Hi Charles, These files do contain file meta data in my understanding but they are in a data format, as you notice. I am not sure how file info can be obtained in a human readable format from the file.
Paul On Mon, 2014-02-24 at 01:44 +0100, Charles Stroom wrote: > Hi Paul, > > you say that files in the gnutar-lists contain the datestamp/stat > of the files backed-up? How can I see the contents of these files, > because they are binary files? Is there a specific am* command? > > Regards, Charles > > > > > On Fri, 21 Feb 2014 19:06:40 -0800 > Paul Yeatman <pyeat...@zmanda.com> wrote: > > > Hi Charles, > > > > This can happen from time to time although I have yet to see tar be > > the reason for the blame. Also, the tar being used is the one on the > > client so, presuming this is unchanged, it should be working the same > > as it was on that client with the previous Amanda server and version. > > > > Amanda calls tar with the --listed-incremental flag pointing to a file > > in typically the ~amanda/gnutar-lists directory on the client. During > > an incremental backup, tar is comparing files on disk to the metadata > > recorded in this file during the last backup at the previous level. > > > > Thus, places to start are to ensure that these files are being > > correctly created and updated. The file name will contain the name > > of the host and directory being backed up with a _# extension > > indicating the backup level. A level 1 is comparing against the _0 > > file, a level 2 against the _1 file, etc. If this file doesn't exist > > or is empty, tar will back everything up again. > > > > If these files are being created and updated correctly, tar is using > > this information to compare the stats of the files currently on disk > > to the ones recorded in the incremental list file of the previous > > level. Thus, you want to compare a "stat" of one of the files > > recently backed up to the stat output of when it is backed up at the > > previous level. > > > > Thus, for instance, force a level 0 backup for one of the DLEs. Might > > as well force a level 0 backup for all DLEs for which level 1 was the > > same size as level 0! Just before the level 0 backup starts, run > > "stat" on a few of the files contained in the DLE. As soon as Amanda > > goes to the next level, check if one of the files you ran stat on was > > backed up during the level 1 backup. If so, compare its stat output > > as it is currently disk to the output before the level 0 backup. As > > long as tar is given the incremental list correctly by Amanda, > > something in the stats of the file must have changed (other than > > access time and device number if --no-check-device flag was given by > > Amanda). > > > > Paul > > > > >
