clyde@ut-sally.UUCP (Clyde Hoover) (02/03/84)
Does anybody have a 4.2 (or 4.1c) version of 'dump(8)' that
logs which files it writes to tape? We really need to have this
sort of thing in order to find which backup tape some lusers'
file is on. Making dump do this looks like a non-trivial
job, so I hope that someone else has done so.
Thanks in advance... (Of course, if nobody has one, then no
thanks will be in order).
Yes, I know that a 'restore t' can be done after the dump
to get a table of contents - I was wondering if someone
has taught dump to do one as it goes.
--
Clyde W. Hoover @ Univ. of Texas Computation Center; Austin, Texas
(Shouter-To-Dead-Parrots)
clyde@ut-ngp.{UUCP,ARPA} clyde@ut-sally.{UUCP,ARPA} ihnp4!ut-ngp!clydemark@umcp-cs.UUCP (02/05/84)
We have one of these at Maryland. When users lose a file they
run the program 'whichtape filename' which tells them what
old versions we have backed up. Dump automatically builds
the database of names as it is run.
--
Mark Weiser
UUCP: {seismo,allegra,brl-bmd}!umcp-cs!mark
CSNet: mark@umcp-cs ARPA: mark@marylandpetec@umcp-cs.UUCP (02/09/84)
>>>>We have one of these at Maryland. When users lose a file they >>>>run the program 'whichtape filename' which tells them what >>>>old versions we have backed up. Dump automatically builds >>>>the database of names as it is run. This is a more complete description of the procedures and utilities at UoMD. -------------------------------------------------------------------- Here at Maryland, we have an incremental backup daily, and on every other Wednesday we have a complete dump that is run in single-user mode. To help us keep track of which files are on tape, we have delevoped several programs. (Actually, we only worry about the names of files on the daily dump tapes, since we know that all the files exist on the bi-weekly). We have not hacked at 'dump' to list which files it is dumping, so we cannot provide you with a verbose dump. What we do do is this (it is run with a shell script): do incremental dump dumpdir > /tmp/dumpdir filtapmgr < /tmp/dumpdir This is only an outline and does not show the true syntax of the commands. Dumpdir tells us what files are on the tape (this seems easier that making 'dump' verbose). Filtapmgr is a program which builds and maintains a file database. The database is actually a giant hierarchical tree structure with each node containing information about a file. The info included is the file inodes, names of the last 3 dump tapes, and whether the file is a directory. As tapes are reused, old information is deleted automatically. To access the data-base, the program 'whichtape' may be used. This simply finds the node with information on that file and prints it. Various options may be given to vary the amount of information retrieved. It has a recursive option that will list sub-trees of the database. We dump 4 file systems: /u, /usr, /g, and / . The average size of the database file is 275-325 1k blocks, and may contain information on 12000-15000 different files. The 2 programs 'filtapmgr' and 'whichtape' go hand-in-hand with each other and require the investment of a certain amount of disk space for the database. However, as the restorer of lost files, I have found that it is usually a trivial matter; I just run 'whichtape' and mount that tape, or I retrieve a file off of the bi-weekly dump tape. I am freed from tracking down the whereabouts of a lost file. There seem to have been enough requests that we will post the sources to filtapmgr and whichtape, along with man entries and a skeleton shell script. Just give us a little time to get it all together. -- Call-Me: Pete Cottrell, Univ. of Md. Comp. Sci. Dept. UUCP: {seismo,allegra,brl-bmd}!umcp-cs!petec CSNet: petec@umcp-cs ARPA: petec.umcp-cs@CSNet-Relay