Hi,
I have some questions on relabeling tapes, but first the background. We
have several tapes that have been marked full prematurely. This happens
from time to time. I say prematurely in the sense that these volumes
would normally have at least 100 GB or so native, but they filled up at
like 30 GB, or in some cases way less, which I'm guessing was when
NetWorker encountered some kind of problem and wrote an eof on the tape.
This has happened on both LTO and SDLT media but doesn't happen that
often. The majority of the savesets on these tapes, or at least the
majority of the space that's being used up on the tape, appears to be
from these "bogus" savesets. I'm calling them bogus because I don't
think I can recover anything from them. 'mminfo' shows the following
values for ssflags, sumflags and clflags respectively for these save
sets:
ssflags sumflags clflags saveset_recover_window_shows
======================================================
vrEiF cE i (aborted)
vrEiF cE s (aborted suspect)
vrEi cE s (in-progress suspect)
vrEi cE i (in-progress)
It is interesting to note that none of these savesets has been cloned
even though in some cases, the clflags value does show 's' which I
thought didn't get set until you tried to read the tape. Regardless,
many of these savesets were automatically re-run at same level and
seemed to complete just fine. Others were not, but I then scheduled
re-runs for them at the same respective level, and those copies all look
good, i.e. ssflags shows something like: vF and clflags shows nothing.
I'm thinking to simply clone the good savesets (i.e. ones with values
like vF or vrF) on the affected tapes so as to conglomerate them all
onto one (or at least fewer) tape(s) so I can then re-label all the
affected tapes. I should note that just about all of the "bogus"
savesets are completely contained on their volumes. It seems a shame to
keep some volume that only had like 4 GB written to it when maybe only a
few hundred MB of savesets on these are non-bogus. Could just clone
those off and then re-use the tape. Anyway, this brings up some
questions:
1. Might it be better to simply try to clone all the savesets on the
tape, and whatever nsrclone is able to do, keep that, and whatever it
can't (nsrclone probably will have problems when it runs against ones
with values like: "vrEiF cE i") then I probably couldn't recover data
from it anyway, so just wipe the tape once done?
2. Can you typically recover data from a saveset wherein one or more
pieces are identified with one of the above values like: "vrEiF cE i",
"vrEiF cE s", etc.?
3. In the event that one of the bogus savesets is not completely
contained on the volume and one or more of the other pieces appears okay
(i.e. flags look good) then is there a reliable way to actually recover
data from one of the good pieces? As has been mentioned in other
posting(s), it is possible to use scanner to write the data from a piece
of a saveset onto disk, but could you do anything with it at that point?
I've never tried.
I'm just thinking that since I have valid copies of these bad savesets
-- in many cases they were run immediately after the preceding failure,
but for others I ran them maybe 2-3 days later -- then I can probably
delete those bad ones as they're just about all completely contained on
the given volume. Now, for the very few that are not completely
contained on there, and for which there is at least one good piece, it
is always possible that there could be some data on one of those good
pieces that might be salvageable and might have been deleted before the
subsequent re-run. Might have to just take that chance if getting
anything back from the good piece is too involved?
Thanks.
George
--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listserv.temple DOT edu or visit the list's Web site at
http://listserv.temple.edu/archives/networker.html where you can
also view and post messages to the list. Questions regarding this list
should be sent to stan AT temple DOT edu
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
|
