[bareos-dir : job] - How to modify an existing job with bconsole

[4kn...@gmail.com]

Hello

I start with the bareos so I'm sorry in advance if my question seem stupid.

How to modify an existing job with bconsole such as
- to modify : pool, fileset (to modify or to add new fileset), schedule and so on (by modifying the JobDefs parameter)
- to add a new client (of course if it is possible that a job can have several clients)

Thank you in advance for your help

Regards
Knoax

[Bruno Friedmann]

On vendredi, 12 mai 2017 11.21:00 h CEST 4kn...@gmail.com wrote:
> Hello
>
> I start with the bareos so I'm sorry in advance if my question seem stupid.
>
> How to modify an existing job with bconsole such as
> - to modify : pool, fileset (to modify or to add new fileset), schedule and
> so on (by modifying the JobDefs parameter)

Not sure what you mean by this, at manual run job you can change all
parameters and thus override whatever is defined in that job (and the jobDefs
heritage

> >- to add a new client (of course
> if it is possible that a job can have several clients)
>

You need a job by client (or several jobs by client).
Everything else than the client name target can be defined in JobDefs,
and then you mainly have to change your JobDefs reload and all jobs using it
will be changed accordly.

> Thank you in advance for your help
>
> Regards
> Knoax

--

Bruno Friedmann
Ioda-Net Sàrl www.ioda-net.ch
Bareos Partner, openSUSE Member, fsfe fellowship
GPG KEY : D5C9B751C4653227
irc: tigerfoot

openSUSE Tumbleweed
Linux 4.10.13-1-default x86_64 GNU/Linux, nvidia: 375.66
Qt: 5.7.1, KDE Frameworks: 5.33.0, Plasma: 5.9.5, kmail2 5.4.3

[4kn...@gmail.com]

Hello

Thank you very much for your reply Bruno.

Sorry I was not very clear in my request.

In fact, I wanted to know if it was possible to modify the configuration of an already existing job by simply modifying the config file (only for scheduled run and not manual run).

If I understand your answer correctly, it seems impossible to add a new client to a job. A job can include only one client, not multiple clients.

Currently, my Arkeia backup solution allows me to start a job that backups several clients at the same time.
All I need to do is add the new client to the existing job so that the new client can be backuped like any other server already configured (full backup with a retention of one week and at a specific time of the day)

It allows me to back up all my production servers every night at the same time.
I wanted to know if it was possible to do the same thing with bareos.

By reading different answers in the "bareos-user" mailling list and in the bareos-manual-main-reference, it seems complicated to backup several clients at the same time.
You must either create multiple devices, or make interlaced backups, or use the data spooling method.
Each previous method has disadvantages.

I think I'm not the only one wanting to save several clients at the same time so I'd like to know what the best method would be to use.

Thanks again for your help

Regards
Knoax

[Bruno Friedmann]

It's not complicated, some efforts has to be made to understand the system,
and translate it to configuration.

The system is so flexible, that yes you have to decide what you want to do.

A typical installation for me, will be.
Managing media
Have a full pool, and full storage, retention will be 2 months.
Have a Inc pool, and Inc storage, retention 2 months.
We will only do incremental, accurate backup and full on 1st sunday.
I always used a spool media


I define a schedule : Daily
I define a JobDefs : containing Pools definition, Storage definition, pre-post
execution, Fileset, and the Schedule.

For any client I define a job with client name, joddefs.
And because the job contain the jobdef, which contain the schedule all my
clients will start their backup at the same time.

But there's several ways of achieving the same global result.
All my client run as the same time.

In another installation, where everything as to be split, we have one pool,
one media type, one storage per client
and again a global schedule for all.

etc... ;-)

Also you have the configure command in bconsole

http://doc.bareos.org/master/html/bareos-manual-main-reference.html#ConfigurationIncludeDirectory

--

Bruno Friedmann
Ioda-Net Sàrl www.ioda-net.ch
Bareos Partner, openSUSE Member, fsfe fellowship
GPG KEY : D5C9B751C4653227
irc: tigerfoot

openSUSE Tumbleweed
Linux 4.10.13-1-default x86_64 GNU/Linux, nvidia: 375.66

Qt: 5.7.1, KDE Frameworks: 5.33.0, Plasma: 5.9.5, kmail2 5.5.0

[4kn...@gmail.com]

Hello

Thanks you very much for your explanation Bruno.
I understand better how bareos works.

My objective is to backup multiple production servers on disk (in a specific directory) at the same time and several development servers on disk too (in a directory different from those in production). So I was only interested in disk backups.

Before your answer, I studied the reference manual on bareos, I concluded that a job can only backup one client.

If I want to backup several clients then I have to create as many jobs as I have client to backup.

If I want to backup several clients at the same time, I must run several jobs at the same time.

As bareos works as if it was connected to a tape drive then by default (one device with a single storage and without data spooling) it can not run several jobs at the same time because of the concurrent disk jobs.

According to reference manual, to resolve the concurrent disk jobs problem, several devices must be created at the storage daemon.

As I have several production servers (about 20), I will have to create 20 devices then an additional device for each new production servers to backup.

So, I would like to know if bareos server have a software limit on the number of device to create.

Thank you for your help

Regards
Knoax

[Bruno Friedmann]

There's no such limit. But going the multiple device will certainly make your
daily operation more complicated (except if you have to implement complex
retention scenario).


I would go with spooling, bring all production to one device with spooling,
and all devs to another stream.

[4kn...@gmail.com]

Hello

Thank you very much for your answer, Bruno.

Indeed I read in the reference manual and on the bareos-user mailinglist that it is useful to use data spooling.

However, I did not really understand the usefulness of data spooling for my case.
In the manual it is explained that the data spooling backup firstly on disk then write on tape.
Moreover I'm not sure to understand the terms "spool" and "despool" in "Chapter 21 : Data Spooling". Sorry I'm french and my english is poor T_T. Even google doesn't want to be my friend ^_^.
So, for me it seems more complicated to use data spooling than to use multiple device to run several job at the same time (because I didn't really understand the data spooling method).

In my infrastructure I haven't a tape drive. I backup everything on a disk array.
I even plan to outsource a copy of the backups to another disk array on another site.

As I want to understand all the opportunities offered by bareos I would like to know if someone would have more information, links or more document on data spooling.

Thank you in advance for your help

Regards
Knoax

[4kn...@gmail.com]

Hello

I'm sorry but I think I was not clear.

My need is this:
- to do backups only on disk because I haven't any tape drive
- to have one job per volume (to backup data from one server only in one volume)
- to run several jobs at the same time (the job run must be independent and simultaneous with other runs, ie a job run must not wait until another run is completed)

Is it possible to configure bareos director with these criteria?

If yes... so what is the method/config you would recommend me?

Thank you for your help

Regards
Knoax

[Bruno Friedmann]

Beside taking some trainings and consulting, the answer is yes to all :-)

Your point 2 will force you to have one pool, one device, one volume per
client which will make the configuration complex once you got more clients.

Otherwise, spooling is usefull for disk too.
Backups volume goes to "cheap" harddrive raid bay so the cost per TB is low,
and you use raid0 of ssd to do the spooling for example.

They you got a bundle of jobs on the same volume. For restore, bareos will
find the right volume alone, you don't have to think about it.

ps: I speak french too in case of ;-)

[4kn...@gmail.com]

Hello

Thank you very much Bruno for your answers and your time spent helping me.

I know that Bareos finds his volumes alone and that I do not have to worry about.
But by reading the manual, it is indicated that if several jobs are run at the same time then the data are interleaved. And this interleaving of data will have the impact on restorations. These would be more slow.
I don't know exactlythe impact of the restoration slowdown so I preferred to choose the possibility offered by bareos to make a backup in a file. That means having one job per volume.

Moreover, as we plan to outsource our backups to another site, it will allow us to outsource only the files containing the priority production backups.

I will test several configurations to compare them.

Thank you again for your help
Regards
Knoax