Backup Central

I have a client that has a (to me anyway) 'strange' scenario:

"Think about this scenario.
Day 1: Directory x has files, a, b & c. Amanda does a full backup
Day 2: rm file a. Amanda does a differential backup
Day 3: rm file b. Amanda does a differential backup,
Day 4: Directory x has only file c. We remove Directory x. And
restore Directory x. . From what you are saying, the restored
Directory would have files a, b & c. I would expect to restore
Directory x with only file c -- the state of the directory upon the
last amanda backup."

Question: what will be in Directory x after doing an amrecover of
Directory x? Will files a and b be restored or not? And why or why
not? What is considered the correct answer to this partitular thought
exercise? And why?

As a long time computer user *my* expectation would be for Directory x
to have all three files, a, b & c in it after the recover. Is my
expectation wrong? Why?

Note: this is *separate* from a full restore in the event of a disk
crash.

Side question: would the above answers *change* if the backups were done
with gnutar instead of dump?


--
Robert Heller -- 978-544-6933 / heller < at > deepsoft.com
Deepwoods Software -- http://www.deepsoft.com/
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments

The long and the short of the answer is this: I'd expect to be able to
recover any file: a, b, or c in any combination from up to dumpcycle
days past. the default behaviour should of course be to restore it to
exactly the state the user requests (as of the backup date the user
requests to recover from) which may require recovering from a full
backup, then deleting files as part of recovering differentials.
of course with amanda, it would probably be a level 0 and a pair of
level 1s, so the number of "differential" recovers required is 1,
unless there are levels 0, 1, and 2 in use in this 3 backup span.

On 11/14/2010 17:51, Robert Heller wrote:
I have a client that has a (to me anyway) 'strange' scenario:

"Think about this scenario.
Day 1: Directory x has files, a, b& c. Amanda does a full backup
Day 2: rm file a. Amanda does a differential backup
Day 3: rm file b. Amanda does a differential backup,
Day 4: Directory x has only file c. We remove Directory x. And
restore Directory x. . From what you are saying, the restored
Directory would have files a, b& c. I would expect to restore
Directory x with only file c -- the state of the directory upon the
last amanda backup."

Question: what will be in Directory x after doing an amrecover of
Directory x? Will files a and b be restored or not? And why or why
not? What is considered the correct answer to this partitular thought
exercise? And why?

As a long time computer user *my* expectation would be for Directory x
to have all three files, a, b& c in it after the recover. Is my
expectation wrong? Why?

Note: this is *separate* from a full restore in the event of a disk
crash.

Side question: would the above answers *change* if the backups were done
with gnutar instead of dump?

At Sun, 14 Nov 2010 18:03:18 -0800 Christ Schlacta <aarcane < at > aarcane.org> wrote:


The long and the short of the answer is this: I'd expect to be able to
recover any file: a, b, or c in any combination from up to dumpcycle
days past. the default behaviour should of course be to restore it to
exactly the state the user requests (as of the backup date the user
requests to recover from) which may require recovering from a full
backup, then deleting files as part of recovering differentials.
of course with amanda, it would probably be a level 0 and a pair of
level 1s, so the number of "differential" recovers required is 1,
unless there are levels 0, 1, and 2 in use in this 3 backup span.

If I recover Directory x (just specifying the *directory*, not the
specific files in the directory) up to the last incremental, should a
and b be there are not? We actually did a test and several 'old' files
where restored and were wondering if amanda messed up or not. The last
incremental (a level 2) was dated *after* the files were known to have
been deleted. Does amanda *on its own* delete files as it goes from
full to incremental to incremental ..., or does it restore a union of
the files up to the specificed backup date, expecting human judgement
about the 'extra' files?


On 11/14/2010 17:51, Robert Heller wrote:
I have a client that has a (to me anyway) 'strange' scenario:

"Think about this scenario.
Day 1: Directory x has files, a, b& c. Amanda does a full backup
Day 2: rm file a. Amanda does a differential backup
Day 3: rm file b. Amanda does a differential backup,
Day 4: Directory x has only file c. We remove Directory x. And
restore Directory x. . From what you are saying, the restored
Directory would have files a, b& c. I would expect to restore
Directory x with only file c -- the state of the directory upon the
last amanda backup."

Question: what will be in Directory x after doing an amrecover of
Directory x? Will files a and b be restored or not? And why or why
not? What is considered the correct answer to this partitular thought
exercise? And why?

As a long time computer user *my* expectation would be for Directory x
to have all three files, a, b& c in it after the recover. Is my
expectation wrong? Why?

Note: this is *separate* from a full restore in the event of a disk
crash.

Side question: would the above answers *change* if the backups were done
with gnutar instead of dump?





--
Robert Heller -- 978-544-6933 / heller < at > deepsoft.com
Deepwoods Software -- http://www.deepsoft.com/
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments

On Sun, Nov 14, 2010 at 08:51:27PM -0500, Robert Heller wrote:
I have a client that has a (to me anyway) 'strange' scenario:

"Think about this scenario.
Day 1: Directory x has files, a, b & c. Amanda does a full backup
Day 2: rm file a. Amanda does a differential backup
Day 3: rm file b. Amanda does a differential backup,
Day 4: Directory x has only file c. We remove Directory x. And
restore Directory x. . From what you are saying, the restored
Directory would have files a, b & c. I would expect to restore
Directory x with only file c -- the state of the directory upon the
last amanda backup."

Question: what will be in Directory x after doing an amrecover of
Directory x? Will files a and b be restored or not? And why or why
not? What is considered the correct answer to this partitular thought
exercise? And why?

As a long time computer user *my* expectation would be for Directory x
to have all three files, a, b & c in it after the recover. Is my
expectation wrong? Why?

Note: this is *separate* from a full restore in the event of a disk
crash.

Side question: would the above answers *change* if the backups were done
with gnutar instead of dump?

I've only dealt with tar. Thus, I'm only assuming that dump results
would be similar. Also, I never restore into the original directory
tree. Instead I restore into an empty directory and copy over what
I want to the original location.

One thing you seem to be missing is that amrecover works with an
"as of date". You tell amrecover to recover the files as they
existed on a specific date. It has been a while since I did not
specify a specific date and accepted the default, but IIRC, the
default is the current date (or maybe the date of last dump of
that DLE).

In your sample scenario, I would expect an amrecover of directory X,
using a default date, after either the day 3 or day 4 dumps to
include only file C in directory X.

jl
--
Jon H. LaBadie jon < at > jgcomp.com
JG Computing
12027 Creekbend Drive (703) 787-0884
Reston, VA 20194 (703) 787-0922 (fax)

amrecover will recover the full dump followed by the dump of day 3.

The full dump will restore files a, b & c
The dump of day 3 will delete file a & b

The result is you get get only the file c.
You get the filesystem as it was while the latest restored backup was done.

Jean-Louis

Robert Heller wrote:
I have a client that has a (to me anyway) 'strange' scenario:

"Think about this scenario.
Day 1: Directory x has files, a, b & c. Amanda does a full backup
Day 2: rm file a. Amanda does a differential backup
Day 3: rm file b. Amanda does a differential backup,
Day 4: Directory x has only file c. We remove Directory x. And
restore Directory x. . From what you are saying, the restored
Directory would have files a, b & c. I would expect to restore
Directory x with only file c -- the state of the directory upon the
last amanda backup."

Question: what will be in Directory x after doing an amrecover of
Directory x? Will files a and b be restored or not? And why or why
not? What is considered the correct answer to this partitular thought
exercise? And why?

As a long time computer user *my* expectation would be for Directory x
to have all three files, a, b & c in it after the recover. Is my
expectation wrong? Why?

Note: this is *separate* from a full restore in the event of a disk
crash.

Side question: would the above answers *change* if the backups were done
with gnutar instead of dump?