Is there a way to restore an incremental save without deleting the account
and restoring the full?
I know it prompts you at the end of an account restore whether you want to
restore the incremental but is there a way to bypass the regular account
restore and get right to the part where it restores the incremental stuff?

bob wrote:
system (the from) does not exist on unless it is restored.

HTH, Patrick, <;=)

*Your answer seems to assume my situation involves a single system*
Let's say I have two systems, one live and one development which are
connected by vpn.
I want to refresh the development machine at the end of each work day but I
don't want to:
a) blow away the entire dev machine with a full restore, or
b) manually restore each data account that I want to refresh
Also I'd rather keep the filesave small to use less bandwidth when
downloading.

When you attempt an account restore, I realize it will stop if the account
already exists.
What I was hoping for was some undocumented option for this.
No reason D3 has to really restore that account because the situation will
be as if it had just completed the full account restore.

Does anyone have a better way to do this?
Right now best way I can think of to do what I want is to write a program
which saves each account then deletes them and then restores them just to
get to that incremental prompt at which point it grabs the incremental save
from the live system.
The reason I don't like this solution so much is because each step depends
on the step before it not failing.
These are Linux systems I'm talking about but in my experience with windows
(fsi) systems, accounts can get into a state where they just refuse to be
deleted. I don't think I've run into that on a Linux system but still I'd
rather keep this process as simple as possible.


"(latimerp)" <"(latimerp)"@comcast.net> wrote

Hot-backup Production to Development. Just log the data files that are
important to you.

This will require at least one "sync"-ing operation to put the two machines
into a state where hot-backup will work.

Mark Brown


"bob" <bgf66 (AT) comcast (DOT) net> wrote

Hot-backup is a valid answer but it might be a bit extreme for the
stated purpose, especially since RD charges a lot for it.

Bob, you originally asked about incremental saves. As we learn more
about your environment the suggestions may change. If we know exactly
what you need up front then we (probably) won't make suggestions that
seem inadequate.

Incremental save, transaction logging, and hot-backup are used to
recreate an environment, not to merge two environments. It seems like
you're looking for a sort of Concurrent Versioning System that stops
developers from over-writing one another, while ensuring backups of
anything new. For this custom purpose you need a custom method. In
short I would not do this synchronously with triggers, but
asynchronously in case there are errors. I might write this like
optimistic lock handling, checking the remote system to ensure it
hasn't been updated before copying new updates. This may or may not
involve triggers and the OSFI to copy data from one system to another.

HTH
T

"Mark Brown" wrote:

It's not so much that I'm worried about developers clobbering each other. As
far as code, I have a system in place to track changes and do version
control. This is more a matter of wanting to keep a fresh copy of live data
on the development machine without having to download such a large psuedo
floppy file every night. The incremental save appeals to me since it's so
much smaller and faster. Plus just one disconnect during the 1 gig file
download and it's got to start all over.
I agree, hot backup isn't really the answer I'm looking for. Too many
dependencies.
Right now I'm thinking my best bet is just to simulate what I really want to
do by deleting and restoring the account for the purpose of getting to that
incremental restore prompt.

Suggestion about giving more context noted but I wanted to ask just a
specific question without troubling anyone with the reason I'm asking for
the sake of keeping it brief. The question about incremental saves was just
a thought I had about a something I might be able use to acheive my goal
without getting so much into what the goal was.
Sometimes there are undocumented options that people have stumbled upon and
was hoping that might be the case here.

Anyway, thanks to all for the help.







"Tony Gravagno" <g6q3x9lu53001 (AT) sneakemail (DOT) com.invalid> wrote

There is a TLOG-RESTORE command that will restore data from transacton
logger. If you're only interested in things that have changed, this might
be a way to go.

Mark Brown


"bob" <bgf66 (AT) comcast (DOT) net> wrote

that sounds promising, thanks


"Mark Brown" <mbrown (AT) drexelmgt (DOT) com> wrote

"bob" wrote:
If you can ensure that people aren't going to be changing data in the
development environment then you can easily do an incremental save on
the production system and restore it to the development system.

I think an obvious problem is that people like to change data in a
development system, so how are you going to reconcile live data with
test data? If you can guarantee that the test data will remain in its
restored state then incrementals will work just fine. If not then the
target data will be out of sync with the updates - there's not a lot
that can be done here.

Also, I did suggest triggers and OSFI. I think that's really your
best bet because you can write your own code to sync the environments
rather than leaving it to a simple save/restore tool.

T

Yeah sorry, I did see the suggestion about using triggers.
I am using them (in the form of callx) as a logging facility to trace the
changes users make to key files.
Forgive my lack of knowlege regarding OSFI, is that something I would use in
a Linux system? I know about the VME and FSI in NT but haven't dealt with it
at all in Linux.

But getting back to callx, I wanted something a little more "hands-off" if
that makes sense.
I'd rather take an approach where I wouldn't impact the working system and
possibly introduce a catalog issue or some other sort of bug that might
impact the client's daytime activity. Not trying to be lazy but this is just
one of those things instinct tells me to keep as simple as possible. I
figure if D3 is tracking updates using dirty bits anyhow, then ideally I
could take advantage of that and pull that information out whether that's
through looking at an incremental save, tx-log or even if I can write code
to check those dirty bits and do the synch. It's something that can be done
after hours and nothing new is introduced into the business code.

I was looking for some documentation that would explain to me the tape
format since I can do readt's on the incremental pseudo floppy file but
haven't had any luck finding it yet.

Also maybe there is a simpler way to work with triggers that I'm not aware
of. Something I could put in system wide might work.
Regarding the transaction log idea, tlog-restore, I was hoping that was
something that accessed those dirty bits but seems that's something also
that requires me to start and stop transaction logging and set up files that
I want to include. Is that true?
tlog-restore was asking me for a tape device but I'm not clear on how I
would create that tape.

Again thanks for all your time and suggestions on this and also to the
others who have made suggestions. I really appreciate it.




"Tony Gravagno" <g6q3x9lu53001 (AT) sneakemail (DOT) com.invalid> wrote