Copying ingres checkpoints to another server

[Ingres Forums]

I would like to set up DR for our Live ingres installation. We hav
two identical servers, with identical ingres installations, onl
different databases on each. I believe this can be accomplished b
copying checkpoint and related files from one server to another, int
identically named locations, which we have. From posts on this site
have seen this can be done, indeed people are moving onto an incrementa
method. Just to start with, however, I wonder if someone could g
through the steps for copying checkpoints and related files to anothe
server, and/or advise me of any documenation on this subject issued b
ingres

--
douglas_
-----------------------------------------------------------------------
douglas_h's Profile: http://community.ingres.com/forum/member.php?userid=4454
View this thread: http://community.ingres.com/forum/showthread.php?t=1206

[Karl Schendel]

I would like to set up DR for our Live ingres installation. We have
two identical servers, with identical ingres installations, only
different databases on each. I believe this can be accomplished by
copying checkpoint and related files from one server to another, into
identically named locations, which we have. ...

What Ingres version, and what platform? The Ingres version does
not make a huge difference but it would be nice to know.

If you run "infodb databasename" on the database(s) of interest,
you'll see the checkpoint and dump locations listed. You need
*at least* those two. You'll probably want the journal location as
well.

An outline of the general procedure would be:
- create a database on the backup machine with the same name
and owner as the primary. Extend that database to all the same
data and work locations as the primary database. (infodb will
list any extra data and work locations.) If you want to be extra
slick, after creating the database, sql -u'$ingres' +U iidbdb on
the backup and fix the database's db_id in the iidatabase
catalog to match what's on the primary. (This has no effect
in running the database, but keeps ipm a lot happier!)

- copy the checkpoint and dump areas from primary to backup.
Delete the contents of the data area(a) on the backup side
(being careful to make sure you are doing this on the right
machine!!!) Copy the aaaaaaaa.cnf file in the root data
area (infodb again) to the backup. (This is not strictly necessary
but makes some Ingres versions happier, and is good
superstition anyway.)

- rollforwarddb +c -j databasename on the backup

If you want to throw journals into the mix, copy the journal
area and use a +c +j rollforward. There are additional
things you can do, such as issuing a DM1305+DM1314
trace point on the primary to force an archiver cycle, so that the
journals are as up to date as possible. (WIth the most
recent Ingres versions, you can use the incremental
journaling features, which I am unfortunately not all
that familiar with.)

Another thing to keep in mind is that you will probably need
the same Ingres users on the primary as on the backup,
with the same installation privileges, roles, etc. In other
words all of the iidbdb stuff.

I have one of those "I'm forgetting something" feelings, but
if so, surely someone else will chime in to fill the gaps.

Karl

[Ingres Forums]

Karl, that's incredibly helpful and will enable me to proceed wit
testing on a small database. We are running ingres 2006 version I
9.1.1 (int.lnx/103) NPTL on Red Hat Enterprise Linux AS release
(Nahant Update 8). Both machines are identical, also in terms of thei
ingres installations. We are not running journalised currently. I nee
to think about that. Users, etc on iidbdb should be identical. I thin
I should be able to make some progress. As you say, if there is anythin
else to consider perhaps someone else will chime in. Thanks again

--
douglas_
-----------------------------------------------------------------------
douglas_h's Profile: http://community.ingres.com/forum/member.php?userid=4454
View this thread: http://community.ingres.com/forum/showthread.php?t=1206

[Michael Dyer]

Douglas,
Do you have a support contract or a POC agreement?
If so, then please log an issue with ServiceDesk.
Michael
(Ingres Technical Support)

-----Original Message-----
From: info-ingres-***@kettleriverconsulting.com
[mailto:info-ingres-***@kettleriverconsulting.com] On Behalf Of
Ingres Forums
Sent: 18 May 2010 18:50
To: info-***@kettleriverconsulting.com
Subject: Re: [Info-Ingres] Copying ingres checkpoints to another server


Karl, that's incredibly helpful and will enable me to proceed with
testing on a small database. We are running ingres 2006 version II
9.1.1 (int.lnx/103) NPTL on Red Hat Enterprise Linux AS release 4
(Nahant Update 8). Both machines are identical, also in terms of their
ingres installations. We are not running journalised currently. I need
to think about that. Users, etc on iidbdb should be identical. I think
I should be able to make some progress. As you say, if there is anything
else to consider perhaps someone else will chime in. Thanks again.

[Ingres Forums]

Michael, thanks, we have 3rd party ingres support but they will charg
to set up this DR facility. As it appears to be widely used within th
industry I thought and hoped I could find the information I need to se
up the DR mechanism for my organisation myself, making some financia
savings. I'm also generally researching this topic. If I have an
problems with the implementation it would have to be passed to our 3r
party ingres support

--
douglas_
-----------------------------------------------------------------------
douglas_h's Profile: http://community.ingres.com/forum/member.php?userid=4454
View this thread: http://community.ingres.com/forum/showthread.php?t=1206

[Jean-Pierre Zuate, La Fage Conseil]

Hi (Karl and others),

"I'm forgetting something", perhap's :
1/ Both databases must have the same location name and area (due to
aaaaaaaa.cnf content)
2/ If you want to replay journal files and dump stuff, logging systems must
be tuned on the same values

I've one of this on a production machine (Ingres 2.6 on HP) and sometime, on
the destination machine, the archiver process (dmfacp) fail because it try
to create a journal file that already exists (because it was created on the
source machine), even if the destination machine is doing nothing in term of
transaction (timer somewhere to generate journal files ?).
The workaround is to tune the acpexit to restart dmfacp after cleanning
journal location on the destination machine.

Hope this help,
--
Jean-Pierre Zuate
La Fage Conseil
[phone]
jean-***@lafageconseil.fr
http://lafageconseil.fr/

Post by Karl Schendel

I would like to set up DR for our Live ingres installation. We have
two identical servers, with identical ingres installations, only
different databases on each. I believe this can be accomplished by
copying checkpoint and related files from one server to another, into
identically named locations, which we have. ...

What Ingres version, and what platform? The Ingres version does
not make a huge difference but it would be nice to know.
If you run "infodb databasename" on the database(s) of interest,
you'll see the checkpoint and dump locations listed. You need
*at least* those two. You'll probably want the journal location as
well.
- create a database on the backup machine with the same name
and owner as the primary. Extend that database to all the same
data and work locations as the primary database. (infodb will
list any extra data and work locations.) If you want to be extra
slick, after creating the database, sql -u'$ingres' +U iidbdb on
the backup and fix the database's db_id in the iidatabase
catalog to match what's on the primary. (This has no effect
in running the database, but keeps ipm a lot happier!)
- copy the checkpoint and dump areas from primary to backup.
Delete the contents of the data area(a) on the backup side
(being careful to make sure you are doing this on the right
machine!!!) Copy the aaaaaaaa.cnf file in the root data
area (infodb again) to the backup. (This is not strictly necessary
but makes some Ingres versions happier, and is good
superstition anyway.)
- rollforwarddb +c -j databasename on the backup
If you want to throw journals into the mix, copy the journal
area and use a +c +j rollforward. There are additional
things you can do, such as issuing a DM1305+DM1314
trace point on the primary to force an archiver cycle, so that the
journals are as up to date as possible. (WIth the most
recent Ingres versions, you can use the incremental
journaling features, which I am unfortunately not all
that familiar with.)
Another thing to keep in mind is that you will probably need
the same Ingres users on the primary as on the backup,
with the same installation privileges, roles, etc. In other
words all of the iidbdb stuff.
I have one of those "I'm forgetting something" feelings, but
if so, surely someone else will chime in to fill the gaps.
Karl
_______________________________________________
Info-Ingres mailing list
http://ext-cando.kettleriverconsulting.com/mailman/listinfo/info-ingres