usage of fixed-space and removable storage very flexible and comfortable. It is simply necessary to have this if one does not want to put all jobs on the array (e.g. one has multi-tiered backup layers).

 

Would you be willing to include vchanger in the container?

 

 

 

 

 

[VRx]

11 hours ago, b0m541 said:

I have a request for the next version of this container: please include vchanger

I tried to build it from the sources myself within the container, but there is (understandably) no compiler.

 

https://sourceforge.net/projects/vchanger/files/?source=navbar
 

Why would we want this in the container? There is a very good but also quite lenghty explanation in the source archive.

TL;DR for backups to the unRAID array this is not a necessary tool, as the array can be upscaled transparently from the perspective of Bacula.

But if you have another storage tier for backups that is not on the array, but e.g. on external USB storage, you run into a nasty limitation of Bacula:

Bacula is not able to handle more than 1 logical device per job. If you write all jobs to the same device it will fill up at some point and you have no way to upscale it, you can only recycle volumes at that point.

 

vchanger makes usage of fixed-space and removable storage very flexible and comfortable. It is simply necessary to have this if one does not want to put all jobs on the array (e.g. one has multi-tiered backup layers).

 

Would you be willing to include vchanger in the container?

 

 

 

 

 

 

I would have to test it. Sources say about the last update in May 2020. At the beginning of 2021, bacula 11 was released. The conclusion is that this solution does not necessarily work with bacula 11, and as I wrote above, the container with bacula 9.x will not be developed anymore.
Also, you have to remember that this is a containerized solution where mounting physical disks is more complicated.
Currently, due to recent problems with the bacula repository, I am testing a new way of building the container.
In the next version I plan to add support for sending email notifications - this is almost ready solution, but due to problems with the bacula repository, I had to deal with the CD process in the first place.

Then I will be able to work on the test implementation of the vchanger.

[b0m541]

I will check out whether vchanger 1.0.2 should work with bacula 11. As long as the autochanger commands haven't changed - I guess - it should.

But I will come back with a more definitive response.

[b0m541]

I found someone who is successfully using changer 1.0.2 with Bacula 11. So implementing vchanger in the container should be feasible.

[b0m541]

Actually in the git there is vchanger 1.0.3:

git clone https://git.code.sf.net/p/vchanger/code vchanger

 

Update: the downloads have been updated to also show 1.0.3

Edited July 2, 2022 by b0m541

[b0m541]

the latest version of the container seems not to use UID 101 any more for the volume files it is accessing. Since previous containers did write volumes with UID 101 the new container is unable to read and append to the volumes written by the previous container versions. This seems not to be a bigger problem as Bacula just creates a new volume if it cannot append to an older volume. However, this leads to warning messages in the log file which might irritate the user.

 

Warning: mount.c:216 Open of File device "redacted" (/mnt/redacted) Volume "redacted" failed: ERR=file_dev.c:189 Could not open(/mnt/redacted,OPEN_READ_WRITE,0640): ERR=Permission denied

Marking Volume "redacted" Read-Only in Catalog.

Created new Volume="redacted", Pool="redacted", MediaType="redacted" in catalog.

 

Still, I would think such a change is something we should strive to avoid.

 

I am not sure what actual improvement is brought by the latest container version.

 

I was hoping it would be including vchanger, alas, it does not. That would have been desirable.

 

Edited July 10, 2022 by b0m541

[VRx]

10 hours ago, b0m541 said:

This seems not to be a bigger problem as Bacula just creates a new volume if it cannot append

This is big problem, because Bacula can not restore anything from backup.

 

  

10 hours ago, b0m541 said:

I am not sure what actual improvement is brought by the latest container version.

You have all informations in the Change Log on CommunityApplications page. Just click Info on Bacula-Server App

 

10 hours ago, b0m541 said:

I was hoping it would be including vchanger, alas, it does not. That would have been desirable.

I was hoping you read my previous post
Especially the part about the upcoming development plans for this container.
I'm

 

10 hours ago, b0m541 said:

Warning: mount.c:216 Open of File device "redacted" (/mnt/redacted)

You changed the container path, of course you can to do so, but I think you should be able to handle a simple chown in this situation as well.
The fix is currently being tested, the image should build and publish within 1 hour

*BTW I'm still wondering about the sense of implementing vchanger, because adding another disk requires stopping the container as well as configuring the vchanger from shell. It seems to me that if someone wants to configure something in shell, they will not choose UnRAID, but install their favorite distribution, configure it for themselves, build a bacula and use it from bconsole, just like I do at work.

[b0m541]

with disk volumes there simply is no alternative to vchanger if not everything fits on 1 device. This is relevant for people who have Bacula volumes on disks that are not assigned to the array.

Edited July 11, 2022 by b0m541

[VRx]

2 hours ago, b0m541 said:

with disk volumes there simply is no alternative to vchanger if not everything fits on 1 device. This is relevant for people who have Bacula volumes on disks that are not assigned to the array.

There is, and it  requires some work with shell commands too.
You can do symlinks on the second disk to volumes saved on first (shell), and then make changes in storage settings (web UI).