Now that we've been through full and incremental backups, I can summarize.
I originally asked:
|>Greetings!
|>
|>To those of you using Digital's PolyCenter Network Save and Restore (NSR)
|>either single server or full server, I ask the following:
|>
|>Have you gotten NSR to obey directives not to backup certain *local* mount
|>points?
|>
|>I'm running Networker_V3.2A-06/07 Single Server, but these exact results can
|>be seen on Networker_V3.2-03/29 (full server).
|>
|>What I've got is a CDROM mounted at /cdrom, and /misc as its own mounted
|>temporary partition, which gets used a lot. I don't want to backup /misc.
|>
|>
|>my /.nsr file contains:
|>
|>null: cdrom
|>null: misc
|>
|>
|>This fails, NSR ignores this and backs up my CDROM each time I get a full save,
|>and /misc files are always getting backed up as incremental or full depending
on
|>the file creation/modification.
|>
|>To keep /misc files from getting backed up, I had to place a /misc/.nsr file
|>with:
|>
|>null: .*
|>null: *
|>
|>since the /.nsr entry for misc was ignored. This appears broken...
|>
|>Now, since /cdrom is the mount point for a CDROM, I can't add a
|>/cdrom/.nsr file. I have also tried editing the Unix standard directives
|>to include:
|>
|><< /cdrom >>
|> +skip: .?* *
|>
|>but this fails to be obeyed also.
|>
|>
|>
|>So, how does one get NSR to *really* understand that I don't want to backup
|>CDROMs or other mount points as it should? I don't consider the necessity
|>of creating .nsr files with null: .* and null: * in each mount point directory
|>a reasonable solution when I interpret the documentation to mean I simply
|>need a .nsr file up one level with the null: <mount-point> to exclude
|>all files under that mount point.
|>
|>
|>Also, can anybody explain the following:
|>
|>When the Unix standard directive has:
|><< /tmp >>
|> +skip: .?* *
|>and there is no /tmp/.nsr file, I get files backed up (in this case a single):
|>
|>--- Successful Save Sets ---
|>...
|> icefog.sois.alaska.edu: /tmp level=incr, 1 KB 00:00:14 1 file
|>...
|>
|>But, when the Unix standard directive has:
|><< /tmp >>
|> +skip: .?* *
|>and there is a /tmp/.nsr file containing:
|>null: .*
|>null: *
|>
|>I get 0 files, 0 KB backed up:
|>
|>--- Successful Save Sets ---
|>...
|> icefog.sois.alaska.edu: /tmp level=incr, 0 KB 00:00:57 0 files
|>...
|>
|>and, yes, there was a file in /tmp created minutes before this backup ran.
|>
|>
|>Any ideas, comments, fixes? I think this is broken, but I may be mistaken.
To which I summarized the solution as this:
One must use directives (create a new one or add to an existing one) to keep
from backing up read-only mount-point filesystems. The directive must have
the following syntax (verify in the /nsr/res/nsr.res file, and directives
screen):
<< /<mount-point >>
(tab)+skip: .?* *
without the (tab) starting the second line, this will fail.
To keep from backing up read-write mount-point filesystems, the .nsr file
must be in the top level of that filesystem. i.e. for the /misc mount-point
/misc/.nsr needs to exist and have an entry for each and every file and
directory below it which you wish to avoid taking a backup of, or wildcard
the entries to get everything.
One can also use directives for read-write mount-point filesystems as outlined
in the read-only mount-point filesystems summarization.
This will be fully tested this evening when we get our standard backup.
Thanks to:
Mandell Degerness of ITSD 389-3539 <MDEGERNESS_at_galaxy.gov.bc.ca>
for pointing out the location of the .nsr file must be at the root level
of the filesystem in concern (not above it). I had my /.nsr file attempting
to exclude /misc, where I needed to have a /misc/.nsr file with wildcard
entries to exclude /misc/*.
"Sheila Hollenbaugh" <shollen_at_valhalla.cs.wright.edu>
for pointing out the <tab> requirement in the directives - 2 of our 4
CDROMs had spaces, and two had tabs, but all four were getting backed up
each time we took a full backup. These four CDROM directive entries were the
final four entries in that directive. The two with bad syntax (spaces
instead of tabs) happened to precede the two with correct syntax. Perhaps
this is significant, and is perhaps, why all four had been getting backed
up. Now, with all four with correct syntax, none of the four are getting
backed up.
Tom Webster <webster_at_europa.mdc.com>
for suggesting the flip side of the coin - rather than use directives to
exclude, enter filesets to include in the client setup. This was not an
option for us.
Randy M. Hayman
haymanr_at_icefog.alaska.edu
Received on Sun Nov 17 1996 - 21:56:38 NZDT