NDMP Backups with Bareos
NDMP Basics
NDMP
is the abbreviation for Network Data Management Protocol.
is a protocol that transports data between Network Attached Storages (NAS) and backup devices.
is widely used by storage product vendors and OS vendors like NetApp, Isilon, EMC, Oracle.
information is available at https://www.snia.org/ndmp.
version is currently (2016) NDMP Version 4.
uses TCP/IP and XDR (External Data Representation) for the communication.
The Bareos NDMP implementation is based on the NDMJOB NDMP reference implementation of Traakan, Inc., Los Altos, CA which has a BSD style license (2 clause one) with some enhancements.
In NDMP, there are different components (called “agents”) involved in doing backups. The most important agents are:
- Data Management Agent (DMA)
is the part that controls the NDMP backup or recover operation.
- Data Agent
(or Primary Storage System) is the part that reads the data from the Filesystem during Backup and writes data to the Filesystem during recover.
- Tape Agent
(or Secondary Storage System) is the part that writes NDMP blocks to the Tape during backup and reads them during recover.
- Robot Agent
is the part that controls the media changer. It loads/unloads tapes and gets the inventory of the Changer. The use of a robot agent is optional. It is possible to run backups on a single tape drive.
All elements involved talk to each other via the NDMP protocol which is usually transported via TCP/IP port 10000.
The Data Management Agent is part of the Backup Application.
The Data Agent is part of the (NAS)-System being backed up and recovered.
The Tape Agent and Robot Agent can
run on the system being backed up
run as part of the backup application
or even run independently on a third system
This flexibility leads to different topologies how NDMP backups can be done.
NDMP Topologies
When looking at the different topologies, the location of the Robot Agent is not specially considered, as the data needed to control the robot is minimal compared to the backup data.
So the parts considered are
the Data Management Agent controlling the operation
the Data Agent as source of the backup data and
the Tape Agent as destination of the backup data
The Data Management Agent always controls both Data Agent and Tape Agent over the Network via NDMP.
The Tape Agent can either
run on a separate system
run on the same system
as the Data Agent.
NDMP 3-way Backup: Data Agent and Tape Agent running on different systems
--+--------------- NETWORK ----+-------------------+----
| -->---->---->-->-->-->\ | //==>==>==>==>\\ |
| / | | || (2) || |
| | | | || || |
/----------\ /----------\ /----------\
| | | | | |
| DMA | DISK====>| DATA | | Tape |====>TAPE DRIVE
| | (1) | Agent | | Agent | (3)
\----------/ \----------/ \----------/
The data path consists of three ways
From Disk to Data Agent (1)
From Data Agent over the Network to the Tape Agent (2)
From Tape Agent to the Tape (3)
and is called NDMP 3-way Backup.
NDMP 2-way Backup: Data Agent and Tape Agent running on the same system
--+--------------- NETWORK ----+---------
| -->---->---->-->-->-->\ |
| / | |
| | | |
/----------\ /----------\
| | | Data |
| DMA | DISK====>| Agent |
| | (1) | |
\----------/ | Tape | (2)
| Agent |====>TAPE DRIVE
\----------/
Data Agent and Tape Agent are both part of the same process on the system, so the data path consists of two ways:
From Disk to Data Agent (1)
From Tape Agent to the Tape (2)
and is called NDMP 2-way Backup, also sometimes referred as NDMP local backup.
Properties of the different NDMP Backup topologies
NDMP 3-way backup:
The data can be send to a different location over the network
No need to attach a tape drive to the NAS system.
The backup speed is usually slower than with 2-way backup as the data is being sent over the network and processed multiple times.
NDMP 2-way backup:
The data is directly copied from the NAS system to the Tape
Usually the fastest way to do a NDMP backup
tape drives need to be attached to the NAS System
NDMP Backup in Bareos
Bareos offers two types of NDMP integration:
NDMP_NATIVE NDMP_BAREOS
In both cases,
Bareos Director acts as Data Management Agent.
The Data Agent is part of the storage system and must be provided by the storage vendor.
The main difference is which Tape Agent is used.
When using NDMP_BAREOS, the Bareos Storage Daemon acts as Tape Agent.
When using NDMP_NATIVE, the Tape Agent must be provided by some other systems. Some storage vendors provide it with there storages, or offer it as an option, e.g. Isilon with their “Isilon Backup Accelerator”.
Data Management Agent |
Bareos Director |
Bareos Director |
Tape Agent |
Bareos Storage Daemon |
external |
Requires external Tape Agent |
✓ |
|
Backup to tape (and VTL) |
✓ |
✓ |
Backup to other |
✓ |
|
2-way backup |
✓ |
|
3-way backup |
✓ |
untested |
Full Backups |
✓ |
✓ |
Differential Backups |
✓ |
✓ |
Incremental Backups |
✓ (8) |
✓ (8) |
Single File Restore |
✓ |
✓ |
DAR |
✓ |
|
DDAR |
✓ |
|
✓ |
NDMP_BAREOS
Bareos implements the Data Management Agent inside of the Bareos Director and a Tape Agent in the Bareos Storage Daemon.
The Tape Agent in the Bareos Storage Daemon emulates a NDMP tape drive that has an infinite tape. Because of the infinite tape, no Robot Agent is required and therefore not implemented. The blocks being written to the NDMP tape are wrapped into a normal Bareos backup stream and then stored into the volumes managed by Bareos.
There is always a pair of storage resource definitions:
a conventional Bareos storage resource and
a NDMP storage resource
These two are linked together. Data that is received by the Tape Agent inside of the Bareos Storage Daemon is then stored as Bareos backup stream inside of the paired conventional Bareos storage resource.
On restore, the data is read by the conventional resource, and then recovered as NDMP stream from the NDMP resource.
Note
Copying and migrating a NDMP_BAREOS job is not an NDMP operation. NDMP jobs are copied and migrated just like every other Bareos job.
Example Setup for NDMP_BAREOS backup
This example starts from a clean default Bareos installation.
Enable NDMP on your storage appliance
The storage appliance needs to be configured to allow NDMP connections. Therefore usually the NDMP service needs to be enabled and configured with a username and password.
Bareos Director: Configure NDMP Client Resource
Add a Client resource to the Bareos Director configuration and configure it to access your NDMP storage system (Primary Storage System/Data Agent).
Protocol (Dir->Client)
must be either NDMPv2, NDMPv3 or NDMPv4.Port (Dir->Client)
is set to the NDMP Port (usually 10000).Username (Dir->Client)
andPassword (Dir->Client)
are used for the authentication against the NDMP Storage System.Auth Type (Dir->Client)
is either Cleartext or MD5. NDMP supports both.
In our example we connect to a Isilon storage appliance emulator:
Verify, that you can access your Primary Storage System via Bareos:
*status client=ndmp-client
Data Agent isilon.example.com NDMPv4
Host info
hostname isilonsim-1
os_type Isilon OneFS
os_vers v7.1.1.5
hostid 005056ad8483ba43cc55a711cd384506e3c1
Server info
vendor Isilon
product Isilon NDMP
revision 2.2
auths (2) NDMP4_AUTH_TEXT NDMP4_AUTH_MD5
Connection types
addr_types (2) NDMP4_ADDR_TCP NDMP4_ADDR_LOCAL
Backup type info of tar format
attrs 0x7fe
set FILESYSTEM=/ifs
set FILES=
set EXCLUDE=
set PER_DIRECTORY_MATCHING=N
set HIST=f
set DIRECT=N
set LEVEL=0
set UPDATE=Y
set RECURSIVE=Y
set ENCODING=UTF-8
set ENFORCE_UNIQUE_NODE=N
set PATHNAME_SEPARATOR=/
set DMP_NAME=
set BASE_DATE=0
set NDMP_UNICODE_FH=N
Backup type info of dump format
attrs 0x7fe
set FILESYSTEM=/ifs
set FILES=
set EXCLUDE=
set PER_DIRECTORY_MATCHING=N
set HIST=f
set DIRECT=N
set LEVEL=0
set UPDATE=Y
set RECURSIVE=Y
set ENCODING=UTF-8
set ENFORCE_UNIQUE_NODE=N
set PATHNAME_SEPARATOR=/
set DMP_NAME=
set BASE_DATE=0
set NDMP_UNICODE_FH=N
File system /ifs
physdev OneFS
unsupported 0x0
type NFS
status
space 12182519808 total, 686768128 used, 11495751680 avail
inodes 17664000 total, 16997501 used
set MNTOPTS=
set MNTTIME=00:00:00 00:00:00
This output shows that the access to the storage appliance was successful.
Bareos Storage Daemon: Configure NDMP
Enabling NDMP
To enable the NDMP Tape Agent inside of the Bareos Storage Daemon, set NDMP Enable (Sd->Storage)
=yes:
Add a NDMP resource
Additionally, we need to define the access credentials for our NDMP TAPE AGENT (Secondary Storage) inside of this Storage Daemon.
These are configured by adding a NDMP resource to Bareos Storage Daemon configuration:
Username and Password can be anything, but they will have to match the settings in the Bareos Director NDMP Storage resource we configure next.
Now restart the Bareos Storage Daemon. If everything is correct, the Bareos Storage Daemon starts and listens now on the usual port (9103) and additionally on port 10000 (ndmp).
root@host:~# netstat -lntp | grep bareos-sd
tcp 0 0 0.0.0.0:9103 0.0.0.0:* LISTEN 10661/bareos-sd
tcp 0 0 0.0.0.0:10000 0.0.0.0:* LISTEN 10661/bareos-sd
Bareos Director: Configure a Paired Storage
For NDMP Backups, we always need two storages that are paired together. The default configuration already has a Storage File (Dir->Storage)
defined:
We now add a paired storage to the already existing File (Dir->Storage)
storage:
The settings of Username and Password need to match the settings of the Bareos Storage Daemon’s NDMP resource we added before. The address will be used by the storage appliance’s NDMP Daemon to connect to the Bareos Storage Daemon via NDMP. Make sure that the Storage appliance can resolve the name or use an IP address.
Now save the director resource and restart the Bareos Director. Verify that the configuration is correct:
*status storage=NDMPFile
Connecting to Storage daemon File at bareos:9103
bareos-sd Version: 15.2.2 (16 November 2015) x86_64-redhat-linux-gnu redhat Red Hat Enterprise Linux Server release 7.0 (Maipo)
Daemon started 14-Jan-16 10:10. Jobs: run=0, running=0.
Heap: heap=135,168 smbytes=34,085 max_bytes=91,589 bufs=75 max_bufs=77
Sizes: boffset_t=8 size_t=8 int32_t=4 int64_t=8 mode=0 bwlimit=0kB/s
Running Jobs:
No Jobs running.
====
Jobs waiting to reserve a drive:
====
Terminated Jobs:
====
Device status:
Device "FileStorage" (/var/lib/bareos/storage) is not open.
==
====
Used Volume status:
====
====
*
The output looks the same, as if a status storage=File would have been called.
Bareos Director: Configure NDMP Fileset
To specify what files and directories from the storage appliance should be backed up, a Fileset needs to be specified. In our example, we decided to backup /ifs/home
directory.
The specified directory needs to be a filesystem or a subdirectory of a filesystem which can be accessed by NDMP. Which filesystems are available is showed in the status client output of the NDMP client.
Additionally, NDMP can be configured via NDMP environment variables. These can be specified in the Options Block of the Fileset with the Meta keyword. Which variables are available is partly depending on the NDMP implementation of the Storage Appliance.
Warning
Normally (Protocol (Dir->Client)
=Native) Filesets get handled by the bareosFd. When connecting directly to a NDMP Clients (Protocol (Dir->Client)
=NDMP*), no Bareos File Daemon is involved and therefore most Fileset options can’t be used. Instead, parameters are handled via Options - Meta from Include (Dir->Fileset)
.
Warning
Avoid using multiple Include (Dir->Fileset)
File directives.
The Bareos Director would try to handle them by running multiple NDMP jobs in a single Bareos job.
Even if this is working fine during backup, restore jobs will cause trouble.
Some NDMP environment variables are set automatically by the DMA in the Bareos Director. The following environment variables are currently set automatically:
- FILESYSTEM
is set to the
Include (Dir->Fileset)
File directive.- HIST
= Y Specifies the file history format:
- Y
Specifies the default file history format determined by your NDMP backup settings.
- N
Disables file history. Without file hostory, single file restore is not possible with Bareos.
Some NDMP environments (eg. Isilon OneFS) allow additional parameter:
- F
Specifies path-based file history. This is the most efficient with Bareos.
- D
Specifies directory or node file history.
- LEVEL
is set accordingly to NDMP Backup Level.
PREFIX
- TYPE
is set accordingly to BUTYPE. Default “DUMP”.
- UPDATE
= Y
Example NDMP Fileset to backup a subset of a NDMP filesystem
The following fileset is intended to backup all files and directories matching /ifs/home/users/a*
. It has been tested against Isilon OneFS 7.2.0.1. See Isilon OneFS Administration Guide, section quote{NDMP environment variables} for details about the supported NDMP environment variables. Excludes are not used in this example.
Bareos Director: Configure NDMP Jobs
To do NDMP backups and restores, some special settings need to be configured. We define special Backup and Restore jobs for NDMP.
Backup Format (Dir->Job)
=dump is used in our example. Other Backup Formats have other advantages/disadvantages.
Run NDMP Backup
Now we are ready to do our first NDMP backup:
*run job=ndmp-backup-job
Using Catalog "MyCatalog"
Run Backup job
JobName: ndmp-backup-job
Level: Incremental
Client: ndmp-client
Format: dump
FileSet: NDMP Fileset
Pool: Full (From Job resource)
Storage: NDMPFile (From Job resource)
When: 2016-01-14 10:48:04
Priority: 10
OK to run? (yes/mod/no): yes
Job queued. JobId=1
*wait jobid=1
JobId=1
JobStatus=OK (T)
*list joblog jobid=1
2016-01-14 10:57:53 bareos-dir JobId 1: Start NDMP Backup JobId 1, Job=NDMPJob.2016-01-14_10.57.51_04
2016-01-14 10:57:53 bareos-dir JobId 1: Created new Volume "Full-0001" in catalog.
2016-01-14 10:57:53 bareos-dir JobId 1: Using Device "FileStorage" to write.
2016-01-14 10:57:53 bareos-dir JobId 1: Opening tape drive LPDA-DEJC-ENJL-AHAI-JCBD-LICP-LKHL-IEDK@/ifs/home%0 read/write
2016-01-14 10:57:53 bareos-sd JobId 1: Labeled new Volume "Full-0001" on device "FileStorage" (/var/lib/bareos/storage).
2016-01-14 10:57:53 bareos-sd JobId 1: Wrote label to prelabeled Volume "Full-0001" on device "FileStorage" (/var/lib/bareos/storage)
2016-01-14 10:57:53 bareos-dir JobId 1: Commanding tape drive to rewind
2016-01-14 10:57:53 bareos-dir JobId 1: Waiting for operation to start
2016-01-14 10:57:53 bareos-dir JobId 1: Async request NDMP4_LOG_MESSAGE
2016-01-14 10:57:53 bareos-dir JobId 1: Operation started
2016-01-14 10:57:53 bareos-dir JobId 1: Monitoring backup
2016-01-14 10:57:53 bareos-dir JobId 1: LOG_MESSAGE: 'Filetransfer: Transferred 5632 bytes in 0.087 seconds throughput of 63.133 KB/s'
2016-01-14 10:57:53 bareos-dir JobId 1: LOG_MESSAGE: 'Filetransfer: Transferred 5632 total bytes '
2016-01-14 10:57:53 bareos-dir JobId 1: LOG_MESSAGE: 'CPU user=0.016416 sys=0.029437 ft=0.077296 cdb=0.000000'
2016-01-14 10:57:53 bareos-dir JobId 1: LOG_MESSAGE: 'maxrss=14576 in=13 out=22 vol=155 inv=72'
2016-01-14 10:57:53 bareos-dir JobId 1: LOG_MESSAGE: '
Objects (scanned/included):
----------------------------
Regular Files: (1/1)
Sparse Files: (0/0)
Stub Files: (0/0)
Directories: (2/2)
ADS Entries: (0/0)
ADS Containers: (0/0)
Soft Links: (0/0)
Hard Links: (0/0)
Block Device: (0/0)
Char Device: (0/0)
FIFO: (0/0)
Socket: (0/0)
Whiteout: (0/0)
Unknown: (0/0)'
2016-01-14 10:57:53 bareos-dir JobId 1: LOG_MESSAGE: '
Dir Depth (count)
----------------------------
Total Dirs: 2
Max Depth: 1
File Size (count)
----------------------------
== 0 0
<= 8k 1
<= 64k 0
<= 1M 0
<= 20M 0
<= 100M 0
<= 1G 0
> 1G 0
-------------------------
Total Files: 1
Total Bytes: 643
Max Size: 643
Mean Size: 643'
2016-01-14 10:57:53 bareos-dir JobId 1: LOG_MESSAGE: '
File History
----------------------------
Num FH_HIST_FILE messages: 0
Num FH_HIST_DIR messages: 6
Num FH_HIST_NODE messages: 3'
2016-01-14 10:57:54 bareos-dir JobId 1: Async request NDMP4_NOTIFY_MOVER_HALTED
2016-01-14 10:57:54 bareos-dir JobId 1: DATA: bytes 2053KB MOVER: written 2079KB record 33
2016-01-14 10:57:54 bareos-dir JobId 1: Operation done, cleaning up
2016-01-14 10:57:54 bareos-dir JobId 1: Waiting for operation to halt
2016-01-14 10:57:54 bareos-dir JobId 1: Commanding tape drive to NDMP9_MTIO_EOF 2 times
2016-01-14 10:57:54 bareos-dir JobId 1: Commanding tape drive to rewind
2016-01-14 10:57:54 bareos-dir JobId 1: Closing tape drive LPDA-DEJC-ENJL-AHAI-JCBD-LICP-LKHL-IEDK@/ifs/home%0
2016-01-14 10:57:54 bareos-dir JobId 1: Operation halted, stopping
2016-01-14 10:57:54 bareos-dir JobId 1: Operation ended OKAY
2016-01-14 10:57:54 bareos-sd JobId 1: Elapsed time=00:00:01, Transfer rate=2.128 M Bytes/second
2016-01-14 10:57:54 bareos-dir JobId 1: Bareos bareos-dir 15.2.2 (16Nov15):
Build OS: x86_64-redhat-linux-gnu redhat Red Hat Enterprise Linux Server release 7.0 (Maipo)
JobId: 1
Job: ndmp-backup-job.2016-01-14_10.57.51_04
Backup Level: Full
Client: "ndmp-client"
FileSet: "NDMP Fileset" 2016-01-14 10:57:51
Pool: "Full" (From Job resource)
Catalog: "MyCatalog" (From Client resource)
Storage: "NDMPFile" (From Job resource)
Scheduled time: 14-Jan-2016 10:57:51
Start time: 14-Jan-2016 10:57:53
End time: 14-Jan-2016 10:57:54
Elapsed time: 1 sec
Priority: 10
NDMP Files Written: 3
SD Files Written: 1
NDMP Bytes Written: 2,102,784 (2.102 MB)
SD Bytes Written: 2,128,987 (2.128 MB)
Rate: 2102.8 KB/s
Volume name(s): Full-0001
Volume Session Id: 4
Volume Session Time: 1452764858
Last Volume Bytes: 2,131,177 (2.131 MB)
Termination: Backup OK
We have successfully created our first NDMP backup.
Let us have a look what files are in our backup:
*list files jobid=1
/@NDMP/ifs/home%0
/ifs/home/
/ifs/home/admin/
/ifs/home/admin/.zshrc
The real backup data is stored in the file /@NDMP/ifs/home%0
(we will refer to it as “NDMP main backup file” or “main backup file” later on). One NDMP main backup file is created for every directory specified in the used Fileset. The other files show the file history and are hardlinks to the backup file.
Run NDMP Restore
Now that we have a NDMP backup, we of course also want to restore some data from the backup. If the backup we just did saved the Filehistory, we are able to select single files for restore. Otherwise, we will only be able to restore the whole backup.
Full Restore
Either select all files or the main backup file (/@NDMP/ifs/home%0
). If file history is not included in the backup job, than only the main backup file is available.
Restore files to original path
*restore jobid=1
You have selected the following JobId: 1
Building directory tree for JobId(s) 1 ...
2 files inserted into the tree.
You are now entering file selection mode where you add (mark) and
remove (unmark) files to be restored. No files are initially added, unless
you used the "all" keyword on the command line.
Enter "done" to leave this mode.
cwd is: /
$ mark /ifs/home/admin/.zshrc
$ done
Bootstrap records written to /var/lib/bareos/bareos-dir.restore.1.bsr
The job will require the following
Volume(s) Storage(s) SD Device(s)
===========================================================================
Full-0001 File FileStorage
Volumes marked with "*" are online.
1 file selected to be restored.
The defined Restore Job resources are:
1: RestoreFiles
2: ndmp-restore-job
Select Restore Job (1-2): 2
Defined Clients:
1: bareos-fd
2: ndmp-client
Select the Client (1-2): 2
Run Restore job
JobName: ndmp-backup-job
Bootstrap: /var/lib/bareos/bareos-dir.restore.1.bsr
Where: /
Replace: Always
FileSet: NDMP Fileset
Backup Client: ndmp-client
Restore Client: ndmp-client
Format: dump
Storage: File
When: 2016-01-14 11:04:46
Catalog: MyCatalog
Priority: 10
Plugin Options: *None*
OK to run? (yes/mod/no): yes
Job queued. JobId=2
*wait jobid=2
JobId=2
JobStatus=OK (T)
*list joblog jobid=2
14-Jan 11:04 bareos-dir JobId 2: Start Restore Job ndmp-backup-job.2016-01-14_11.04.53_05
14-Jan 11:04 bareos-dir JobId 2: Using Device "FileStorage" to read.
14-Jan 11:04 bareos-dir JobId 2: Opening tape drive KKAE-IMLO-NHJD-GOCO-GJCO-GEHB-BODL-ADNG@/ifs/home read-only
14-Jan 11:04 bareos-dir JobId 2: Commanding tape drive to rewind
14-Jan 11:04 bareos-dir JobId 2: Waiting for operation to start
14-Jan 11:04 bareos-sd JobId 2: Ready to read from volume "Full-0001" on device "FileStorage" (/var/lib/bareos/storage).
14-Jan 11:04 bareos-sd JobId 2: Forward spacing Volume "Full-0001" to file:block 0:194.
14-Jan 11:04 bareos-dir JobId 2: Async request NDMP4_LOG_MESSAGE
14-Jan 11:04 bareos-dir JobId 2: Operation started
14-Jan 11:04 bareos-dir JobId 2: Monitoring recover
14-Jan 11:04 bareos-dir JobId 2: DATA: bytes 0KB MOVER: read 0KB record 0
14-Jan 11:04 bareos-dir JobId 2: LOG_MESSAGE: 'Filetransfer: Transferred 1048576 bytes in 0.135 seconds throughput of 7557.139 KB/s'
14-Jan 11:04 bareos-dir JobId 2: OK: /admin/.zshrc
14-Jan 11:04 bareos-dir JobId 2: LOG_MESSAGE: '
Objects:
----------------------------
Regular Files: (1)
Stub Files: (0)
Directories: (0)
ADS Entries: (0)
Soft Links: (0)
Hard Links: (0)
Block Device: (0)
Char Device: (0)
FIFO: (0)
Socket: (0)
Unknown: (0)'
14-Jan 11:04 bareos-dir JobId 2: LOG_MESSAGE: '
File Size (count)
----------------------------
== 0 0
<= 8k 1
<= 64k 0
<= 1M 0
<= 20M 0
<= 100M 0
<= 1G 0
> 1G 0
-------------------------
Total Files: 1
Total Bytes: 643
Max Size: 643
Mean Size: 643'
14-Jan 11:04 bareos-dir JobId 2: Async request NDMP4_NOTIFY_MOVER_PAUSED
14-Jan 11:04 bareos-dir JobId 2: DATA: bytes 1024KB MOVER: read 2079KB record 32
14-Jan 11:04 bareos-dir JobId 2: Mover paused, reason=NDMP9_MOVER_PAUSE_EOF
14-Jan 11:04 bareos-dir JobId 2: End of tapes
14-Jan 11:04 bareos-dir JobId 2: DATA: bytes 1024KB MOVER: read 2079KB record 32
14-Jan 11:04 bareos-dir JobId 2: Operation done, cleaning up
14-Jan 11:04 bareos-dir JobId 2: Waiting for operation to halt
14-Jan 11:04 bareos-dir JobId 2: Commanding tape drive to rewind
14-Jan 11:04 bareos-dir JobId 2: Closing tape drive KKAE-IMLO-NHJD-GOCO-GJCO-GEHB-BODL-ADNG@/ifs/home
14-Jan 11:04 bareos-dir JobId 2: Operation halted, stopping
14-Jan 11:04 bareos-dir JobId 2: Operation ended OKAY
14-Jan 11:04 bareos-dir JobId 2: LOG_FILE messages: 1 OK, 0 ERROR, total 1 of 1
14-Jan 11:04 bareos-dir JobId 2: Bareos bareos-dir 15.2.2 (16Nov15):
Build OS: x86_64-redhat-linux-gnu redhat Red Hat Enterprise Linux Server release 7.0 (Maipo)
JobId: 2
Job: ndmp-backup-job.2016-01-14_11.04.53_05
Restore Client: ndmp-client
Start time: 14-Jan-2016 11:04:55
End time: 14-Jan-2016 11:04:57
Elapsed time: 2 secs
Files Expected: 1
Files Restored: 1
Bytes Restored: 1,048,576
Rate: 524.3 KB/s
SD termination status: OK
Termination: Restore OK
Restore files to different path
The restore location is determined by the Where (Dir->Job)
setting of the restore job. In NDMP, this parameter works in a special manner, the prefix can be either “relative” to the filesystem or “absolute”. If a prefix is set in form of a directory (like /bareos-restores
), it will be a relative prefix and will be added between the filesystem and the filename. This is needed to make sure that the
data is restored in a different directory, but into the same filesystem. If the prefix is set with a leading caret (^), it will be an absolute prefix and will be put at the front of the restore path. This is needed if the restored data should be stored into a different filesystem.
Example:
original file name |
where |
restored file |
---|---|---|
|
|
|
|
|
|
NDMP Copy Jobs
To be able to do copy jobs, we need to have a second storage resource where we can copy the data to. Depending on your requirements, this resource can be added to the existing Bareos Storage Daemon (e.g. autochanger-0 (Sd->Storage)
for tape based backups) or to an additional Bareos Storage Daemon.
We set up an additional Bareos Storage Daemon on a host named bareos-sd2.example.com with the default FileStorage (Sd->Storage)
device.
When this is done, add a second storage resource File2 (Dir->Storage)
to the bareos-dir.conf
:
Copy Jobs copy data from one pool to another (see Migration and Copy). So we need to define a pool where the copies will be written to:
Add a Pool that the copies will run to:
Then we need to define the just defined pool as the Next Pool (Dir->Pool)
of the pool that actually holds the data to be copied.
In our case this is the Full (Dir->Pool)
Pool:
Finally, we need to define a Copy Job that will select the jobs that are in the Full (Dir->Pool)
pool and copy them over to the Copy (Dir->Pool)
pool reading the data via the File (Dir->Storage)
Storage and writing the data via the File2 (Dir->Storage)
Storage:
After restarting the director and storage daemon, we can run the Copy job:
*run job=NDMPCopy
Run Copy job
JobName: NDMPCopy
Bootstrap: *None*
Pool: Full (From Job resource)
NextPool: Copy (From unknown source)
Write Storage: File2 (From Storage from Run NextPool override)
JobId: *None*
When: 2016-01-21 09:19:49
Catalog: MyCatalog
Priority: 10
OK to run? (yes/mod/no): yes
Job queued. JobId=74
*wait jobid=74
JobId=74
JobStatus=OK (T)
*list joblog jobid=74
21-Jan 09:19 bareos-dir JobId 74: The following 1 JobId was chosen to be copied: 73
21-Jan 09:19 bareos-dir JobId 74: Automatically selected Catalog: MyCatalog
21-Jan 09:19 bareos-dir JobId 74: Using Catalog "MyCatalog"
21-Jan 09:19 bareos-dir JobId 75: Copying using JobId=73 Job=NDMPJob.2016-01-21_09.18.50_49
21-Jan 09:19 bareos-dir JobId 75: Bootstrap records written to /var/lib/bareos/bareos-dir.restore.20.bsr
21-Jan 09:19 bareos-dir JobId 74: Job queued. JobId=75
21-Jan 09:19 bareos-dir JobId 74: Copying JobId 75 started.
21-Jan 09:19 bareos-dir JobId 74: Bareos bareos-dir 15.2.2 (16Nov15):
Build OS: x86_64-redhat-linux-gnu redhat Red Hat Enterprise Linux Server release 7.0 (Maipo)
Current JobId: 74
Current Job: NDMPCopy.2016-01-21_09.19.50_50
Catalog: "MyCatalog" (From Default catalog)
Start time: 21-Jan-2016 09:19:52
End time: 21-Jan-2016 09:19:52
Elapsed time: 0 secs
Priority: 10
Termination: Copying -- OK
21-Jan 09:19 bareos-dir JobId 75: Start Copying JobId 75, Job=NDMPCopy.2016-01-21_09.19.52_51
21-Jan 09:19 bareos-dir JobId 75: Using Device "FileStorage" to read.
21-Jan 09:19 bareos-dir JobId 76: Using Device "FileStorage2" to write.
21-Jan 09:19 bareos-sd JobId 75: Ready to read from volume "Full-0001" on device "FileStorage" (/var/lib/bareos/storage).
21-Jan 09:19 bareos-sd JobId 76: Volume "Copy-0004" previously written, moving to end of data.
21-Jan 09:19 bareos-sd JobId 76: Ready to append to end of Volume "Copy-0004" size=78177310
21-Jan 09:19 bareos-sd JobId 75: Forward spacing Volume "Full-0001" to file:block 0:78177310.
21-Jan 09:19 bareos-sd JobId 75: End of Volume at file 0 on device "FileStorage" (/var/lib/bareos/storage), Volume "Full-0001"
21-Jan 09:19 bareos-sd JobId 75: End of all volumes.
21-Jan 09:19 bareos-sd JobId 76: Elapsed time=00:00:01, Transfer rate=64.61 K Bytes/second
21-Jan 09:19 bareos-dir JobId 75: Bareos bareos-dir 15.2.2 (16Nov15):
Build OS: x86_64-redhat-linux-gnu redhat Red Hat Enterprise Linux Server release 7.0 (Maipo)
Prev Backup JobId: 73
Prev Backup Job: NDMPJob.2016-01-21_09.18.50_49
New Backup JobId: 76
Current JobId: 75
Current Job: NDMPCopy.2016-01-21_09.19.52_51
Backup Level: Incremental
Client: ndmp-client
FileSet: "NDMP Fileset"
Read Pool: "Full" (From Job resource)
Read Storage: "NDMPFile" (From Job resource)
Write Pool: "Copy" (From Job Pool's NextPool resource)
Write Storage: "File2" (From Storage from Pool's NextPool resource)
Next Pool: "Copy" (From Job Pool's NextPool resource)
Catalog: "MyCatalog" (From Default catalog)
Start time: 21-Jan-2016 09:19:54
End time: 21-Jan-2016 09:19:54
Elapsed time: 0 secs
Priority: 10
SD Files Written: 1
SD Bytes Written: 64,614 (64.61 KB)
Rate: 0.0 KB/s
Volume name(s): Copy-0004
Volume Session Id: 43
Volume Session Time: 1453307753
Last Volume Bytes: 78,242,384 (78.24 MB)
SD Errors: 0
SD termination status: OK
Termination: Copying OK
Now we successfully copied over the NDMP job.
Warning
list jobs will only show the number of main backup files as JobFiles. However, with list files jobid=… all files are visible.
Restore to NDMP Primary Storage System
Unfortunately, we are not able to restore the copied data to our NDMP storage. If we try we get this message:
To be able to do NDMP operations from the storage that was used to store the copies, we need to define a NDMP storage that is paired with it. The definition is very similar to our NDMPFile (Dir->Storage)
Storage, as we want to restore the data to the same NDMP Storage system:
Also we have to configure NDMP on the Bareos Storage Daemon bareos-sd2.example.com. For this follow the instruction from Bareos Storage Daemon: Configure NDMP.
After this, a restore from bareos-sd2.example.com directly to the NDMP Primary Storage System is possible.
Limitations
This list the specific limitiations of the NDMP_BAREOS protocol. For limitation for all Bareos NDMP implementation, see Bareos NDMP Common Limitations.
NDMP Job limitations when scanning in volumes
For NDMP jobs, all data is stored into a single big file. The file and directory information (File History in NDMP Terms) is stored as hardlinks to this big file.
Limitation - NDMP: File information are not available in the Bareos backup stream.
As hardlink information is only stored in the Bareos database, but not int the backup stream itself, it is not possible to recover the file history information from the NDMP stream with bscan.
As storing the database dump for disaster recovery and storing the bootstrap file offsite is recommended anyway (see Steps to Take Before Disaster Strikes), this should be not a big problem in correctly setup environments.
For the same reason, the information about the number of files of a job (e.g. JobFiles with list jobs command) is limited to the number of NDMP backup files in copied jobs.
Restore always transfers the full main backup file to the Primary Storage System
Contrary to NDMP_NATIVE, the NDMP_BAREOS implementation do not support NDMP “Direct Access Restore” (DAR).
On restore, the full main backup file (@NDMP/...%.
) is always transfered back to the Primary Storage System, together with a description, what files to restore.
The reason for this is that the Primary Storage System handles the backup data by itself. Bareos will not modify the backup data it receives from the Primary Storage System.
NDMP_NATIVE
The NDMP_NATIVE protocol is implemented since Bareos Version >= 17.2.3.
Bareos implements the Data Management Agent inside of the Bareos Director and is the only Bareos Daemon involved in the backups.
When using NDMP_NATIVE, the Tape Agent must be provided by some other systems. Some storage vendors provide it with there storages, or offer it as an option, e.g. Isilon with there “Isilon Backup Accelerator”.
Example Setup for NDMP_NATIVE backup
Configure a NDMP Client
This defines the connection to the NDMP Data Agent.
Verify, that you can access your Primary Storage System (Tape Agent) via Bareos:
*status client=isilon
Data Agent isilon.example.com NDMPv4
Host info
hostname isilon
os_type Isilon OneFS
os_vers v7.2.1.4
hostid xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Server info
vendor Isilon
product Isilon NDMP
revision 2.2.1
auths (2) NDMP4_AUTH_TEXT NDMP4_AUTH_MD5
Connection types
addr_types (2) NDMP4_ADDR_TCP NDMP4_ADDR_LOCAL
Backup type info of tar format
attrs 0x7fe
set FILESYSTEM=/ifs
set FILES=
set EXCLUDE=
set PER_DIRECTORY_MATCHING=N
set HIST=f
set DIRECT=N
set LEVEL=0
set UPDATE=Y
set RECURSIVE=Y
set ENCODING=UTF-8
set ENFORCE_UNIQUE_NODE=N
set PATHNAME_SEPARATOR=/
set DMP_NAME=
set BASE_DATE=0
set NDMP_UNICODE_FH=N
Backup type info of dump format
attrs 0x7fe
set FILESYSTEM=/ifs
set FILES=
set EXCLUDE=
set PER_DIRECTORY_MATCHING=N
set HIST=f
set DIRECT=N
set LEVEL=0
set UPDATE=Y
set RECURSIVE=Y
set ENCODING=UTF-8
set ENFORCE_UNIQUE_NODE=N
set PATHNAME_SEPARATOR=/
set DMP_NAME=
set BASE_DATE=0
set NDMP_UNICODE_FH=N
File system /ifs
physdev OneFS
unsupported 0x0
type NFS
status
space 224681156345856 total, 126267678720 used, 224554888667136 avail
inodes 324102912000 total, 323964781836 used
set MNTOPTS=
set MNTTIME=00:00:00 00:00:00
Configure a NDMP Fileset
This determines what filesystem to backup and configures the NDMP environment to use in the meta options for it.
The setting of "DIRECT = Y"
is required for Direct Access Recovery.
For more information, see Bareos Director: Configure NDMP Fileset.
Configure a NDMP Storage
This defines now to connect to the Tape and Robot Agents and what devices to use.
As we do not yet now the device names, we can put a placeholder string in Device (Dir->Storage)
and NDMP Changer Device (Dir->Storage)
:
Verify that the connection to the NDMP Tape Agent and Robot Agent work, by running the status storage command.
The Tape Agent will return information about the available tape drives. The Robot Agent will return information about the available tape changer device.
*status storage=isilon
Tape Agent isilon.bareos.com NDMPv4
Host info
hostname isilon
os_type Isilon OneFS
os_vers v7.2.1.4
hostid abcdefg
Server info
vendor Isilon
product Isilon NDMP
revision 2.2.1
auths (2) NDMP4_AUTH_TEXT NDMP4_AUTH_MD5
Connection types
addr_types (2) NDMP4_ADDR_TCP NDMP4_ADDR_LOCAL
tape HP Ultrium 5-SCSI I30Z
device HP-TLD-004-01
attr 0x4
set EXECUTE_CDB=t
set SERIAL_NUMBER=123456
tape HP Ultrium 5-SCSI I30Z
device HP-TLD-004-02
attr 0x4
set EXECUTE_CDB=t
set SERIAL_NUMBER=1234567
Robot Agent isilon.bareos.com NDMPv4
Host info
hostname isilon
os_type Isilon OneFS
os_vers v7.2.1.4
hostid 001517db7e38f40dbb4dfc0b823f29a31e09
Server info
vendor Isilon
product Isilon NDMP
revision 2.2.1
auths (2) NDMP4_AUTH_TEXT NDMP4_AUTH_MD5
scsi QUANTUM Scalar i6000 605A
device mc001
set SERIAL_NUMBER=VL002CX1252BVE01177
The interesting parts of the output is the device information both of the Tape Agent and Robot Agent.
As each NDMP backup or recovery operation always involves exactly one tape and at one robot agent.
We now know the device names and can configure what robot and what tape to use when this storage is used by bareos by updating the isilon (Sd->Storage)
resource:
Configure a Pool for the NDMP Tapes
Configure NDMP Jobs
To be able to do scheduled backups, we need to configure a backup job that will use the NDMP client and NDMP storage resources:
As we also need to be able to do a restore of the backed up data, we also need to define an adequate restore job:
Label Tapes
Before we can really start do do backups, first we need to label the tapes that should be used.
First we check if our robot has tapes with barcodes by running status slots:
*status slots
Slot | Volume Name | Status | Media Type | Pool |
------+------------------+-----------+----------------+--------------------------|
1@| ? | ? | ? | ? |
2@| ? | ? | ? | ? |
3@| ? | ? | ? | ? |
4@| ? | ? | ? | ? |
[...]
251*| BT0001 | ? | ? | ? |
252*| BT0002 | ? | ? | ? |
253*| BT0003 | ? | ? | ? |
254*| BT0004 | ? | ? | ? |
255*| BT0005 | ? | ? | ? |
256*| BT0006 | ? | ? | ? |
257*| BT0007 | ? | ? | ? |
[...]
Now we can label these tapes and add them to the pool that we have created for NDMP Tapes:
*label storage=isilon barcodes slots=251-257
Automatically selected Storage: isilon
Select Drive:
1: Drive 0
2: Drive 1
Select drive (1-12): 1
get ndmp_vol_list...
The following Volumes will be labeled:
Slot Volume
==============
251 BT0001
252 BT0002
253 BT0003
254 BT0004
255 BT0005
256 BT0006
257 BT0007
Do you want to label these Volumes? (yes|no): yes
Defined Pools:
1: Scratch
2: NDMP-Tape
3: Incremental
4: Full
5: Differential
Select the Pool (1-5): 2
ndmp_send_label_request: VolumeName=BT0001 MediaType=NDMP-Tape PoolName=NDMP-Tape drive=0
Catalog record for Volume "BT0001", Slot 251 successfully created.
ndmp_send_label_request: VolumeName=BT0002 MediaType=NDMP-Tape PoolName=NDMP-Tape drive=0
Catalog record for Volume "BT0002", Slot 252 successfully created.
ndmp_send_label_request: VolumeName=BT0003 MediaType=NDMP-Tape PoolName=NDMP-Tape drive=0
Catalog record for Volume "BT0003", Slot 253 successfully created.
ndmp_send_label_request: VolumeName=BT0004 MediaType=NDMP-Tape PoolName=NDMP-Tape drive=0
Catalog record for Volume "BT0004", Slot 254 successfully created.
ndmp_send_label_request: VolumeName=BT0005 MediaType=NDMP-Tape PoolName=NDMP-Tape drive=0
Catalog record for Volume "BT0005", Slot 255 successfully created.
ndmp_send_label_request: VolumeName=BT0006 MediaType=NDMP-Tape PoolName=NDMP-Tape drive=0
Catalog record for Volume "BT0006", Slot 256 successfully created.
ndmp_send_label_request: VolumeName=BT0007 MediaType=NDMP-Tape PoolName=NDMP-Tape drive=0
Catalog record for Volume "BT0007", Slot 257 successfully created.
We have now 7 volumes in our NDMP-Tape Pool that were labeled and can be used for NDMP Backups.
Run NDMP_NATIVE Backup
*run job=ndmp-native-backup-job yes
JobId 1: Start NDMP Backup JobId 1, Job=ndmp.2017-04-07_01.40.31_10
JobId 1: Using Data host isilon.bareos.com
JobId 1: Using Tape host:device isilon.bareos.com:HP-TLD-004-01
JobId 1: Using Robot host:device isilon.bareos.com:mc001
JobId 1: Using Tape record size 64512
JobId 1: Found volume for append: BT0001
JobId 1: Commanding robot to load slot @4146 into drive @256
JobId 1: robot moving @4146 to @256
JobId 1: robot move OK @4146 to @256
JobId 1: Opening tape drive HP-TLD-004-01 read/write
JobId 1: Commanding tape drive to rewind
JobId 1: Checking tape label, expect 'BT0001'
JobId 1: Reading label
JobId 1: Commanding tape drive to rewind
JobId 1: Commanding tape drive to NDMP9_MTIO_FSF 1 times
JobId 1: Waiting for operation to start
JobId 1: Operation started
JobId 1: Monitoring backup
JobId 1: DATA: bytes 3703831KB MOVER: written 3703644KB record 58788
JobId 1: LOG_MESSAGE: 'End of medium reached.'
JobId 1: DATA: bytes 4834614KB MOVER: written 4834053KB record 76731
JobId 1: Mover paused, reason=NDMP9_MOVER_PAUSE_EOM
JobId 1: Operation requires next tape
JobId 1: At EOM, not writing filemarks
JobId 1: Commanding tape drive to rewind
JobId 1: Closing tape drive HP-TLD-004-01
JobId 1: Commanding robot to unload drive @256 to slot @4146
JobId 1: robot moving @256 to @4146
JobId 1: robot move OK @256 to @4146
JobId 1: Found volume for append: BT0002
JobId 1: Commanding robot to load slot @4147 into drive @256
JobId 1: robot moving @4147 to @256
JobId 1: robot move OK @4147 to @256
JobId 1: Opening tape drive HP-TLD-004-01 read/write
JobId 1: Commanding tape drive to rewind
JobId 1: Checking tape label, expect 'BT0002'
JobId 1: Reading label
JobId 1: Commanding tape drive to rewind
JobId 1: Commanding tape drive to NDMP9_MTIO_FSF 1 times
JobId 1: Operation resuming
JobId 1: DATA: bytes 6047457KB MOVER: written 6047244KB record 95988
JobId 1: LOG_MESSAGE: 'End of medium reached.'
JobId 1: DATA: bytes 9668679KB MOVER: written 9668106KB record 153462
JobId 1: Mover paused, reason=NDMP9_MOVER_PAUSE_EOM
JobId 1: Operation requires next tape
JobId 1: At EOM, not writing filemarks
JobId 1: Commanding tape drive to rewind
JobId 1: Closing tape drive HP-TLD-004-01
JobId 1: Commanding robot to unload drive @256 to slot @4147
JobId 1: robot moving @256 to @4147
JobId 1: robot move OK @256 to @4147
JobId 1: Found volume for append: BT0003
JobId 1: Commanding robot to load slot @4148 into drive @256
JobId 1: robot moving @4148 to @256
JobId 1: robot move OK @4148 to @256
JobId 1: Opening tape drive HP-TLD-004-01 read/write
JobId 1: Commanding tape drive to rewind
JobId 1: Checking tape label, expect 'BT0003'
JobId 1: Reading label
JobId 1: Commanding tape drive to rewind
JobId 1: Commanding tape drive to NDMP9_MTIO_FSF 1 times
JobId 1: Operation resuming
JobId 1: LOG_MESSAGE: 'Filetransfer: Transferred 10833593344 bytes in 87.187 seconds throughput of 121345.079 KB/s'
JobId 1: LOG_MESSAGE: 'Filetransfer: Transferred 10833593344 total bytes '
JobId 1: LOG_MESSAGE: 'CPU user=0.528118 sys=54.575536 ft=87.182576 cdb=0.000000'
JobId 1: LOG_MESSAGE: 'maxrss=171972 in=1323908 out=17 vol=199273 inv=5883'
JobId 1: LOG_MESSAGE: '
Objects (scanned/included):
----------------------------
Regular Files: (2765/2765)
Sparse Files: (0/0)
Stub Files: (0/0)
Directories: (447/447)
ADS Entries: (0/0)
ADS Containers: (0/0)
Soft Links: (0/0)
Hard Links: (0/0)
Block Device: (0/0)
Char Device: (0/0)
FIFO: (0/0)
Socket: (0/0)
Whiteout: (0/0)
Unknown: (0/0)'
JobId 1: LOG_MESSAGE: '
Dir Depth (count)
----------------------------
Total Dirs: 447
Max Depth: 10
File Size (count)
----------------------------
== 0 14
<= 8k 1814
<= 64k 658
<= 1M 267
<= 20M 10
<= 100M 0
<= 1G 0
> 1G 2
-------------------------
Total Files: 2765
Total Bytes: 10827843824
Max Size: 5368709120
Mean Size: 3916037'
JobId 1: LOG_MESSAGE: '
File History
----------------------------
Num FH_HIST_FILE messages: 3212
Num FH_HIST_DIR messages: 0
Num FH_HIST_NODE messages: 0'
JobId 1: Async request NDMP4_NOTIFY_MOVER_HALTED
JobId 1: DATA: bytes 10581729KB MOVER: written 10581732KB record 167964
JobId 1: Operation done, cleaning up
JobId 1: Waiting for operation to halt
JobId 1: Commanding tape drive to NDMP9_MTIO_EOF 2 times
JobId 1: Commanding tape drive to rewind
JobId 1: Closing tape drive HP-TLD-004-01
JobId 1: Commanding robot to unload drive @256 to slot @4148
JobId 1: robot moving @256 to @4148
JobId 1: robot move OK @256 to @4148
JobId 1: Operation halted, stopping
JobId 1: Operation ended OKAY
JobId 1: ERR-CONN NDMP4_CONNECT_CLOSE exchange-failed
JobId 1: media #1 BT0001+1/4834053K@4146
JobId 1: valid label=Y filemark=Y n_bytes=Y slot=Y
JobId 1: media used=Y written=Y eof=N eom=Y io_error=N
JobId 1: label read=Y written=N io_error=N mismatch=N
JobId 1: fm_error=N nb_determined=Y nb_aligned=N
JobId 1: slot empty=N bad=N missing=N
JobId 1: media #2 BT0002+1/4834053K@4147
JobId 1: valid label=Y filemark=Y n_bytes=Y slot=Y
JobId 1: media used=Y written=Y eof=N eom=Y io_error=N
JobId 1: label read=Y written=N io_error=N mismatch=N
JobId 1: fm_error=N nb_determined=Y nb_aligned=N
JobId 1: slot empty=N bad=N missing=N
JobId 1: media #3 BT0003+1/913626K@4148
JobId 1: valid label=Y filemark=Y n_bytes=Y slot=Y
JobId 1: media used=Y written=Y eof=N eom=N io_error=N
JobId 1: label read=Y written=N io_error=N mismatch=N
JobId 1: fm_error=N nb_determined=Y nb_aligned=N
JobId 1: slot empty=N bad=N missing=N
JobId 1: Media: BT0001+1/4834053K@251
JobId 1: Media: BT0002+1/4834053K@252
JobId 1: Media: BT0003+1/913626K@253
JobId 1: ndmp_fhdb_lmdb.c:675 Now processing lmdb database
JobId 1: ndmp_fhdb_lmdb.c:679 Processing lmdb database done
JobId 1: Bareos bareos-dir 17.2.3:
Build OS: x86_64-unknown-linux-gnu redhat Red Hat Enterprise Linux Server release 6.8 (Santiago)
JobId: 1
Job: ndmp.2017-04-07_01.40.31_10
Backup Level: Full
Client: "isilon"
FileSet: "isilon" 2017-04-07 01:40:31
Pool: "NDMP-Tape" (From Job resource)
Catalog: "MyCatalog" (From Client resource)
Storage: "isilon" (From Job resource)
Scheduled time: 07-Apr-2017 01:40:31
Start time: 07-Apr-2017 01:40:33
End time: 07-Apr-2017 01:42:03
Elapsed time: 1 min 30 secs
Priority: 10
NDMP Files Written: 3,212
NDMP Bytes Written: 10,835,690,496 (10.83 GB)
Rate: 120396.6 KB/s
Volume name(s): BT0001|BT0002|BT0003
Volume Session Id: 0
Volume Session Time: 0
Last Volume Bytes: 935,553,024 (935.5 MB)
Termination: Backup OK
Run NDMP_NATIVE Restore
Now we want to restore some files from the backup we just did:
*restore
[...]
cwd is: /
: mark /ifs/home/testdata/git/bareos/src/console/bconsole
1 file marked.
: mark /ifs/home/testdatrandom5G-2
1 file marked.
$ done
Connecting to Director bareos:9101
1000 OK: bareos-dir Version: 17.2.3
Enter a period to cancel a command.
list joblog jobid=2
Automatically selected Catalog: MyCatalog
Using Catalog "MyCatalog"
JobId 2: Start Restore Job ndmp-restore.2017-04-07_01.48.23_13
JobId 2: Namelist add: node:6033532893, info:5464882688, name:"/ifs/home/testdata/random5G-2"
JobId 2: Namelist add: node:6033077461, info:40076288, name:"/ifs/home/testdata/git/bareos/src/console/bconsole"
JobId 2: Record size is 64512
JobId 2: Media: BT0001+1/4834053K@251
JobId 2: Media: BT0002+1/4834053K@252
JobId 2: Media: BT0003+1/913626K@253
JobId 2: Logical slot for volume BT0001 is 251
JobId 2: Physical(NDMP) slot for volume BT0001 is 4146
JobId 2: Media Index of volume BT0001 is 1
JobId 2: Logical slot for volume BT0002 is 252
JobId 2: Physical(NDMP) slot for volume BT0002 is 4147
JobId 2: Media Index of volume BT0002 is 2
JobId 2: Logical slot for volume BT0003 is 253
JobId 2: Physical(NDMP) slot for volume BT0003 is 4148
JobId 2: Media Index of volume BT0003 is 3
JobId 2: Commanding robot to load slot @4146 into drive @256
JobId 2: robot moving @4146 to @256
JobId 2: robot move OK @4146 to @256
JobId 2: Opening tape drive HP-TLD-004-01 read-only
JobId 2: Commanding tape drive to rewind
JobId 2: Checking tape label, expect 'BT0001'
JobId 2: Reading label
JobId 2: Commanding tape drive to rewind
JobId 2: Commanding tape drive to NDMP9_MTIO_FSF 1 times
JobId 2: Waiting for operation to start
JobId 2: Operation started
JobId 2: Monitoring recover
JobId 2: DATA: bytes 0KB MOVER: read 0KB record 0
JobId 2: DATA: bytes 11KB MOVER: read 11KB record 622
JobId 2: Mover paused, reason=NDMP9_MOVER_PAUSE_SEEK
JobId 2: Operation requires a different tape
JobId 2: Commanding tape drive to rewind
JobId 2: Closing tape drive HP-TLD-004-01
JobId 2: Commanding robot to unload drive @256 to slot @4146
JobId 2: robot moving @256 to @4146
JobId 2: robot move OK @256 to @4146
JobId 2: Commanding robot to load slot @4147 into drive @256
JobId 2: robot moving @4147 to @256
JobId 2: robot move OK @4147 to @256
JobId 2: Opening tape drive HP-TLD-004-01 read-only
JobId 2: Commanding tape drive to rewind
JobId 2: Checking tape label, expect 'BT0002'
JobId 2: Reading label
JobId 2: Commanding tape drive to rewind
JobId 2: Commanding tape drive to NDMP9_MTIO_FSF 1 times
JobId 2: Operation resuming
JobId 2: DATA: bytes 79884KB MOVER: read 79884KB record 85979
JobId 2: DATA: bytes 201740KB MOVER: read 201740KB record 87914
JobId 2: DATA: bytes 321548KB MOVER: read 321548KB record 89815
JobId 2: DATA: bytes 440332KB MOVER: read 440332KB record 91701
JobId 2: DATA: bytes 556044KB MOVER: read 556044KB record 93538
JobId 2: DATA: bytes 674828KB MOVER: read 674828KB record 95423
JobId 2: DATA: bytes 796684KB MOVER: read 796684KB record 97357
JobId 2: DATA: bytes 915468KB MOVER: read 915468KB record 99243
JobId 2: DATA: bytes 1036300KB MOVER: read 1036300KB record 101161
JobId 2: DATA: bytes 1157132KB MOVER: read 1157132KB record 103079
JobId 2: DATA: bytes 1277964KB MOVER: read 1277964KB record 104997
JobId 2: DATA: bytes 1398796KB MOVER: read 1398796KB record 106915
JobId 2: DATA: bytes 1518604KB MOVER: read 1518604KB record 108816
JobId 2: DATA: bytes 1622028KB MOVER: read 1622028KB record 110458
JobId 2: DATA: bytes 1741836KB MOVER: read 1741836KB record 112360
JobId 2: DATA: bytes 1859596KB MOVER: read 1859596KB record 114229
JobId 2: DATA: bytes 1981452KB MOVER: read 1981452KB record 116163
JobId 2: DATA: bytes 2094092KB MOVER: read 2094092KB record 117951
JobId 2: DATA: bytes 2207756KB MOVER: read 2207756KB record 119755
JobId 2: DATA: bytes 2328588KB MOVER: read 2328588KB record 121673
JobId 2: DATA: bytes 2448396KB MOVER: read 2448396KB record 123575
JobId 2: DATA: bytes 2569228KB MOVER: read 2569228KB record 125493
JobId 2: DATA: bytes 2689036KB MOVER: read 2689036KB record 127395
JobId 2: DATA: bytes 2810892KB MOVER: read 2810892KB record 129329
JobId 2: DATA: bytes 2926604KB MOVER: read 2926604KB record 131165
JobId 2: DATA: bytes 3043340KB MOVER: read 3043340KB record 133018
JobId 2: DATA: bytes 3163148KB MOVER: read 3163148KB record 134920
JobId 2: DATA: bytes 3279884KB MOVER: read 3279884KB record 136773
JobId 2: DATA: bytes 3400716KB MOVER: read 3400716KB record 138691
JobId 2: DATA: bytes 3518476KB MOVER: read 3518476KB record 140560
JobId 2: DATA: bytes 3636236KB MOVER: read 3636236KB record 142429
JobId 2: DATA: bytes 3757068KB MOVER: read 3757068KB record 144347
JobId 2: DATA: bytes 3877900KB MOVER: read 3877900KB record 146265
JobId 2: DATA: bytes 3994636KB MOVER: read 3994636KB record 148118
JobId 2: DATA: bytes 4116492KB MOVER: read 4116492KB record 150053
JobId 2: DATA: bytes 4237324KB MOVER: read 4237324KB record 151971
JobId 2: DATA: bytes 4331317KB MOVER: read 4331317KB record 153462
JobId 2: Mover paused, reason=NDMP9_MOVER_PAUSE_SEEK
JobId 2: Operation requires a different tape
JobId 2: Commanding tape drive to rewind
JobId 2: Closing tape drive HP-TLD-004-01
JobId 2: Commanding robot to unload drive @256 to slot @4147
JobId 2: robot moving @256 to @4147
JobId 2: robot move OK @256 to @4147
JobId 2: Commanding robot to load slot @4148 into drive @256
JobId 2: robot moving @4148 to @256
JobId 2: robot move OK @4148 to @256
JobId 2: Opening tape drive HP-TLD-004-01 read-only
JobId 2: Commanding tape drive to rewind
JobId 2: Checking tape label, expect 'BT0003'
JobId 2: Reading label
JobId 2: Commanding tape drive to rewind
JobId 2: Commanding tape drive to NDMP9_MTIO_FSF 1 times
JobId 2: Operation resuming
JobId 2: DATA: bytes 4424716KB MOVER: read 4424716KB record 154945
JobId 2: DATA: bytes 4544524KB MOVER: read 4544524KB record 156847
JobId 2: DATA: bytes 4663308KB MOVER: read 4663308KB record 158732
JobId 2: DATA: bytes 4781068KB MOVER: read 4781068KB record 160601
JobId 2: DATA: bytes 4902924KB MOVER: read 4902924KB record 162536
JobId 2: DATA: bytes 5022732KB MOVER: read 5022732KB record 164437
JobId 2: DATA: bytes 5138444KB MOVER: read 5138444KB record 166274
JobId 2: OK: /testdata/git/bareos/src/console/bconsole
JobId 2: OK: /testdata/random5G-2
JobId 2: LOG_MESSAGE: 'Filetransfer: Transferred 5368721181 bytes in 223.436 seconds throughput of 23464.803 KB/s'
JobId 2: LOG_MESSAGE: '
Objects:
----------------------------
Regular Files: (2)
Stub Files: (0)
Directories: (0)
ADS Entries: (0)
Soft Links: (0)
Hard Links: (0)
Block Device: (0)
Char Device: (0)
FIFO: (0)
Socket: (0)
Unknown: (0)'
JobId 2: LOG_MESSAGE: '
File Size (count)
----------------------------
== 0 0
<= 8k 1
<= 64k 0
<= 1M 0
<= 20M 0
<= 100M 0
<= 1G 0
> 1G 1
-------------------------
Total Files: 2
Total Bytes: 5368716925
Max Size: 5368709120
Mean Size: 2684358462'
JobId 2: Async request NDMP4_NOTIFY_MOVER_HALTED
JobId 2: DATA: bytes 5242893KB MOVER: read 5242893KB record 167932
JobId 2: Operation done, cleaning up
JobId 2: Waiting for operation to halt
JobId 2: Commanding tape drive to rewind
JobId 2: Closing tape drive HP-TLD-004-01
JobId 2: Commanding robot to unload drive @256 to slot @4148
JobId 2: robot moving @256 to @4148
JobId 2: robot move OK @256 to @4148
JobId 2: Operation halted, stopping
JobId 2: Operation ended OKAY
JobId 2: ERR-CONN NDMP4_CONNECT_CLOSE exchange-failed
JobId 2: LOG_FILE messages: 2 OK, 0 ERROR, total 2 of 2
JobId 2: media #1 BT0001+1/4834053K@4146
JobId 2: valid label=Y filemark=Y n_bytes=Y slot=Y
JobId 2: media used=Y written=N eof=N eom=N io_error=N
JobId 2: label read=Y written=N io_error=N mismatch=N
JobId 2: fm_error=N nb_determined=N nb_aligned=N
JobId 2: slot empty=N bad=N missing=N
JobId 2: media #2 BT0002+1/4834053K@4147
JobId 2: valid label=Y filemark=Y n_bytes=Y slot=Y
JobId 2: media used=Y written=N eof=N eom=N io_error=N
JobId 2: label read=Y written=N io_error=N mismatch=N
JobId 2: fm_error=N nb_determined=N nb_aligned=N
JobId 2: slot empty=N bad=N missing=N
JobId 2: media #3 BT0003+1/911610K@4148
JobId 2: valid label=Y filemark=Y n_bytes=Y slot=Y
JobId 2: media used=Y written=N eof=N eom=N io_error=N
JobId 2: label read=Y written=N io_error=N mismatch=N
JobId 2: fm_error=N nb_determined=Y nb_aligned=N
JobId 2: slot empty=N bad=N missing=N
JobId 2: Bareos bareos-dir 17.2.3:
Build OS: x86_64-unknown-linux-gnu redhat Red Hat Enterprise Linux Server release 6.8 (Santiago)
JobId: 2
Job: ndmp-restore.2017-04-07_01.48.23_13
Restore Client: isilon
Start time: 07-Apr-2017 01:48:25
End time: 07-Apr-2017 01:52:11
Elapsed time: 3 mins 46 secs
Files Expected: 2
Files Restored: 1
Bytes Restored: 5,368,722,944
Rate: 23755.4 KB/s
NDMP Common
This section contains additional information about the Bareos NDMP implementation that are valid for all Bareos NDMP protocols.
NDMP Backup Level
The trailing number in the main backup file (after the %
character) indicates the NDMP backup level:
Level |
Description |
---|---|
0 |
Full NDMP backup. |
1 |
Differential or first Incremental backup. |
2-9 |
second to ninth Incremental backup. |
Differential Backups
are supported. The NDMP backup level will be 1, visible as trailing number in the backup file (/@NDMP/ifs/home%1
).
Incremental Backups
are supported. The NDMP backup level will increment with each run, until a Full (0) or Differential (1) will be made. The maximum backup level will be 9. Additional Incremental backups will result in a failed job and the message:
NDMP Debugging
To debug the NDMP backups, these settings can be adapted:
This will create a lot of debugging output that will help to find the problem during NDMP backups.
Bareos NDMP Common Limitations
NDMP Fileset limitations
Limitation - NDMP: A NDMP fileset should only contain a single File directive and Meta options.
Using multiple Include (Dir->Fileset)
File directives should be avoided.
The Bareos Director would try to handle them by running multiple NDMP jobs in a single Bareos job.
Even if this is working fine during backup, restore jobs will cause trouble.
Normally (Protocol (Dir->Client)
=Native) Filesets get handled by the bareosFd. When connecting directly to a NDMP Clients (Protocol (Dir->Client)
=NDMP*), no Bareos File Daemon is involved and therefore most Fileset options can’t be used. Instead, parameters are handled via Options - Meta from Include (Dir->Fileset)
.
Single file restore on incremental backups
Limitation - NDMP: No single file restore on merged backups.
Unfortunately, it is currently (bareos-15.2.2) not possible to restore a chain of Full and Incremental backups at once. The workaround for that problem is to restore the full backup and each incremental each in a single restore operation.
Temporary memory mapped database
Limitation - NDMP: 64-bit system recommended.
The Bareos Director uses a memory mapped database (LMBD) to temporarily store NDMP file information.
On some 32-bit systems the default File History Size (Dir->Job)
requires a larger memory area than available.
In this case, you either have to lower the File History Size (Dir->Job)
or preferably run the Bareos Director on a 64-bit system.
Tested Environments
Bareos NDMP support have been tested against:
Vendor |
Product |
NDMP Subsystem |
Bareos version |
Tape Agent |
Features |
Remarks |
---|---|---|---|---|---|---|
Isilon |
Isilon OneFS v7.2.1.4 |
Isilon NDMP 2.2.1 |
bareos-17.2.3 |
Isilon Backup Accelerator |
Protocol: NDMP_NATIVE |
|
Isilon |
Isilon OneFS v7.2.0.1 |
Isilon NDMP 2.2 |
bareos-16.2.6 |
Bareos Storage Daemon |
||
Isilon |
Isilon OneFS v7.1.1.5 |
Isilon NDMP 2.2 |
bareos-15.2.2 |
Bareos Storage Daemon |
||
NetApp |
Release 8.2.3 7-Mode |
bareos-15.2.2 |
Bareos Storage Daemon |
|||
Oracle/Sun |
ZFS Storage Appliance |
OS 8.3 |
bareos-15.2.2 |
Bareos Storage Daemon |