Data is growing unbelievably fast.
I now have arrays on two different servers amounting to 12TB, 20TB and 20TB. That's capacity, not
filled yet. But, it is in addition to the multi-TB capacity housed in the servers themselves. I've
been taking huge directories of data and breaking them down into manageable pieces for backup using
a combination of include and exclude lists. I typically do this by using du to determine the size of
all the sub directories and then grouping those to keep within a size limit that seems to make sense
based on holding disk capacity and so on. Suppose there are a, b, c, d, e, and f. One DLE each might
have an include for a, b, c and e. Then another might do the overall directory and exclude a, b, c
and e. That would "guarantee" that I would catch everything.
The two 20TB arrays are new and not very full. The 12TB array is 60% full, and I have about a two
page list of DLEs specifying how it is to be backed up. While I have been very careful with this, it
is obviously an error prone strategy. It's easy to imagine being asked to recover something and
having to come back with, "sorry, I seem to have missed that in my backup specification."
I know that I have seen posts from people talking about auto generating lists
of DLEs.
What about the alternative of having Amanda, or a custom script, look at the disklist and analyze
what may have been missed? I do proof read my disklist meticulously and periodically, but . . .
--
---------------
Chris Hoogendyk
-
O__ ---- Systems Administrator
c/ /'_ --- Biology & Geology Departments
(*) \(*) -- 317 Morrill Science Center
~~~~~~~~~~ - University of Massachusetts, Amherst
<[email protected]>
---------------
Erdös 4
