[Info-Ingres] Incremental rollforward

[Paul White]

Hello all,

I'm trying to rollforward incrementally from an old checkpoint because I'm
missing checkpoint 6.
This command is rejected

rollforwarddb mydb \#c5 +c -j -incremental -norollback

with error

Mon Nov 16 15:39:25 2009 E_DM1373_RFP_INCR_ARG Invalid argument(s) #c for
incremental rollforwarddb.
Mon Nov 16 15:39:25 2009 E_DM1356_RFP_INVALID_ARGUMENT An invalid
rollforwarddb argument has been specified.
Mon Nov 16 15:39:25 2009 E_DM1301_RFP_ROLL_FORWARD An error occurred
that prevents further processing.

I don't have a copy of the aaaaaaaa.cnf from the time of ckp 5.

any suggestions?

Paul

[Martin Bowes]

Hi Paul,

IIRC you can only do an incremental recovery from the most recent
checkpoint. Ergo '#cN' wil be rejected.

But if its an incremental recovery this implies that the source database
is still running perfectly well. Why not just force a new checkpoint of
the source and restart the incremental recovery from the new backup?

Martin Bowes

Hello all,

with error

any suggestions?

Paul

_______________________________________________
Info-Ingres mailing list
Info-...@kettleriverconsulting.com
http://ext-cando.kettleriverconsulting.com/mailman/listinfo/info-ingres

[Paul White]

Thanks Marty,

The databases are around 250KM apart connected by a 512KB link with a
transfer time of around 5-8 hours depending on WAN traffic.

I've previously transferred full checkpoints (compressed) and journals each
night. Occasionally there is a comms problem and a checkpoint is not
transferred with no chance during the day to try another copy, in that case
I rely just on journals. In the event of a failover or disaster, I have
been able to run a normal rollforwarddb from any replicated checkpoint
providing including restoring to a point in time for recovering data
accidentally deleted.

Now I'm trying out the incremental rollforward (9.2) which works great until
it reaches the last journal before the nightly checkpoint where it complains
about the open journal. The procedure also comes unstuck if I run a random
checkpoint during the day just for safe keeping.

So in this scenario
Wangaratta Melbourne
ckp 4
ckp 5 ckp 5
jnl 51, 52,... jnl 51, 52,...
ckp 6
jnl 61, 62,... jnl 61, 62,...
ckp 7
jnl 71, 72,... jnl 71, 72,...

ckp 5 is a full online checkpoint followed by a full replication.
After that, I am replicating only journals and dumps.

In Melbourne, the incremental roll using ckp 5 works fine.
Incremental journal roll works but stops at the last journal prior to ckp6.
E_DM1377_RFP_INCR_JNL_WARNING ROLLFORWARDDB -incremental -norollback for
database mydb, ignoring open journal file sequence xxx

After that, the only DR option I have is to re-roll ckp 5 and all the
journals.

Is this a bug or feature? I've just read Nick's mail 22Jun2009 "closed
journals and incremental rollforwarddb". Looks like an enhancement waiting
to happen.

Paul

[Martin Bowes]

Hi Paul,

> Is this a bug or feature? I've just read Nick's mail 22Jun2009
"closed
journals and incremental rollforwarddb". Looks like an enhancement
waiting
to happen.

Last I heard from Mike Flower was that it's a shortfall they were hoping
to address.
I've cc'd him for an update.

Marty

PS. Wangaratta....God how I miss the North East of Vic.

[Michael Flower]

Hello Marty,

Thanks for the prod.This rings a bell - Incremental Rollforward does not detect the last closed jnl before the next ckp?

 

I'll chase with level 2.

 

Regards, Mike

---

From: Martin Bowes [mailto:martin...@ctsu.ox.ac.uk]

Sent: Wed 18/11/2009 14:26
To: Paul White; Michael Flower
Cc: Ingres and related product discussion forum
Subject: RE: [Info-Ingres] Incremental rollforward

[Michael Flower]

Hello Marty,

Nick Bolton has just given me a timely reminder about this (timely because I was looking at a fix only last week).

The bug number for this is 122216.

 

There’s also a related bug – 122328 a fix to prevent a checkpoint against a database when it is in “INCR_RFP” state.

Regards, Mike