**Problem Description**
Trying to get the following configuration to work (the real one is more
complex, but this is a minimal example, although it's still quite long):
Imagine having 3 hosts, all linux.
Host Alice: Runs bacula-fd sending local data to Bob.
Host Bob: Runs bacula-fd, bacula-sd, bacula-dir, pointing to a NAS
Host Carl: Runs Bacula-SD, pointing to local zfs filesystem.
I'm trying to get the backups that are made daily from host 'Alice' to host
'Bob' to be copied so they are also stored in host 'Carl' for redundancy.
Searching, Bacula seems to have a [copy job][1] made for this.
I cannot get these to work. Instead of copying files, these copy jobs cause the
job (and everything that depends on it) to deadlock.
**Configuration**
The definition of the job, and copy job, stored on host Bob at
`Bob:/etc/bacula/jobdefs/alice.conf` looks like the following:
Job {
Name = "alice-job"
Client = "alice-fd"
FileSet = alice-fileset
Pool = Default
Full Backup Pool = alice-full-pool
Incremental Backup Pool = alice-inc-pool
Differential Backup Pool = alice-diff-pool
Enabled = Yes
Type = Backup
Schedule = WeeklyCycle
Messages = Standard
Storage = sdef-bob
RunScript {
RunsOnClient = No
Command = "/etc/bacula/runCopyJob.sh -n %n -i %i -p %p -l %l"
RunsWhen = After
}
Maximum Concurrent Jobs = 63 ## Testing various values here.
}
# Supposed to copy the Full backups originally from Alice from Bob to Carl
# Not working yet...
Job {
Name = "alice-copy-full"
Client = "bob-fd"
FileSet = alice-fileset
Type = Copy
Pool = alice-full-pool
# Copies all jobs that have not yet been copied.
Selection Type = PoolUncopiedJobs
Messages = Standard
Enabled = Yes
# Note: This is the source, not the target storage
Storage = bob-storage
Maximum Concurrent Jobs = 19 ## Trying out Pool jobs x 2 + 1
}
Job {
Name = "alice-copy-inc"
Client = "bob-fd"
FileSet = alice-fileset
Type = Copy
Pool = alice-inc-pool
Selection Type = PoolUncopiedJobs
Messages = Standard
Enabled = Yes
Storage = bob-storage
Maximum Concurrent Jobs = 63
}
Job {
Name = "alice-copy-diff"
Client = "bob-fd"
FileSet = alice-fileset
Type = Copy
Pool = alice-diff-pool
Selection Type = PoolUncopiedJobs
Messages = Standard
Enabled = Yes
Storage = bob-storage # Note: This is the source, not the target storage
Maximum Concurrent Jobs = 63
}
I use a little bash script `runCopyJob.sh` that manually calls the copy job
after the regular job completes. This does the same as running the job manually
via the console. (Suppose the job is #23 in the list, you use the commandline
to run: `bconsole`, `run` then `23`)
The `WeeklyCycle` schedule does a Full backup each month on the 1st, then
incremental/differential backups daily on weekdays.
Side note: `63 = 31 * 2 + 1`. There are 31 days in the longest month. Each copy
task spawns two jobs, and the main task is another job. Thus there are at most
63 jobs if the 'copy job' functionality is enabled for a client on the 31st if
backups run at most daily.
I've been experimenting with the Maximum Concurrent Jobs value but to no avail.
It can be put in many places, and I'm not sure which overrides which. Trial and
Error for the endless combinations possible is taking too long.
Next, there's also some Backup pools defined for Alice (a total of 6). These
are defined on Bob at `Bob:/etc/bacula/pooldefs/`
Pool {
Name = alice-full-pool
Pool Type = Backup
Recycle = yes
AutoPrune = yes
Volume Retention = 6 months
Maximum Volume Jobs = 1
Maximum Volumes = 9
Label Format = alice-full-
Next Pool = alice-full-pool-carl
}
Pool {
Name = alice-inc-pool
Pool Type = Backup
Recycle = yes
AutoPrune = yes
Volume Retention = 20 days
Maximum Volume Jobs = 6
Maximum Volumes = 7
Label Format = alice-inc-
Next Pool = alice-inc-pool-carl
}
Pool {
Name = alice-diff-pool
Pool Type = Backup
Recycle = yes
AutoPrune = yes
Volume Retention = 40 days
Maximum Volume Jobs = 1
Maximum Volumes = 10
Label Format = alice-diff-
Next Pool = alice-diff-pool-carl
}
Pool {
Name = alice-full-pool-carl
Pool Type = Backup
Recycle = yes
AutoPrune = yes
Volume Retention = 6 months
Maximum Volume Jobs = 1
Maximum Volumes = 9
Label Format = alice-full-
Storage = sdef-carl
}
Pool {
Name = alice-inc-pool-carl
Pool Type = Backup
Recycle = yes
AutoPrune = yes
Volume Retention = 20 days
Maximum Volume Jobs = 6
Maximum Volumes = 7
Label Format = alice-inc-
Storage = sdef-carl
}
Pool {
Name = alice-diff-pool-carl
Pool Type = Backup
Recycle = yes
AutoPrune = yes
Volume Retention = 40 days
Maximum Volume Jobs = 1
Maximum Volumes = 10
Label Format = alice-diff-
Storage = sdef-carl
}
The director (@ Bob) is configured via `Bob:/etc/bacula/bacula-dir.conf`like
so:
Director {
Name = bob-dir
DIRport = 9101
QueryFile = "/etc/bacula/query.sql"
WorkingDirectory = "/opt/bacula/working"
PidDirectory = "/var/run"
Maximum Concurrent Jobs = 63 ## Trying out higher values here...
Password = "dir-bob-password"
Messages = Daemon
TLS Certificate = /etc/bacula/certs/Bob.example.com.crt
TLS Key = /etc/bacula/certs/Bob.example.com.key
TLS CA Certificate File= /etc/bacula/certs/myca.crt
TLS Enable = yes
TLS Require = yes
TLS Verify Peer = yes
TLS Allowed CN = bob.example.com
}
Schedule {
Name = "WeeklyCycle"
Run = Full 1st sun at 23:05
Run = Differential 2nd-5th sun at 23:05
Run = Incremental mon-sat at 23:05
}
The Storage Demon at Bob is configured via `Bob:/etc/bacula/bacula-sd.conf`:
Storage {
Name = bob-sd
SDPort = 9103 # Director's port
WorkingDirectory = "/opt/bacula/working"
Pid Directory = "/var/run"
Maximum Concurrent Jobs = 40
SDAddress = bob.example.com
TLS Certificate = /etc/bacula/certs/bob.example.com.crt
TLS Key = /etc/bacula/certs/bob.example.com.key
TLS CA Certificate File = /etc/bacula/certs/myca.crt
TLS Enable = yes
TLS Require = yes
TLS Verify Peer = yes
TLS Allowed CN = bob.example.com, carl.example.com
}
Director {
Name = bob-dir
Password = "sd-bob-dir-bob-pw"
}
Device {
Name = bob-storage
Media Type = File
Archive Device = /FileStorage01 # Path where the NAS is mounted.
LabelMedia = yes;
Random Access = Yes;
AutomaticMount = yes;
RemovableMedia = no;
AlwaysOpen = no;
Maximum Concurrent Jobs = 20
}
The storage demon at 'carl' is configured with the same file on this host;
Storage { # definition of myself
Name = carl-sd
SDPort = 9103 # Director's port
WorkingDirectory = "/var/lib/bacula"
Pid Directory = "/run/bacula"
Plugin Directory = "/usr/lib/bacula"
Maximum Concurrent Jobs = 63
SDAddress = 0.0.0.0
TLS Enable = Yes
TLS Require = Yes
TLS CA Certificate File = /etc/bacula/certs/myca.crt
TLS Certificate = /etc/bacula/certs/carl.example.com.crt
TLS Key = /etc/bacula/certs/carl.example.com.key
TLS Verify Peer = Yes
TLS Allowed CN = bob.example.com
}
#
# List Directors who are permitted to contact Storage daemon
#
Director {
Name = bob-dir
Password = "sd-carl-dir-bob-pw"
TLS Enable = Yes
TLS Require = Yes
TLS CA Certificate File = /etc/bacula/certs/myca.crt
TLS Certificate = /etc/bacula/certs/carl.example.com.crt
TLS Key = /etc/bacula/certs/carl.example.com.key
}
Device {
Name = carl-storage
Media Type = File
Archive Device = /zfs1/external/bob/backups
LabelMedia = yes; # lets Bacula label unlabeled media
Random Access = Yes;
AutomaticMount = yes; # when device opened, read it
RemovableMedia = no;
AlwaysOpen = no;
Maximum Concurrent Jobs = 20
}
Both storage demons are in a file `/etc/bacula/storagedefs/file.conf` so the
director over at Bob can find them and orchestrate the file transfer.
Storage {
Name = sdef-bob
# Do not use "localhost" here
# Note: Use a fully qualified name here
Address = bob.example.com
SDPort = 9103
Password ="sd-bob-dir-bob-pw"
Device = bob-storage
Media Type = File
Maximum Concurrent Jobs = 63
}
Storage {
Name = sdef-carl
Address = carl.example.com
SDPort = 9103
Password = "sd-carl-dir-bob-pw"
Device = carl-storage
Media Type = File
Maximum Concurrent Jobs = 63
TLS Enable = yes
TLS Require = yes
TLS Certificate = /etc/bacula/certs/carl.example.com.crt
TLS Key = /etc/bacula/certs/carl.example.com.key
TLS CA Certificate File = /etc/bacula/certs/myca.crt
}
The original backup and the copy job are done by FileDaemons. The configuration
for Bob's file daemon is at `bob:/etc/bacula/bacula-fd.conf`:
#
# List Directors who are permitted to contact this File daemon
#
Director {
Name = bacula01-dir
Password = "fd-bob-dir-bob-pw"
Address = bob.example.com
TLS Certificate = /etc/bacula/certs/bob.example.com.crt
TLS Key = /etc/bacula/certs/bob.example.com.key
TLS CA Certificate File = /etc/bacula/certs/myca.crt
TLS Enable = yes
TLS Require = yes
# TLS Allowed CN = “bob.example.com”
}
#
# Restricted Director, used by tray-monitor to get the
# status of the file daemon
#
Director {
Name = bacula01-mon
Password = ""
Monitor = yes
}
#
# "Global" File daemon configuration specifications
#
FileDaemon { # this is me
Name = bob-fd
FDport = 9102 # where we listen for the director
WorkingDirectory = /opt/bacula/working
Pid Directory = /var/run
Maximum Concurrent Jobs = 20
FDAddress = bob.example.com
TLS Enable = yes
TLS Require = yes
TLS CA Certificate File = /etc/bacula/certs/myca.crt
TLS Certificate = /etc/bacula/certs/bob.example.com.crt
TLS Key = /etc/bacula/certs/bob.example.com.key
}
And Alice's FileDaemon is configured at `alice:/etc/bacula/bacula-fd.conf`
#
# List Directors who are permitted to contact this File daemon
#
Director {
Name = bob-dir
Password = "alice-fd-bob-dir-pw" # same pass as on bob
/etc/bacula/clientdefs/HOSTNAME-fd.conf
TLS Enable = yes
TLS Require = yes
TLS Verify Peer = yes
TLS CA Certificate File = /etc/bacula/certs/myca.crt
TLS Certificate = /etc/bacula/certs/alice.example.com.crt
TLS Key = /etc/bacula/certs/alice.example.com.key
}
#
# Restricted Director, used by tray-monitor to get the
# status of the file daemon
#
Director {
Name = bacula01-mon
Password = "RandomSecretDataHere"
Monitor = yes
}
#
# "Global" File daemon configuration specifications
#
FileDaemon { # this is me
Name = "alice-fd"
FDport = 9102 # where we listen for the director
WorkingDirectory = /opt/bacula/working
Pid Directory = /var/run
Maximum Concurrent Jobs = 20
# Plugin Directory = /usr/lib
FDAddress = alice.example.com
TLS Enable = yes
TLS Require = yes
TLS CA Certificate File = /etc/bacula/certs/myca.crt
TLS Certificate = /etc/bacula/certs/bob.example.com.crt
TLS Key = /etc/bacula/certs/bob.example.com.key
}
# Send all messages except skipped files back to Director
Messages {
Name = Standard
director = bob-dir = all, !skipped, !restored
}
The director knows about Alice's File Daemon because of what's in
`bob:/etc/bacula/clientDefs/alice.example.com-fd.conf`:
Client {
Name = "alice-fd"
Address = alice.example.com
FDPort = 9102
Catalog = MyCatalog
Password = "alice-fd-bob-dir-pw"
File Retention = 30 days
Job Retention = 6 months
AutoPrune = yes
TLS Enable = yes
TLS Require = yes
TLS Certificate = /etc/bacula/certs/alice.example.com.crt
TLS Key = /etc/bacula/certs/alice.example.com.key
TLS CA Certificate File = /etc/bacula/certs/myca.crt
}
Similarly for its own file daemon at
`/etc/bacula/clientdefs/bob.example.com-fd.conf`:
Client {
Name = "bob-fd"
Address = bob.example.com
FDPort = 9102
Catalog = MyCatalog
Password = "bob-fd-bob-dir-pw"
File Retention = 14 days
Job Retention = 1 months
AutoPrune = yes
TLS Enable = yes
TLS Require = yes
TLS Certificate = /etc/bacula/certs/bob.example.com.crt
TLS Key = /etc/bacula/certs/bob.example.com.key
TLS CA Certificate File = /etc/bacula/certs/myca.crt
}
**Problem Symptoms**
When running this type of copy job, the job never completes
Running , within `bconsole> status dir`, produces output akin to:
JobId Type Level Files Bytes Name Status
======================================================================
63642 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63643 Copy Full 0 0 alice-copy-inc is running
63644 Back Full 0 0 alice-job is running
63645 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63646 Back Full 0 0 alice-job is waiting on Storage
"sdef-carl"
63647 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63648 Back Full 0 0 alice-job is waiting on Storage
"sdef-carl"
63649 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63650 Back Full 0 0 alice-job is waiting on Storage
"sdef-carl"
63651 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63652 Back Full 0 0 alice-job is waiting on Storage
"sdef-carl"
63653 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63654 Back Full 0 0 alice-job is waiting on Storage
"sdef-carl"
63655 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63656 Back Full 0 0 alice-job is waiting on Storage
"sdef-carl"
63657 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63658 Back Full 0 0 alice-job is waiting on Storage
"sdef-carl"
63659 Copy Full 0 0 alice-copy-inc is waiting on Storage
"sdef-bob"
63660 Back Full 0 0 alice-job is waiting on Storage
"sdef-carl"
The number of lines depends on the day of the month, two for every incremental
backup plus 1 `2n+1` created, except on the first, where the last backup is a
Full backup; in that case, it should give 7 lines. Once the first copy job has
run and all the data is transferred, the next job should only ever produce 3
lines.
Noticing how everyone is waiting on these storage, running `status storage` for
say carl's storage outputs:
carl-sd Version: 9.6.7 (10 December 2020) x86_64-pc-linux-gnu debian
bullseye/sid
Daemon started 21-Jun-22 12:08. Jobs: run=0, running=2.
Heap: heap=401,408 smbytes=722,480 max_bytes=722,674 bufs=674 max_bufs=676
Sizes: boffset_t=8 size_t=8 int32_t=4 int64_t=8 mode=0,0 newbsr=0
Res: ndevices=2 nautochgr=0
Running Jobs:
Writing: Full Backup job alice-job JobId=63555 Volume=""
pool="alice-Inc-Pool-carl" device="carl-storage"
(/zfs1/external/bob/backups)
spooling=0 despooling=0 despool_wait=0
Files=0 Bytes=0 AveBytes/sec=0 LastBytes/sec=0
FDReadSeqNo=5 in_msg=5 out_msg=4 fd=6
Writing: Full Backup job alice-job JobId=63644 Volume=""
pool="alice-Inc-Pool-carl" device="carl-storage"
(/zfs1/external/bob/backups)
spooling=0 despooling=0 despool_wait=0
Files=0 Bytes=0 AveBytes/sec=0 LastBytes/sec=0
FDReadSeqNo=5 in_msg=5 out_msg=4 fd=15
====
Jobs waiting to reserve a drive:
====
Terminated Jobs:
JobId Level Files Bytes Status Finished Name
===================================================================
63350 Full 0 0 Created 20-Jun-22 14:21 alice-job
63352 Full 0 0 Created 20-Jun-22 14:21 alice-job
63354 Full 0 0 Created 20-Jun-22 14:21 alice-job
63355 Full 0 0 Created 20-Jun-22 14:21 alice-job
63358 Full 0 0 Cancel 20-Jun-22 14:37 alice-job
63360 Full 0 0 Cancel 20-Jun-22 14:55 alice-job
63384 Full 0 0 Cancel 20-Jun-22 15:27 alice-job
63459 Full 0 0 Cancel 21-Jun-22 09:57 alice-job
63483 Full 0 0 Cancel 21-Jun-22 09:58 alice-job
63396 Full 0 0 Cancel 21-Jun-22 12:08 alice-job
====
Device status:
Device File: "carl-storage" (/zfs1/external/bob/backups) is not open.
Device is BLOCKED waiting to create a volume for:
Pool: alice-Inc-Pool-carl
Media type: File
Available Space=28.15 TB
==
====
Used Volume status:
====
====
No, manually creating a volume (which is supposed to automatically create
itself) does not make the problem go away.
It's possible to stop the director and run it in debug mode by redirecting its
output.
By executing `bacula-dir -d 201 -f > /var/log/bacula/run_debug_2.log 2>&1 &`
it's possible to view some more information alongside the C code.
Except... that results in hundreds of megabytes of logs which won't fit on this
page. I'd like to know more what to look for..
One thing I found earlier was that at some point the code fails due to a
variable `rncj` not being high enough. I looked at the C file in question, and
it contains the following snippet;
bool inc_read_store(JCR *jcr)
{
P(rstore_mutex);
int num = jcr->rstore->getNumConcurrentJobs();
int numread = jcr->rstore->getNumConcurrentReadJobs();
int maxread = jcr->rstore->MaxConcurrentReadJobs;
if (num < jcr->rstore->MaxConcurrentJobs &&
(jcr->getJobType() == JT_RESTORE ||
numread == 0 ||
maxread == 0 || /* No limit set */
numread < maxread)) /* Below the limit */
{
num++;
numread++;
jcr->rstore->setNumConcurrentReadJobs(numread);
jcr->rstore->setNumConcurrentJobs(num);
Dmsg1(200, "Inc rncj=%d\n", num);
V(rstore_mutex);
return true;
}
V(rstore_mutex);
return false;
}
The line `Inc rncj` would be missing, while the line before (in the job
scheduling process) would be printed. Therefore the Job isn't allowed to
proceed because it couldn't read. So I tried increasing the MaxConcurrentJobs
variable, but it'll just keep cycling and never actually start the job, and
rncj seems to increase without bound (there isn't even 63 total jobs queued).
[1]: https://www.bacula.org/11.0.x-manuals/en/main/Migration_Copy.html
_______________________________________________
Bacula-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bacula-users