SAVEINDEX(8) Legato NetWorker 3.0 SAVEINDEX(8)
NAME
saveindex - save a NetWorker server's indexes
recoverindex - recover a NetWorker server's indexes
SYNOPSIS
saveindex
[ -q|-v ] [ -E ] [ -n ] [ -p ] [ -R ] [ -s server ] [ -P
printer ]
[ -l level ] [ -t date ] [ -W width ] [ -c client ]
recoverindex
[ -q|-v ] [ -s server ]
DESCRIPTION
These shell scripts are used in preparing for and recovering from the
loss of a NetWorker server's indexes. Both the on-line file index
and media indexes are saved and recovered. Typical events causing
such disasters are accidental removal of an index by a user or a disk
crash. Saveindex is used to produce a copy of the indexes, recov
erindex is used to recover the indexes. See nsrcrash(8) for a dis
cussion of general issues and procedures for NetWorker client and
server crash recovery.
SAVEINDEX
Saveindex should be run at regular intervals to insure that up-to-
date copies of the NetWorker indexes are available in the event of a
disaster. Saveindex is automatically run by savegroup(8) each night.
As each NetWorker client completes its saves, savegroup uses savein
dex to save the client's on-line file index. The same schedule con
trolling the client's save levels is used to control saving the
client's on-line file index.
Client files saved after saveindex was last run will not appear in a
recovered index. Index entries for these files may be generated with
scanner(8). Since savegroup automatically runs saveindex after it
saves a client's files, only those files saved via the explicit use
of the save command would fall into this category. If large numbers
of user files are explicitly backed up by running save, it would be
prudent to run saveindex when the save command completes. This will
assure that the most recent entries in the index are recoverable.
When saving a NetWorker server's index, saveindex will generate a
save set called bootstrap. This save set contains an incremental
save of the server's on-line file index and the server's complete
media index. lp(1) is then used to print a list containing informa
tion from mminfo(8). This information should be kept in a safe
place, it will be needed to run recoverindex. In particular, the
ssid (save set id), and the volume containing the most recent boot
strap save set will be requested by recoverindex. An example of the
mminfo output is shown below:
Apr 1 17:25 1992 NetWorker bootstrap information Page 1
volume client date time size ssid flags level name
Licensed material--property of copyright holder(s) 1
SAVEINDEX(8) Legato NetWorker 3.0 SAVEINDEX(8)
mars.1 mars 03/29/92 04:04:53 44 KB 17850934 incr bootstrap
mars.2 mars 03/30/92 15:18:51 97 KB 17851053 incr bootstrap
mars.2 mars 03/31/92 17:03:37 109 KB 17851140 incr bootstrap
mars.3 mars 04/01/92 17:22:37 113 KB 17851237 incr bootstrap
Before moving the NetWorker service to another machine, saveindex
should be run as the last use of the NetWorker system on the original
machine. This will ensure that a complete index exists and is recov
erable.
RECOVERINDEX
Recoverindex is used to recover a NetWorker server's on-line file and
media indexes. Recoverindex rebuilds the NetWorker server's indexes
from the media. Note that this command will overwrite the NetWorker
server's existing file and media indexes.
The NetWorker system must be fully installed and correctly configured
prior to using this command. If any of the NetWorker software was
lost, re-install NetWorker from the distribution tape before running
recoverindex.
The same release of NetWorker should be used, and it should be
installed in the same location (including the use of any symbolic
links with respect to /nsr and the directories under it) as it was in
the original installation. If the original install directories are
unknown, run recoverindex twice. The first time will fail, but the
path names where it is trying to recover the indexes to will be dis
played. With this information the original installation directories
may be determined. After creating the appropriate directories, run
recoverindex again to actually recover the indexes.
Recoverindex works in two phases. First, it extracts the contents of
a bootstrap save set. This save set contains the NetWorker media and
on-line file indexes. The on-line file index only contains one
entry, for the on-line file index itself. In the second phase,
recoverindex will run recover(8) to completely recover the NetWorker
server's file index. This last phase is performed in the background,
so the operator can respond to NetWorker volume mount requests.
After recoverindex is started, it will ask for the most recent boot
strap save set id. This number is found in the sixth column (labeled
ssid) of the last line of the bootstrap information sheet printed by
saveindex. In the example above, the ssid of the most recent boot
strap save set is `17851237'. Recoverindex will then ask that the
volume, `mars.3' in the example above, containing the selected boot
strap save set be inserted into the drive. Both the ssid and the
volume name will have to be determined by the user from the printed
sheet, recoverindex has no way of determining this information.
If the bootstrap save set spans more than one volume, indicated by an
`h' (head) or `t' (tail) in the F (for flags) column of the bootstrap
information sheet, start with the first volume (the one marked with
an `h' in the F column). A `c' (complete) in the column indicates
the save set completely fit on one volume (see mminfo(8)).
Licensed material--property of copyright holder(s) 2
SAVEINDEX(8) Legato NetWorker 3.0 SAVEINDEX(8)
Recoverindex will then ask for the name of the drive where the volume
has been loaded. The volume will then be scanned, and the bootstrap
save set will be extracted.
After the volume scan completes, recoverindex will complete. A
recover will be running in the background, reconstructing a complete
index from the save sets generated by the server's save schedule.
Since the save sets may be spread across multiple volumes, net
worker(8), or nsrwatch(8) should be run, and the volumes mounted as
they are requested.
When the recover completes, the message ``The on-line index is now
fully recovered'' will be displayed. Once this message has appeared,
recover can be used, in its normal interactive browsing mode, to
recover other server files, including other NetWorker client on-line
file indexes. Once a NetWorker client's index is recovered, that
client can start recovering its files.
OPTIONS
-c client
The name of a NetWorker client who's index should be saved. A
bootstrap save, with its associated printer output, will only
be performed if client is also a NetWorker server. Without
this option, the indexes for all NetWorker clients on this
server will be saved.
-E Estimate. Produce an estimate of the amount of data to be
saved prior to actually saving any data.
-l level
The level of save to perform. There are 12 levels: full, 1
though 9, incr, and skip. Full specifies that all index
records are to be saved. Incr specifies that only index
records created since the last save of any level be saved.
See savefs(8) for more information about levels.
-n No save. Just produce an estimate of the amount of data that
would be saved (as in -E) but don't actually save any data.
-P printer
Specifies the printer used by lp to print the bootstrap infor
mation generated by saveindex. See lp(1) for how the default
printer is selected.
-p Preview. List the name of the indexes which would be saved,
the level of save that would be performed, and the time after
which index entries must have been created to be saved, but
don't actually do the save. This information is gleaned from
the time stamps in each index entry and the ``NSR schedule''
resource.
-q Quiet. Display only error messages.
-R Report success or failure, by echoing a "succeeded" or
Licensed material--property of copyright holder(s) 3
SAVEINDEX(8) Legato NetWorker 3.0 SAVEINDEX(8)
"failed" message as the last act. This is used by savegroup
when it runs saveindex.
-t date
The date, in getdate(3) format, which should be used when
examining the schedule to determine what level of save should
be performed.
-s server
The server to perform the actual command on. If this names a
machine other the the current one, an rsh(1) to execute the
command with the identical command line arguments will be per
formed to the specified machine.
-v Verbose. Cause lots of debugging style information to be gen
erated as things proceed.
-W width
The width the -p option will use when displaying its output.
FILES
/nsr/index/servername/db
The NetWorker server's on-line file index.
/nsr/mm/mmvolume The NetWorker server's media index.
BUGS
To recover files not referred to by the recovered on-line file index
(i.e. they were saved after the last run of saveindex), scanner(8)
must be used to rebuild the media and on-line file indexes from the
contents of the volumes generated between the time of the last run of
saveindex and the loss of the original index.
SEE ALSO
nsrcrash(8), nsr(8), nsrd(8), nsrclient(5), nsrschedule(5),
mminfo(8), recover(8), save(8), savefs(8), savegroup(8),
saveindex(8), scanner(8), networker(8), nsrwatch(8), lp(1), rsh(1),
getdate(3)
Licensed material--property of copyright holder(s) 4