On Jul 2, 2009, at 5:41 PM, randalls AT bioinfo.wsu DOT edu wrote:
> I see that Bacula 3.0 has a duplicate job control feature. Perhaps
> I need to upgrade to 3.0 and use this feature. However I believe I
> have something else wrong in my configuration. I need to read up
> on the configuration options more. Perhaps there is some setting
> that I have missed.
>
> Anyone else have experience with large jobs. I have an overland
> neo 2000 with two, lto 3 tape drives installed.
>
>
I think what limits the number of jobs is Maximum Concurrent Jobs
defined in the Director or Storage or Pool or Job resource. However,
I'm finding with Bacula 3.0.1 this configuration option is
_completely_ ignored.
Here's part of my configuration file
Director {
Name = Logger-Director-01
Password = ******
Dir Addresses = {
ipv4 = {
addr = 127.0.0.1
port = 9101
}
}
Working Directory = /opt/bacula/var/working
Pid Directory = /opt/bacula/var/working
Query File = /opt/bacula/var/scripts/query.sql
Maximum Concurrent Jobs = 10
Messages = Standard
}
JobDefs {
Name = FileStorage-Main-JobDef-01
Type = Backup
Level = Incremental
Pool =
L180-LTO3-Autochanger-Full-Pool-01
Full Backup Pool = L180-LTO3-Autochanger-Full-
Pool-01
Differential Backup Pool = L180-LTO3-Autochanger-Incremental-
Pool-01
Incremental Backup Pool = L180-LTO3-Autochanger-Incremental-
Pool-01
Priority = 10
Allow Mixed Priority = Yes
Max Start Delay = 28800
Rerun Failed Levels = No
Maximum Concurrent Jobs = 1
Write Bootstrap = /opt/bacula/var/bootstraps/%n.bsr
Messages = Standard
}
Job {
Name = Arking-Job-01
JobDefs = FileStorage-Main-JobDef-01
Client = Filer01-02
File Set = Arking-FileSet-01
Schedule = 1st-Sunday-Cycle
}
Pool {
Name = L180-LTO3-Autochanger-Full-Pool-01
Pool Type = Backup
Maximum Volume Jobs = 0
Recycle = Yes
AutoPrune = Yes
Volume Retention = 75d
ScratchPool = L180-LTO3-Autochanger-Scratch-Pool-01
RecyclePool = L180-LTO3-Autochanger-Scratch-Pool-01
Storage = Logger-L180-LTO3-Autochanger-01
}
Pool {
Name = L180-LTO3-Autochanger-Incremental-Pool-01
Pool Type = Backup
Maximum Volume Jobs = 0
Recycle = Yes
AutoPrune = Yes
Volume Retention = 40d
ScratchPool = L180-LTO3-Autochanger-Scratch-Pool-01
RecyclePool = L180-LTO3-Autochanger-Scratch-Pool-01
Storage = Logger-L180-LTO3-Autochanger-01
}
Storage {
Name = Logger-L180-LTO3-Autochanger-01
Password = ******
Address = logger01.igm.jhmi.edu
SD Port = 9103
Device = L180-Autochanger-02
Media Type = LTO3
Autochanger = Yes
Maximum Concurrent Jobs = 2
}
Client {
Name = Filer01-02
Password = *******
Address = 192.168.2.160
FD Port = 9102
Catalog = Logger-Catalog-01
File Retention = 1825d
Job Retention = 1825d
Auto Prune = Yes
Maximum Concurrent Jobs = 3
}
Here's what you would expect to happen
Run Arking-Job-01
Run Arking-Job-01 again
The second Arking-Job-01 should wait until the first finishes because
Maximum Concurrent Jobs =1 in the JobDef. That's _not_ true,
apparently. Bacula is perfectly happy to run both jobs _at the same
time_.
Okay, so I try this
Run Arking-Job-01
Run Arking-Job-01 again
Run Arking-Job-01 again, again
At least one of these jobs should wait because Maximum Concurrent
Jobs = 2 in the Storage Resource. That's _not_ true, either. Bacula
will run all three concurrently.
Okay, so I try this
Run Arking-Job-01
Run Arking-Job-01 again
Run Arking-Job-01 again, again
Run Arking-Job-01 again, again, again
_Now_ one of the jobs will wait.
So it seems to me that Bacula will pay attention to the Maximum
Concurrent Jobs in the client resource and completely ignore the
configuration item in the Storage and Job resources.
I may be misunderstanding how this all works but I really did go
through the documentation multiple, multiple times. It's pretty
frustrating because I'm constantly watching Bacula so that I can kill
duplicate Full jobs.
John Kloss.
------------------------------------------------------------------------------
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users
|