| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 71.1 | | GHOTI::PROTEAU | Jean-Claude Proteau | Wed Aug 24 1994 12:47 | 12 |
| 71.2 | Current functionality ? | STKHLM::KNORN | 50% of Chip & Dale | Wed Aug 24 1994 17:56 | 14 |
| 71.3 | Yes, these features are available now | BROKE::PROTEAU | Jean-Claude Proteau | Thu Aug 25 1994 09:05 | 10 |
| 71.4 | Confused? | ORAREP::CHEFS::MACLEOD | | Wed Apr 24 1996 13:13 | 12 |
|
-< Yes, these features are available now >-
I'm confused. According to the v6.0 release notes (3.3.7.1) the syntax is
accepted by SQL but ignored by DDD.
I would be grateful for any light.
Thanks
Ferdy
|
| 71.5 | Only part of it is ignored | BROKE::PROTEAU | Jean-Claude Proteau | Wed Apr 24 1996 14:28 | 30 |
|
Ferdy,
I don't have the 6.0 release notes at my finger tips. The
CHECKPOINT clause as implemented in SQL's CREATE TRANSFER statement
has two parts:
CHECKPOINT [EVERY n MINUTES] [RECOVER WITHIN n MINUTES]
It is the EVERY n MINUTES part that is ignored by Data Distributor.
The RECOVER WITHIN n MINUTES part does work.
Checkpointing every n minutes was intended to allow us to checkpoint
in the middle of a transaction. The idfeas was that if you're transfer
a table with a very large number of rows, a failure in the middle of
the transaction wouldn't require us to start the transaction over from
scratch. Well, it turns out that the database systems do not provide
us with the necessary features to allow us to do that.
Recovering within n minutes does work. If a transaction fails, we
detach from the target database, delay a bit, reattach, and try to
restart the transaction. If that succeeds, we continue. If we fail to
attach or start the transaction, often it is because the network
connection is still disabled. Specifying a limit to how long you want
DD to keep retrying is done with the RECOVER WITHIN clause.
Please re-read the release note and let us know if the wording is still
not clear.
Claude
|
| 71.6 | Yes, Recovery is implemented | ORAREP::CHEFS::MACLEOD | | Thu Apr 25 1996 05:40 | 28 |
|
Hi Claude,
Yes, having re-read the documentation it does state that it's the
[EVERY n MINUTES] sub-clause that is not implemented by DDD. It's
just that from reading the earlier notes I thought that both
sub-clauses were implemenetd.
I have a similar situation to the originator of the note. I am trying
to Initialize a replication transfer, from a database with a table
which has a lot of rows (>500,000).
The problem is that the .RUJ and .SNP files grow too large, or I get
an exceeded quota error message when doing the first transfer. The
CHECKPOINT EVERY n MINUTES seemed the solution.
An alternative would be to RMU/backup the source database copy it to
the target site, then restore it, set protections etc. Then create a
Replication transfer to the now existing Database. But this involves
more manual steps.
I would be grateful for any ideas you might have.
Thanks again
Ferdy
|
| 71.7 | Future | BROKE::PROTEAU | Jean-Claude Proteau | Thu Apr 25 1996 10:01 | 7 |
|
I don't have any ready solution for you regarding the use of Data Dis-
tributor. I'll make a note of your problem and make a point to discuss
it with Rdb engineering to see if some solution can be implemented.
However, I can't promise when or if a solution would be available.
Claude
|