This section provides important notes and tips about your Backup software and provides recommendations, where appropriate.
Use the script on Example 1-1 to lock a Clear Case VOB (version object database), do the backup, then unlock the VOB. Your customized script file must reside in the same directory where the Backup save program is installed (for example, on a Solaris system, the save program is installed in the /usr/sbin/nsr directory). You enter the name of the script into the Backup Command attribute of the Client resource configured for the ClearCase VOB. During a scheduled backup, the Backup Command is invoked instead of the usual save program.
#!/bin/sh# export the SHELL that we are going to use SHELL=/bin/sh export SHELL # export the correct PATH so that all the required binaries can be found case $0 in /* ) PATH=/usr/atria/bin:/bin:/usr/bin:\Q/bin/dirname $0\Q c=\Q/bin/basename $0\Q ;; * )PATH=/usr/atria/bin:/bin:/usr/bin:/usr/sbin:/usr/bin/nsr:/usr/sbin/nsr c=$0 ;; esac export PATH # These are the valid statuses which save reports on completion of the backup statuses=" failed. abandoned. succeeded. completed savetime= " # Perform the PRECMD (Lock VOB) /usr/atria/bin/cleartool setview -exec "/usr/atria/bin/cleartoollock -c \ `VOB backups in progress' -vob /cm_data/mis_dev" magic_view > /tmp/voblock.log 2>&1 # Perform backup on client save "$@" > /tmp/saveout$$ 2>& # cat out the save output cat /tmp/saveout$$ # search for the backup status in the output reported by save for i in ${statuses}; do result=\Qgrep "${i}" /tmp/saveout$$\Q if [ $? != 0 ]; then echo ${result} fi done # Perform the POSTCMD (Unlock VOB) /usr/atria/bin/cleartool setview -exec "/usr/atria/bin/cleartoolunlock - vob /cm_data/mis_dev" \ magic_view > /tmp/vobunlock.log 2>& # make sure to gracefully exit out of this shell script exit 0 |
You must include the save command within your script in order for the backup to occur. If the script file is not in the same directory as the save program, the backup will fail.
For Emass/Grau and StorageTek Silo, you need to edit the /etc/init.d/S95networker startup file.
After the lines:
'start') (echo 'starting Backup daemons:') > /dev/console |
For Emass/Grau silos enter the next 8 lines:
DAS_SERVER = name-of-DAS-server export DAS_SERVER DAS_CLIENT = name-of-system-as-defined-to-DAS-server export DAS_CLIENT DAS_MEDIUM = type-of-tape-drive-used (select from one of the following: 3480, OD-Thick, OD-Thin, DECDLT, 8MM, 4MM, D2, VHS, 3590, CD, TRAVAN, DTF, BETACAM, AUDIOTAPE) export DAS_MEDIUM ACI_MEDIA_TYPE = same-as-DAS_MEDIUM export ACI_MEDIA_TYPE |
For StorageTek silos enter the next four lines:
CSI_HOSTNAME = name-of-ACSLS-system export CSI_HOSTNAME /networker-binaries-path/mini_el & /networker-binaries-path/ssi & |
The maximum value for parallelism and devices depends on the Backup product purchased and the number of enabled storage nodes, as shown in Table 1-1. Regardless of the number of enabled storage nodes, the maximum limit for parallelism is 512 and the maximum limit for devices is 256.
Table 1-1
Backup Product |
With Each Enabled Storage Node |
Without Storage Nodes |
---|---|---|
Server Edition |
Parallelism: 32 Maximum = (32 x #nodes) + 32 |
Parallelism: 32 |
Devices: 16 Maximum = (16 x #nodes) + 16 |
Devices: 16 |
|
NetWork Edition |
Parallelism: 32 Maximum = (32 x #nodes) + 32 |
Parallelism: 32 |
Devices: 16 Maximum = (16 x #nodes) + 16 |
Devices: 16 |
|
Power Edition |
Parallelism: 32 Maximum = (32 x #nodes) + 64 |
Parallelism: 64 |
Devices: 16 Maximum = (16 x #nodes) + 32 |
Devices: 32 |
If the setting for the system clock on a Backup client with Backup 5.0 client software installed differs from that of the Backup server by more than one minute, you receive the following message during a scheduled backup that invokes the savegrp program:
Warning: unsynchronized client clock detected |
If the gap in synchronization is less than two minutes, you will see these errors, but the schedule is still honored. (This situation does not occur on machines with prior releases of the Backup client software installed.) To avoid this, keep the server and client clocks synchronized to within a minute of each other.
If you do not name any backup servers when you are prompted for the names of authorized servers to contact the client, all backup servers will be authorized to contact the client. Failure to specify a list of one or more authorized backup servers allows any backup server to contact the client.
When performing a disaster recovery, some scenarios require additional configuration prior to performing the disaster recovery.
If you have more than one media device, you must configure that device prior to beginning the disaster recovery. The default device created by a Backup install is always a 4mm device. If your device is not a 4mm device, you must delete this device from within Backup and create a correct device type.
After reinstalling Backup, and before starting mmrecov, edit the client resource for the server and set its browse and retention policies to a decade. This allows all of the server's records to be recovered by mmrecov. If you do not do this, all records are recovered, but any records that are more than a month old are then discarded (since the browse policy is one month by default).
If you relocated the server's file index path to a new location, you must edit the client index path resource prior to beginning the disaster recovery. The correct steps are to change the index path to match the path used prior to the disaster. Restart the services to alert Backup that the index path has changed. Finally, run nsrck -c from the /usr/sbin/nsr directory to create an empty file index into which the records are recovered.
If you use the -o option twice with the nsrjb command, the first -o will be ignored. For example, if you enter:
# nsrjb -o notreadonly -o notfull -S 2 |
The -o notreadonly option will not be executed. To do this, you must enter the command as follows:
# nsrjb -Y -o notreadonly -S n # nsrjb -Y -o notfull -S n |
The save set recover feature does not currently work on computers running Windows 95. A forthcoming patch will fix save set recover on Windows 95.
The save set recover feature is available on Windows NT clients, but in some cases, the client user interface does not display the correct amount of disk space required for the recover. Disregard the incorrect information and proceed with the recovery.