* John R. Jackson ([EMAIL PROTECTED]) [20010502 16:49] thus spake:
> >? gtar: ./dev/log: socket ignored
>
> These messages are ignored in the current version of Amanda.
Hmmmm, what is the current *stable* version?
>
> >? gtar: ./@ulysses/scratch2: Cannot stat: Stale NFS file handle
yeah, I just took care of this.
I'll see how the backup fares tonight.
>
> This may be "fatal" as far as tar is concerned. See below.
>
> >? gtar: ./opt/amanda/adm/amanda/mrs/index/valeria/dks0d2s0/20010502_0.gz.tmp:
>Warning: Cannot stat: No such file or directory
>
> This implies you are trying to back up the Amanda index area, which
> is almost guaranteed to be active, which tar doesn't like. You might
> consider setting up an exclusion list to not back up this area.
OK!
Would a exclusion pattern like this work?
./opt/amanda/adm/amanda/*
Just a question then: in the event of a catastrophic failure
of the disk holding the amanda files on the server, what's the way
to proceed? If I include the above exclusion pattern, then upon
restore I'll be missing the curinfo, index and log subdirs for
all amanda configurations. Is this the way to go?
Thanks a million,
jf
>
> >sendbackup: error [/bin/gtar returned 2]
>
> I'm not sure which of the above errors tar is not happy with (maybe both).
>
> You can probably make things better with an exclusion list so tar does
> not even descend into the areas that cause it trouble. But you may
> have to rebuild Amanda and get IGNORE_TAR_ERRORS defined, either in
> client-src/sendbackup-tar.c or by setting up CPPFLAGS before running
> ./configure (or any number of other techniques).
>
> Note, however, that this will cause Amanda to ignore **all** tar errors.
> You'll have to watch the reports closely for what is fatal and what
> is not.
>
> >jf
>
> John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
--
