Ok, now that's interesting... I'll agree that the fact that you're using
2.4.3b3 and aren't having problems indicates that you're right.
However... The comments in the amanda.conf file say:
"Note that the `full pathname' of a file within its
filesystem starts with `./', because of the way amanda runs
gnutar: `tar -C $mountpoint -cf - --lots-of-options .' (note
the final dot!) Thus, if you're backing up `/usr' with a
diskfile entry like ``host /usr gnutar-root', but you don't
want to backup /usr/tmp, your exclude list should contain
the pattern `./tmp', as this is relative to the `/usr' above.
Please refer to the man-page of gnutar for more information.
If a relative pathname is specified as the exclude list,
it is searched from within the directory that is
going to be backed up."
The man page says:
" For exclude list, If the file name is relative, the
disk name being backed up is prepended. So if this
is entered:
exclude list ".amanda.excludes"
the actual file use would be /var/.amanda.excludes
for a backup of /var, /usr/local/.amanda.excludes
for a backup of /usr/local, and so on."
These suggest that the exclude pattern, whether as a exclude pattern or a
file containing a list of patterns, is supposed to have *relative*
pathnames, not absolute pathnames. Yes, your list is relative, but your
path for the exclude list file itself is absolute
(/usr/local/etc/amanda/exclude.gtar) which is not quite what the docs
recommend....
I'll try some absolute paths later: if they work, it suggests a
documentation bug, rather than a coding one...
J.
On Mon, 19 Aug 2002, Gene Heskett wrote:
> On Monday 19 August 2002 20:09, John Ouellette wrote:
> >We're going a bit far afield from what Kevin had originally posted
> > but oh well.
> >
> >Are you using version 2.4.2, Gene?
>
> No, 2.4.3b3-20020805 right now. And I'm not having any known
> problems, none. And other than some broken scripting in the
> amstatus command, I've had zilch troubles with amanda in the last 4
> or more months, thru a goodly number of 2.4.3b* versions.
>
> > Kevin and I are using v2.4.3b3
> > and have found the same problem with exclude lists. What I have
> > found (again, with v2.4.3b3) was that the exclude keyword in a
> > dumptype definition *causes the size estimates from tar to fail*.
> > The debug info from Amanda shows that tar is never even called.
> > From one of my tests (sendsize.*.debug):
> >
> >++++++++++++++++++++++++++++++++++++++++++
> >sendsize: debug 1 pid 27066 ruid 499 euid 499 start time Mon Aug
> > 19 14:31:57 2002
> >/home/amanda/libexec/sendsize: version 2.4.3b3
> >sendsize: calculating for amname '/', dirname '/'
> >sendsize: getting size via gnutar for / level 0
> >++++++++++++++++++++++++++++++++++++++++++
>
> Any one of my copious numbers of sendsize.*.debug files is exactly
> like the rest, containing only a copy of the exclude file as
> follows:
> -----
> ./usr/dumps/*
> ./var/pacct
> -----
> So I again submit that the exclude logic isn't broken, but that the
> file itself simply isn't being found due to a lack of a full
> pathlist in the dumptype specification, as is shown that I am using
> below.
>
> >(Note that there were three other partitions called with the same
> >dumptype: the size estimates weren't even attempted by amanda --
> > the above is the entire contents of the debug file.) This with a
> > dumptype definition of:
> >
>
> This is yours
>
> >define dumptype TEST {
> > global
> > comment "test dumptype"
> > program "GNUTAR"
> > compress none
> > index
> > exclude "./amanda" <---whats the current pwd when this
> executes?
> > record no
> >}
> >
> >(The idea of the exclude was the same as Kevin's: to exclude the
> > holding disk areas from those disks.) The result is the same if
> > I use:
> >
> >exclude "./amanda*"
> >exclude "*amanda*
> >exclude list ".amanda.exclude"
> >exclude list "./.amanda.exclude"
> >
> >etc., in the dumptype definition. The exclude keyword seemed to
> > work in v2.4.2p2, but I decided to upgrade to 2.4.3b3 to get
> > other features working (I'm not sure if it was worth it now...).
>
> Thats odd. I also note that you still aren't giving it a full path
> to the exclude file. Its possible that the fact that it can't find
> it might be enough of an error to cause the tar call to fail.
>
> >I've taken a brief look into the code for how exclude lists are
> > handled, but there was such a big change from v2.4.2 to v2.4.3,
> > I'm not sure if I'll be able to figure out what broke the exclude
> > keyword...
> >
> >Note that in the amandad debug file, the exclude list seems to be
> > parsed OK, so it really must be in the call to sendsize that the
> > error is occurring...
> >
> >J.
> >
> >On Mon, 19 Aug 2002, Gene Heskett wrote:
> >> On Monday 19 August 2002 16:12, John Ouellette wrote:
> >> >I've played around with the exclude lists for a while (as
> >> > defined in the dumptypes) and have had no luck in getting any
> >> > configs with exclude lists to work, whether I define the paths
> >> > with the leading ./ or no....
> >> >
> >> >Kevin was using an exclude list defined in the disklist, but I
> >> > assume it is the same code that parses this and passes it on
> >> > to 'tar', so the same bug might exist.
> >> >
> >> >And, according to the docs and examples, you do need the
> >> > quotes... (or get a parse error if you exclude them).
> >>
> >> I'm not here, but I'm using an exclude file too, here it is:
> >> ---------------------
> >> ./usr/dumps/*
> >> ./var/pacct
> >> ---------------------
> >>
> >> Also, I recall the filename didn't give the full path in your
> >> earler example, which should look like this in your amanda.conf:
> >> ------------------------------------------
> >> define dumptype root-tar {
> >> global
> >> program "GNUTAR"
> >> comment "root partitions dumped with tar"
> >> compress none
> >> index yes
> >> exclude list "/usr/local/etc/amanda/exclude.gtar"
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >> priority low
> >> }
> >>
> >> define dumptype user-tar {
> >> root-tar
> >> comment "user partitions dumped with tar"
> >> exclude list "/usr/local/etc/amanda/exclude.gtar"
> Again^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >> compress server best
> >> priority medium
> >> }
> >> -----------------etc, etc taking note that each "dumptype" can
> >> have its own unique name-it-whatever-file.
> >>
> >> >I've started to trudge through the code to see if the bug is
> >> > obvious, but haven't managed to get far.
> >> >
> >> >J.
> >> >
> >> >On Mon, 19 Aug 2002, Gene Heskett wrote:
> >> >> On Monday 19 August 2002 12:00, Kevin Passey wrote:
> >> >> >Thanks John
> >> >> >
> >> >> >That is what I am using.
> >> >> >
> >> >> >I only what to omit the /dumps folders.
> >> >>
> >> >> Did you specify them as "./dumps" (without the quotes of
> >> >> course) The missing leading dot in your line above says its
> >> >> your exclude file thats broken. Tar's exludes are relative
> >> >> to the currently being processed directory, hence the leading
> >> >> ./
> >> >>
> >> >> >I think I will have to create them in a separate partition.
> >> >> >
> >> >> >Thanks for that.
> >> >> >
> >> >> >Regards
> >> >> >
> >> >> >Kevin
> >> >> >
> >> >> >
> >> >> >
> >> >> >-----Original Message-----
> >> >>
> >> >> From: John Ouellette [mailto:ouellet AT amnh DOT org]
> >>
> >> [...]
>
>
--
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
John Ouellette | Ph: 212-313-7919
Department of Astrophysics | Fax: 212-769-5007
American Museum of Natural History | e-mail: ouellet AT amnh DOT org
Central Park West at 79th St. | http://research.amnh.org/astrophysics
New York, NY 10024-5192 |
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|