Data ONTAP dump can write to its standard output (most useful with rsh(1) from a UNIX system), to a remote tape device on a host that supports the rmt(8) remote tape protocol or to a local tape drive, connected directly to the system (na_tape(4)).
The subtree argument specifies a subtree to be dumped. This is one way to allow dump to work with remote tape devices that are limited to 2 GB of data per tape file. The specified subtree may be in the active file system (e.g. /home) or in a snapshot (e.g. /.snapshot/weekly.0/home). If the subtree is in the active file system, dump creates a snapshot named snapshot_for_dump.X where X is a sequentially incrementing integer. This naming convention prevents conflicts between concurrently executing dumps. The dump is run on this snapshot so that its output will be consistent even if the filer is active. If dump does create a snapshot, it automatically deletes the snapshot when it completes.
Another way to allow dump to work with remote tape devices that are limited to 2 GB of data per tape file is to dump to multiple tape files or ``volumes''. The B option to dump specifies the maximum amount of data to be dumped to one volume; when that much data has been written to one volume, dump will start writing to another volume. A list of tape devices can be specified as arguments to the f option, and the volumes will be written to the devices in that list, in order. If there are no more devices in the list, dump will re-use the last device in the list, after prompting the user to indicate that they've put a new tape in that device.
The Data ONTAP dump command handling of end of tape is slightly different than that of the standard UNIX dump. Instead of aborting the entire dump if EOT is reached, the Data ONTAP dump starts a new volume and continues the dump on the next tape. Specifying a large value for the B option will cause dump to utilize the entire tape.
You can enter the dump command on a trusted host through rsh. It is preferable to enter the dump command through rsh if the backup takes a significant amount of time. This is because if you enter the dump command on the console, the filer does not display the console prompt until the backup is finished. During the time when the filer is backing up data, you do not have console access to the filer.
Another advantage of running the dump command through rsh is that you can control backups from UNIX shell scripts or crontab entries.
0-9 Dump levels. A level 0, full backup, guarantees the entire file system is copied. A level number above 0, incremental backup, tells dump to copy all files new or modified since the last dump of a lower level. The default level is 0.
f files Write the backup to the specified files. files may be:
a list of the names of local tape devices, in the form specified in na_tape(4);
a list of the names of tape devices on a remote host, in the form host:devices;
the standard output of the dump command, specified as -.
The list may have a single device or a comma-separated list of devices; note that the list must either contain only local devices or only devices on a remote host and, in the latter case, must refer to devices on one particular remote host, e.g.
tapemachine:/dev/rst0,/dev/rst1
Each file in the list will be used for one dump volume in the order listed; if the dump requires more volumes than the number of names given, the last file name will used for all remaining volumes after prompting for media changes.
Use sysconfig -t for a list of local tape devices. See below for an example of a dump to local tape.
For a dump to a tape device on a remote host, host must support the standard UNIX rmt(8) remote tape protocol.
By default, dump writes to standard output.
B blocks
Set the size of the dump file to the specified number of 1024-byte
blocks, If this amount is exceeded, dump will close the current
file and open the next file in the list specified by the f option.
If there are no more files in that list, dump will re-open the last
file in the list, and prompt for a new tape to be loaded.
It is recommended to be a bit conservative on this option.
This is one way to allow dump to work with remote tape devices that are limited to 2 GB of data per tape file.
toaster> dump 0ufbB adminhost:/dev/rst0 63 2097151 /
To make a level 0 dump of /home on a 2 GB tape to a remote tape device, use:
toaster> dump 0ufbB adminhost:/dev/rst0 63 2097151 /home
To make a level 0 dump of /home on a 2 GB tape to a local tape drive (no rewind device, unit zero, highest density) use:
toaster> dump 0ufbB nrst0a 63 2097151 /home
To make a level 0 dump of the entire file system to a local tape drive (no rewind device, unit zero, highest density), with each tape file in the dump being less than 2 GB in size, without operator intervention, using a tape stacker, with four tape files written per tape, assuming that the dump requires no more than 10GB, use:
toaster> dump 0ufbB nrst0a,nrst0a,nrst0a,urst0a,rst0a 63 2097151 /
This will:
write the first three files to the norewind device, so that they, and the next dump done after them, will appear consecutively on the tape;
write the next file to the unload/reload device. This will cause the stacker to rewind and unload the tape after the file has been written and then load the next tape.
write the last file to the rewind device, so that the tape will be rewound after the dump is complete.
To back up all files and directories in a volume named engineering that are not in a qtree you created, use:
toaster> dump 0ufQ rst0a /vol/engineering
To run the dump command through rsh, enter the following command on a trusted host:
adminhost# rsh toaster dump 0ufbB adminhost:/dev/rst0 63 2097151 /home
Deleting or renaming a snapshot that is currently being backed up is not supported and will lead to dump errors.
Restore
As stated previously, filer dump output format is compatible with SunOS 4.x/Solaris 1.x and SunOS 5.x/Solaris 2.x dump. The filer supports a local restore command (see na_restore(1)), so the restoration process can be performed on the filer. It can also be be performed via a restore done on an NFS client machine; if such a restore is being done, the client system should be checked to ensure it supports SunOS-compatible dump/restore format.
Client Dump and Restore Capability
If a client is to be used for performing filer dump and/or restore, it is important to check what the maximum dump and restore capabilities of your client system are before setting up a dump schedule. There are some client systems which do not support dump and restore of greater than 2 GB while others may support very large dumps and restores. It is especially important to check the restore capability of your system when using the filer local tape dump since the filer supports dumps that are greater than 2 GB.
Tape Capacity and Dump Scheduling
Along with the potential 2-GB restriction of dump or restore on a client system, it is important to consider your tape capacity when planning a dump schedule. For the filer local tape option, the Exabyte 8505 supports an approximate maximum capacity of 10GB per tape using compression. If a client system is used as the target for your dump, the capacity of that tape drive should be checked for dump planning.
If your filer file system exceeds the capacity of the local tape drive or the client system dump/restore, or you choose to dump multiple file system trees to parallelize the restore process with multiple tape drives, you must segment your dump to meet these restrictions.
One way to plan a dump schedule with a UNIX client system is to go to the root mount point of your filer and use the du command to obtain sizes of underlying subtrees on your filer file system. Depending on the restrictions of your client's dump and restore capability or recording capacity of the tape device being used, you should specify a dump schedule that fits these restrictions. If you choose to segment your dump, the norewind device (see na_tape(4)) can be used to dump multiple tape files to one physical tape (again, choose a dump size which meets the criteria of your client restore and capacity of your tape drive).
The following example shows the du output from a filer file system on a client that supports dump and restore that are greater than 2 GB:
client% du -s * 4108 etc 21608 finance 5510100 home 3018520 marketing 6247100 news 3018328 users
CIFS Data
The Data ONTAP dump command dumps the CIFS attributes and 8.3 name data for each file that is backed up. This data will not be backed up by a dump run on an NFS client machine. This data will not be restored by a restore run on an NFS client machine. This data will only be restored if a local restore is done of a backup created by the Data ONTAP dump command.