On Fri, Sep 29, 2006 at 09:00:13AM +0700, Olivier Nicole wrote: > Hi, > > This may be because I am using a prehistoric version of Amanda > (2.4.2p2) but, after I make a dump, I make a verify. > > Some of the DLE cannot be verified becausethe client has been compile > to use a GNU tar that is different from the one used when compiling > the server: different meaning that the tar binary has a diferent path. > > So I get the kind of following results: > > Checked alhazen.desktops._home_mdailey_Projects.20060928.1 > Skipped ufo1000._web.20060928.1 (** Cannot do /usr/local/bin/mygtar dumps) > Skipped mailback._usr.20060928.0 (** Cannot do /usr/local/bin/gtar dumps) > Checked alhazen.desktops._.20060928.0 > > - /usr/local/bin/mygtar is a tar wrapper > > - mailback has been build to use /usr/local/bin/gtar instead of > /usr/local/bin/tar > > By using a tar wrapper, one could provide just any archive format he > wishes and there is no way that the good old tar on the server would > know what to do with it. > > But I beleive that this is a very rare case and that for most of us, > that the client uses /bin/tar, /usr/local/bin/gtar or > /usr/local/bin/tarwrapp or whatever, the file is in tar format; so the > good old tar on the server should be able to amverify it. > > And just in case the file is really not in tar format, then the verify > would issue a "Cannot do /usr/local/bin/mygtar dumps" warning. And > checking if the fileis not in tar format could be done 2 ways: with > the file(1) command or with runing tar and receiving an error. > > OK, that would really improve amverify (and amrestore I beleive) if > that is not included in more recent versions of Amanda. >
If I get your suggestion correctly, amverify currently looks at the tape file header block and finds which command was used to create the archive. if it does not have the same command "path" available on the server that was used to create the archive it only checks readability of the tape. you suggest that instead, it should just try the servers compiled in version of tar. Seems like a good idea, but I'll play devil's advocate. There are many more things that could upset amverify's ability to better test the archive. These include the compession utility used (gzip vs bzip2 or ???), archive program used, and use of encryption of the archive. I think these things will be coming into greater usage at amanda sites as amanda continues to add new features. It may be that amverify will only be fully useful if the client that created the archive can become involved. -- Jon H. LaBadie [EMAIL PROTECTED] JG Computing 4455 Province Line Road (609) 252-0159 Princeton, NJ 08540-4322 (609) 683-7220 (fax)